目录Contents
组会（最新一天：2026-10-17）Group meeting (latest day: 2026-10-17)
成果：Virgo 重提取批处理已在 Virgo 上运行（2026-10-17 05:22Z，PI 在对话中授权）：run_batch.sh PID 1078920/1078931，第 1 个场 0080540101，共 26 个场，预计 62–95 小时；启动时资源门为内存 39.5%、NVMe 494 GB、data8t 810 GB 空闲。红队 RT-22（codexnim，09:4xZ 分诊；Claude 复算）抓到两项实质问题：runner 里那个自定义的耀斑过滤步骤（2.1.filter）实际没有起作用——试点场 0200430301 的 mos1 目标谱曝光 11417.38 s，比 v19 的 10228.74 s 多 11.6%，而试点的 2.1.filter 与 2.filter 干净 GTI 完全相同（14 个区间、11630.0 s）；v19 的 rpc/2.1.filter/ 目录里保存的是自定义的差分谱与高斯面积交叉迭代产品，试点里只有标准的 *-hist.png——runner 的 mode A 走的是标准 espfilt 路径。第二项：retained_fraction 是拿 runner 自己的 2.filter 当基准算的，所以抓不到上面那个问题，试点审计因此通过。决定 D52：批处理继续跑（前面的校准、2.filter、edetect_stack 等上游步骤可以复用），同时发修正单 CORRECTION_02 给重提取批处理的 codexnim 线程——找回并恢复 v19 的自定义过滤器、从 2.1.filter 重跑试点、验收条件为目标谱曝光在 v19 的 2% 以内且 GTI 重叠 \(\ge 98\)%、写好 rerun_from_21filter.sh 但暂不启动，派发方审查后才在批处理场上重跑。重跑通过之前，任何拟合都不使用重提取批处理的谱——所以重提取批处理的产品暂时不能与 v19 比较（R015）。用能谱模拟校准判定阈值 \(\Delta \mathrm{stat} = 2.71\)：试点跑完 40/40 个配置，闭合检验一度未达标（0800731101 的三个节点比多元起始剖面曲面高 +0.34…+0.61 stat、0800730801 有一个节点 +0.059；原因：闭合只用 3 行热启动，而多元起始曲面用 11 行 kT 链）；2026-10-17 15:5xZ R022/D56 更新：CORRECTION_02（多元起始曲面矢量起点 + 正确的 UL 规则）后 stage 0 已被接受（闭合 9/9 达到或低于多元起始曲面，−0.02…−0.17），阈值仍未校准；stage 1 需要 PI 决定（4 个 lang worker 每个真值 17–19 天，T0/T1/rc30 三个真值 51–57 天），不再写「阻断」。（2026-10-17 更正（R016，红队 RT-21）：此前记的「局部极小/热启动」原因不对——差距来自没有使用多元起始曲面保存的起点；覆盖率程序取错交点，试点覆盖率数字撤回。）试点的覆盖曲线形状一度被读作符合预期（已撤回，R016）（A = 0 与 A = A_UL 处覆盖 1.0，A = 2\(\cdot\)A_UL 与 rc30 幅度处降到 0.0），（已撤回，R016）首次显示 \(\Delta \mathrm{stat} = 2.71\) 能覆盖 A_UL；但试点只用 2 个场（生产用 14 个）且每类只有 1 个玩具——是冒烟测试，不是校准，\(\pm 1.5\)% 的覆盖数字是 stage 1 的量。工作：重提取批处理运行中（Virgo，1 个 worker，26 场，62–95 h）；CORRECTION_02 运行中；阈值校准试点驱动 40/40 后仍在跑，stage 0 已被接受、stage 1 等 PI 决定；红队补充（0 → 2）：RT-21（Hermes PID 3822374，审阈值校准闭合/试点，报告仍在写）与 RT-22（codexnim PID 3822387，审重提取批处理 runner 配置，已分诊）；部署：刷新 6 续 B 阶段（03:54Z）与 stage 14 最终版（04:46Z）上线，实时条目 L008–L012 并入。问题：E026 产品暂时不可比（自定义过滤器空操作 + 审计自指，处置 = D52 继续批处理 + 从 2.1.filter 重跑试点）；E028 stage 0 闭合一度未达标（2026-10-17 15:5xZ R022/D56：CORRECTION_02 后 stage 0 已被接受，阈值仍未校准，stage 1 需 PI 决定）；无新增的代理自身错误。待 PI 决定：E028 stage 1 是否跑（每个真值 17–19 天，4 个 lang worker）。
Results: the Virgo re-extraction batch is running on Virgo (2026-10-17 05:22Z, authorised by the PI in chat): run_batch.sh PIDs 1078920/1078931, first field 0080540101, 26 fields, an expected 62–95 h; the start resource gate was memory 39.5 %, NVMe 494 GB and data8t 810 GB free. Red team RT-22 (codexnim, triaged 09:4xZ; Claude reproduced) found two material problems: the runner's custom flare-filter step (2.1.filter) did nothing — the pilot field 0200430301's mos1 obj.pi exposure is 11417.38 s against v19's 10228.74 s (+11.6 %), while the pilot's 2.1.filter and 2.filter clean GTIs are identical (14 intervals, 11630.0 s); v19's rpc/2.1.filter/ directory holds the custom diff-spectrum and gauss1-area-intersection-iter2 products, while the pilot holds only the standard *-hist.png — the runner's mode A ran the standard espfilt path. Second: retained_fraction is computed against the runner's own 2.filter, so it cannot catch the first problem, which is why the pilot audit passed. Decision D52: the batch keeps running (the upstream calibration, 2.filter and edetect_stack steps are reusable), with correction list CORRECTION_02 sent to the re-extraction batch's codexnim thread — find and restore v19's custom filter, re-run the pilot from 2.1.filter, acceptance = obj exposure within 2 % of v19 and GTI overlap \(\ge 98\) %, write rerun_from_21filter.sh but do not launch it, and the dispatcher reviews before any rerun on batch fields. Until that rerun passes, no fit uses any re-extraction-batch spectrum — so the re-extraction batch's products are not v19-comparable for now (R015). Calibrating the decision threshold \(\Delta \mathrm{stat} = 2.71\) with spectral simulation: the pilot finished 40/40 profiles and the closure test once fell short (0800731101's three nodes sit +0.34…+0.61 stat above the multi-start profile surface and 0800730801 has one node +0.059; cause: the closure warm-starts from only 3 rows while the multi-start surface uses an 11-row kT chain); update 2026-10-17 15:5xZ (R022/D56): after CORRECTION_02 (the multi-start-surface vector start plus the corrected UL rule) stage 0 is accepted (closure 9/9 at or below the multi-start surface, −0.02…−0.17), the threshold is still uncalibrated, and stage 1 needs a PI decision (17–19 days per truth on 4 lang workers, 51–57 days for T0/T1/rc30) — no longer written as "blocked". (Correction 2026-10-17 (R016, red team RT-21): the "local-minimum/warm-start" cause recorded earlier is wrong — the gap comes from not using the multi-start surface's saved starting vectors; the coverage code picks the wrong crossing, so the pilot coverage numbers are withdrawn.) The pilot's coverage curve was first read as the expected shape (withdrawn, R016; 1.0 at A = 0 and A = A_UL, falling to 0.0 at A = 2\(\cdot\)A_UL and the rc30 amplitude), (withdrawn, R016) a first indication that \(\Delta \mathrm{stat} = 2.71\) covers A_UL; but the pilot uses 2 fields (14 in production) and only 1 toy per kind — it is a smoke test, not a calibration, and the \(\pm 1.5\) % coverage number is a stage-1 quantity. Work: the re-extraction batch is running (Virgo, one worker, 26 fields, 62–95 h); CORRECTION_02 is running; the threshold calibration's pilot drivers are still running past 40/40, stage 0 accepted and stage 1 awaiting the PI; red teams topped up (0 → 2): RT-21 (Hermes PID 3822374, reviewing the threshold calibration's closure/pilot, report still being written) and RT-22 (codexnim PID 3822387, reviewing the re-extraction batch runner's configuration, triaged); deploys: refresh 6 continued stage B (03:54Z) and stage 14's final version (04:46Z), with live entries L008–L012 folded in. Problems: E026's products are not comparable for now (a no-op custom filter plus a self-referential audit; action = D52 keeps the batch and re-runs the pilot from 2.1.filter); E028 stage 0's closure once fell short (update 2026-10-17 15:5xZ, R022/D56: after CORRECTION_02 stage 0 is accepted, the threshold is still uncalibrated, and stage 1 needs a PI decision); no new agent errors. For the PI: whether to run E028 stage 1 (17–19 days per truth on 4 lang workers).
来源：cognition-os/experiments/E026_virgo_reextraction_d50.md（2026-10-17 05:22Z 批处理启动节与 09:4xZ RT-22 修订节）；cognition-os/04_DECISION_LOG.md D52；cognition-os/07_REVISION_LOG.md R015；cognition-os/08_RUNNING.md（hb #80b 05:22Z、hb #81 06:13Z、hb #85 09:4xZ 附录）；.agent/jobs/2026-10-16-e028-a2-spectral-calibration-stage0/REPORT.codexnim.md 第 6、8 节；cognition-os/06_AGENT_LOG.md 2026-10-17 各行（05:22Z、06:13Z、07:04Z、07:56Z、08:47Z、09:4xZ）；handbook/live_updates.jsonl L013。前一天：2026-10-16（A1d 验收 + A2 基线、RT-08…RT-14、E026 阻断、epoch 故事）Source: the 2026-10-17 05:22Z batch-launch section and the 09:4xZ RT-22 amendment of cognition-os/experiments/E026_virgo_reextraction_d50.md; D52 in cognition-os/04_DECISION_LOG.md; R015 in cognition-os/07_REVISION_LOG.md; cognition-os/08_RUNNING.md (the hb #80b 05:22Z, hb #81 06:13Z and hb #85 09:4xZ addenda); sections 6 and 8 of .agent/jobs/2026-10-16-e028-a2-spectral-calibration-stage0/REPORT.codexnim.md; the 2026-10-17 rows of cognition-os/06_AGENT_LOG.md (05:22Z, 06:13Z, 07:04Z, 07:56Z, 08:47Z, 09:4xZ); handbook/live_updates.jsonl L013. Previous day: 2026-10-16 (A1d accepted + the A2 baseline, RT-08…RT-14, E026 blocking, the epoch story).
最新状态（实时，下次刷新写入正文）Live status (folded into the pages at the next refresh)
以下条目在本次网页刷新之后出现，由 Claude 在分诊时从记录中摘写，未经 Hermes 改写。These items arrived after this site refresh; Claude wrote them from the records at triage time, and Hermes has not yet rewritten them.
- 2026-10-18T12:10Z 关于“早期/晚期”检验（RT-09-T3）的设计，红队预演了它可能失败的地方。结论：这个检验可以做（早期和晚期各至少 3 个配对场），但功效有限：在现实的散布下（单场散布 0.37）只有约 55 % 的把握能看出 0.47 的早晚差，若只取与旧样本共有的 10 个场（差值 0.357）则约 36 %。所以结果很可能是“无法判断”，这一点会在检验前写明，不能把它当成“没有时代效应”。另外，新旧样本只有 10 个场重合，新流量与旧流量也不是同一个量，所以它是一次新的测量，不是对旧斜率的重复。最终判定规则等第二个红队报告后再固定。来源：cognition-os/07_REVISION_LOG.md R045。（可信度：Provisional） 来源：
cognition-os/07_REVISION_LOG.md R045 (RT-45b)For the early/late test (RT-09-T3) a red team rehearsed how it could fail. The test is feasible (at least 3 matched fields on each side) but weakly powered: at a realistic per-field scatter of 0.37 it has only about 55 % chance of seeing a 0.47 early-late difference, and about 36 % if the 10 fields shared with the old sample (difference 0.357) set the size. An 'inconclusive' result is therefore likely and will be declared as such before the test, never read as 'no epoch effect'. The new and old samples share only 10 fields and the new flux is a different quantity from the old one, so this is a new measurement, not a repeat of the old slope. The final decision rule is fixed after the second red-team report. Source: cognition-os/07_REVISION_LOG.md R045. (trust: Provisional) Source:cognition-os/07_REVISION_LOG.md R045 (RT-45b) - 2026-10-18T10:22Z 外围场的流量偏低，红队重新检查了这个“随年份下降”的说法。改正如下：（1）形状更像“早期对晚期”的台阶，而不是平滑下降：23 个场里，一个台阶比直线趋势拟合得更好（AIC 差约 9.5），台阶位置在 2006–2010 年之间，最好的切分在 2007/2008 年，不是严格的 2010。（2）之前说的“台阶不显著（p 值 0.14）”是错的，原因是置换检验时只打乱了流量、没有同步打乱误差；改正后 p 值是 0.018（13 场）和 0.0009（23 场），台阶是显著的。（3）斜率 −0.032 的误差 ±0.007 偏乐观，自助法给 ±0.012，但斜率依然成立。（4）“88% 的差距在 2010 年以前”这个比例无法复现，已删去。（5）E025 掩膜失败不是原因：只用掩膜干净的 13 个场，斜率反而更陡。原因仍未解决（太阳风电荷交换看来不是主因，仪器仍未排除），E022/E023 的外场结论继续是“存疑”。（可信度：Provisional (reworded)） 来源：
cognition-os/04_DECISION_LOG.md D69; 07_REVISION_LOG.md R043 (RT-44)The red team re-checked the claim that outer-field fluxes decline with year. Corrections: (1) the shape looks more like an early-versus-late step than a smooth decline: on the 23 fields a step fits better than a straight trend (AIC difference about 9.5), the step lies between 2006 and 2010, and the best cut is 2007/2008, not strictly 2010. (2) The earlier statement that the step is not significant (p = 0.14) was wrong: the permutation test shuffled the fluxes but not their errors; corrected p is 0.018 (13 fields) and 0.0009 (23 fields), so the step is significant. (3) The error ±0.007 on the slope −0.032 is optimistic; a bootstrap gives ±0.012, but the slope stands. (4) The claim that 88 % of the gap is before 2010 cannot be reproduced and is dropped. (5) The E025 masking failure is not the cause: using only the 13 mask-clean fields the slope is steeper. The cause is still unresolved (solar-wind charge exchange does not look like the main driver; the instrument is not excluded), and the E022/E023 outer-field conclusions stay questioned. (trust: Provisional (reworded)) Source:cognition-os/04_DECISION_LOG.md D69; 07_REVISION_LOG.md R043 (RT-44) - 2026-10-18T09:27Z 同时放开宇宙 X 射线背景的谱指数和 CGM 的丰度后，14 个场的总量是 0.834 ± 0.044，比现行的 0.867 低 0.034（3.9%）。但这个更灵活的模型没有被信息准则（AIC）选中：多出来的 28 个参数几乎正好抵消了拟合改进（ΔAIC 约 0 到 +2，要选中需要 ≤ −10）。按事先登记的规则，现行引用 0.867 不变，0.834 只作为“模型敏感性”报告，不加进误差棒。单个场的流量会动到 ±66%，因为丰度和归一化之间有一条几乎平的山脊；总量稳定是因为这些变化互相抵消。此前“放开谱指数在 AIC 上被偏好”只在和 v19 单起点拟合比较时成立。（可信度：Provisional (sensitivity only)） 来源：
cognition-os/04_DECISION_LOG.md D68; 07_REVISION_LOG.md R042 (E029)Freeing the cosmic X-ray background photon index and the CGM abundance together gives a 14-field total of 0.834 ± 0.044, 0.034 (3.9 %) below the current 0.867. The more flexible model is not selected by the information criterion (AIC): its 28 extra parameters almost exactly cancel the improvement in fit (ΔAIC about 0 to +2; selection needs ≤ −10). By the pre-registered rule the quote 0.867 stands and 0.834 is reported only as a model sensitivity, not added to the error bar. Individual fields move by up to 66 % because abundance and normalisation form a nearly flat ridge; the total is stable because the changes partly cancel. The earlier statement that a free index is AIC-preferred holds only against the single-start v19 fit. (trust: Provisional (sensitivity only)) Source:cognition-os/04_DECISION_LOG.md D68; 07_REVISION_LOG.md R042 (E029)
工作台：M31 的星系周介质（New-ANGELS III）Workbench: the circumgalactic medium of M31 (New-ANGELS III)
这是什么项目？What is this project?
M31（仙女座星系）是离我们最近的大型旋涡星系。宇宙学认为星系的大部分普通物质（重子）不在星系内部，而在星系周围几十万光年的星系周介质（CGM）——百万度的稀薄气体。本项目用 XMM-Newton 的 22 个指向（proposal 080073，PI J. Li）在 M31 内晕（投影半径 10.5–30.8 kpc）逐视线测量弥漫软 X 射线发射，把「银河系前景 + M31 内晕」的总和（记号 CGMsum）测准，再用空间形状限制其中有多少属于 M31。测量已冻结（D48，2026-10-14）：14 个主样本场的 CGMsum 吸收流量引用值为 0.867 \(\pm 0.044\)（统计）\(+0.007/-0.003\)（NH 重拟合）\(\pm 0.027\)（本地热泡分配）（D65，2026-10-18；单位 1e-15 erg cm\(^{-2}\) s\(^{-1}\) arcmin\(^{-2}\)；历史值 0.965 是 v19 单起始基线，已被取代）。这是毛总量（gross）：孔径内未分辨点源也计在里面，D49 的 0.25 逐场项没有加到这 14 个主场上（R033/D60，2026-10-18）。共温 kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\)。
M31 (Andromeda) is the nearest large spiral galaxy. Cosmology says most of a galaxy's ordinary matter (baryons) is not inside the galaxy but in the circumgalactic medium (CGM), million-degree thin gas around it. Using 22 XMM-Newton pointings (proposal 080073, PI J. Li) toward the M31 inner halo (projected radii 10.5–30.8 kpc), this project measures the diffuse soft X-ray emission sightline by sightline, pins down the sum "Milky Way foreground + M31 inner halo" (notation CGMsum), and then limits how much belongs to M31 from its spatial shape. The measurement is frozen (D48, 2026-10-14): the 14 primary fields give the quoted CGMsum absorbed flux 0.867 \(\pm 0.044\) (stat) \(+0.007/-0.003\) (N_H, refit) \(\pm 0.027\) (LHB allocation) (D65, 2026-10-18; units 1e-15 erg cm\(^{-2}\) s\(^{-1}\) arcmin\(^{-2}\); the historical 0.965 was the v19 single-start baseline and is superseded). This is the gross total: unresolved sources inside the apertures are counted in it, and D49's 0.25 per-field term is not applied to these 14 primaries (R033/D60, 2026-10-18). Common kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\).
来源：引用总量 0.867 \(\pm 0.044\)（统计）\(+0.007/-0.003\)（NH，重拟合）\(\pm 0.027\)（本地热泡分配，受 0.08 keV 温度下限条件限制）来自 cognition-os/04_DECISION_LOG.md D65（2026-10-18 05:3xZ），其前身 D57（2026-10-17 16:5xZ）与 07_REVISION_LOG.md R024 给出 0.867 \(\pm 0.044\) 与旧的 \(\pm 0.034\) 固定归一化 NH 项——该旧项已由 D65 撤回；毛总量与权重分解（含未分辨点源、D49 的 0.25 不用于 14 主场、95% 上限 0.087）来自 R033 与 D60（2026-10-18）；0.965 \(\pm 0.044\) 是 v19 单起始历史值（D48，2026-10-14；cognition-os/insights/I033，\(\chi^{2}=11.99/13\)）；kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\) 来自 v19 基线。
Source: the quoted total 0.867 \(\pm 0.044\) (stat) \(+0.007/-0.003\) (N_H, refit) \(\pm 0.027\) (LHB allocation, conditional on the 0.08 keV temperature floor) comes from cognition-os/04_DECISION_LOG.md D65 (2026-10-18 05:3xZ); its predecessor D57 (2026-10-17 16:5xZ) with 07_REVISION_LOG.md R024 gave 0.867 \(\pm 0.044\) with the old \(\pm 0.034\) fixed-normalisation NH term — that old term is withdrawn by D65; the gross-total and weight decomposition (unresolved sources included, D49's 0.25 not applied to the 14 primaries, 95% upper limit 0.087) comes from R033 and D60 (2026-10-18); 0.965 \(\pm 0.044\) is the historical v19 single-start value (D48, 2026-10-14; cognition-os/insights/I033, \(\chi^{2}=11.99/13\)); kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\) from the v19 baseline.
现在进展如何？Where does it stand?
测量冻结；v19 基线（14 场）可信，但单起始分诊（已分诊）确认它是局部单起始基线——同模型多元起始把 14 场均值移到 0.891（−7.7%），最终总量将来自多元起始剖面曲面，模型 B/C 作拓扑系统误差；\(\pm0.044\) 只是统计误差。轮廓在 10.5–29 kpc 内平（\(\chi^{2}=11.99/13\)）。联合轮廓似然基线已在多元起始曲面上重算（临时，阈值未校准）：无 M31 分量被偏好，集中式 M31 \(\beta\) 晕（r_c \(\le 8 \mathrm{kpc}\)，MW 常数）在 R=20 kpc 的单侧 95% 上限（插值版）为 0.049–0.16（\(\beta 0.6\)…0.4），Zhang 名义中值 / 上限（r_c 8，\(\beta 0.4\)）3.5–5.2 倍；晕算术分诊（已分诊）确认算术与阈值用法，但上限主要靠一个视场（仅 v19 曲面）；模型 C 曲面复核（已分诊）：模型 C 曲面下该集中度基本消失——对外措辞「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」；玩具中位数撤回（已分诊）判定玩具蒙特卡洛校准只在简化模型下有效，\(\Delta \mathrm{stat} = 2.71\) 尚未用能谱模拟校准（能谱模拟阈值校准运行中）；插值仍未收敛（线性/样条 0.77–0.82）。历史值仅作记录：第一版 0.047–0.164（对插值敏感 2.0–2.3 倍，插值警告）；第二版（模型 C 曲面，敏感性变体）\(\beta = 0.5\) 时 r_c = 4/8/15/30 kpc 0.111/0.138/0.215/0.484。外样本从 8 个方向扩到 26 个已拟合方向（23 个有双侧误差）：干净外样本（13 场，无未掩展源）吸收流量 IVW 0.846 \(\pm 0.059\) vs 内场 0.965 \(\pm 0.044\)（v19 单起始历史值；引用总量已由 D57 改为 0.867 \(\pm 0.044\)（统计），其 NH 项与本地热泡分配项见 O1 行），固定效应外/内 = 0.877 \(\pm 0.072\)；含未掩展源的全部 23 场 0.831 \(\pm 0.045\)。红队 RT-03（已分诊）：随机效应给外/内 = 0.94 \(\pm 0.11\)，逐一剔除 0.83–0.99——稳妥说法是外场与内场在约 10–15% 内一致，「外场更暗」不稳健。E023 完成：内+外联合拟合把大核 M31 晕在 20 kpc 的单侧 95% 上限压到 0.46（r_c=50）／0.59（r_c=100）／0.90（r_c=200）（\(\beta = 0.5\)）——Zhang 名义中值剖面 0.83 被排除到 r_c \(\lesssim 125/175 \mathrm{kpc}\)（\(\beta = 0.4/0.5\)）；红队 RT-02 后「\(\beta = 0.6\) 时 r_c \(\le 200\) 全部排除」已撤回（15% 外场斑块底噪给 UL 0.90–1.06，该端边际），E023 将在流量空间于谱曲面上重做；EAGLE fiducial 0.51 到 r_c \(\lesssim 45/65/85 \mathrm{kpc}\)（\(\beta = 0.4/0.5/0.6\)）。红队 RT-01 已分诊：E020 模型 C 的「−15%」不是 SP 谱形系统误差。红队 RT-06（已分诊）：D49 原文成立——RASS 只排除宽的 R45 亮超出，OMNI 只排除 OMNI 追踪的高通量 SWCX，NW 四场 SP/QPB 比值约为其余十场 3\(\times\)（事后）——NW 超出仍是「未解释」，但倾向软质子。红队 RT-09/RT-09-T1/RT-12/RT-14（均已分诊）：外场流量随观测年代下降（\(\le2009 1.05\) \(\pm 0.08\) vs \(\ge2010 0.55\) \(\pm 0.09\)，约 2\(\sigma\)）——「怀疑探测器老化」已撤回，1.3 keV 特征是自由的 1.2–1.4 keV 高斯残差，外/内比与基于外场的极限改标「受观测年代混杂」。红队 RT-11（阻断级）：E026 的 runner 会让全部 28 个场失败，修正单已在试点场上通过（O1–O12 关闭）；批处理 2026-10-17 05:22Z 起在 Virgo 上运行（第 1 个场 0080540101，26 个场，62–95 h），但红队 RT-22（09:4xZ，R015）发现自定义耀斑过滤是空操作（试点曝光比 v19 多 11.6%、GTI 与 2.filter 相同），D52 让批处理继续跑并从 2.1.filter 重跑试点——重跑通过前 E026 产品不与 v19 比较、不进任何拟合。已分诊：RT-14-T1、RT-15（判 E028 stage 0 已被接受（R022/D56，2026-10-17 15:5xZ：CORRECTION_02 后闭合 9/9 达到或低于 A1d），阈值仍未校准，stage 1 等 PI 决定）、RT-16、RT-22。E028：试点跑完 40/40 个配置，覆盖曲线形状一度被读作符合预期（已撤回，R016）（A = 0 与 A = A_UL 处为 1.0，更大的真值处降到 0.0）；闭合检验一度未达标（0800731101 三个节点 +0.34…+0.61、0800730801 一个节点 +0.059，原因是 3 行热启动 vs A1d 的 11 行 kT 链），2026-10-17 15:5xZ R022/D56 已把 stage 0 改判为「已被接受」，但阈值仍未校准，stage 1 需要 PI 决定（4 个 lang worker 每个真值 17–19 天，T0/T1/rc30 三个真值 51–57 天）；试点只有 2 个场、每类 1 个玩具——是冒烟测试，不是校准。2026-10-18 D66：PI 已批准 stage 1——三个真值（T0/T1/rc30）在 lang2 上并行跑，每个至少 200 次模拟，受资源约束（E029 结束后约需 17–19 天）；在它完成之前，A2 的上限仍写「阈值未校准」。运行中（待定）：E028 stage 1（PI 已批准，运行中，阈值仍未校准）、E026 的 CORRECTION_02 重跑、E027 1b/1c、RT-09-T3、RT-21。
The measurement is frozen and the v19 baseline (14 fields) is trusted, but red team RT-04 (triaged) confirms it is a local single-start baseline — same-model multi-start moves the 14-field mean to 0.891 (−7.7%), the final total will come from the A1d profile surfaces with Models B/C as a topology systematic, and \(\pm0.044\) is statistical only. The profile is flat over 10.5–29 kpc (\(\chi^{2}=11.99/13\)). The A2 baseline has been recomputed on the A1d surfaces (provisional, threshold uncalibrated): no M31 component is preferred, and the one-sided 95% upper limit (interpolated) on a concentrated M31 beta halo (r_c \(\le 8 \mathrm{kpc}\), MW constant) at R = 20 kpc is 0.049–0.16 (\(\beta 0.6\)…0.4), Zhang's nominal median / UL (r_c 8, \(\beta 0.4\)) is 3.5–5.2\(\times\); red team RT-13 (triaged) confirms the arithmetic and the threshold's usage, but the limit leans on one field on the v19 surfaces only; red team RT-16 (triaged) re-checked it: under the Model C surfaces that concentration largely disappears — the public wording is "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)"; red team RT-08 (triaged) rules that the toy-MC calibration holds only under its simplified model, so \(\Delta \mathrm{stat} = 2.71\) is not yet calibrated by spectral simulation (E028 running); interpolation is still not converged (linear/spline 0.77–0.82). Historical values, for the record only: A2 v1 0.047–0.164 (interpolation-sensitive by 2.0–2.3\(\times\), RT-02); A2 v2 (on the Model-C surfaces, a sensitivity variant) gives, for \(\beta = 0.5\) at r_c = 4/8/15/30 kpc, limits of 0.111/0.138/0.215/0.484. The outer sample has grown from 8 to 26 fitted directions (23 with two-sided errors): the clean outer sample (13 fields without an unmasked extended target) has absorbed-flux IVW 0.846 \(\pm 0.059\) vs the inner 0.965 \(\pm 0.044\) (the historical v19 single-start value; the quoted total is now 0.867 \(\pm 0.044\) (stat) under D57, with its N_H and LHB-allocation terms in the O1 row), fixed-effect outer/inner = 0.877 \(\pm 0.072\); all 23 fields including unmasked targets give 0.831 \(\pm 0.045\). Red team RT-03 (triaged): random effects give outer/inner = 0.94 \(\pm 0.11\) and leave-one-out 0.83–0.99 — the safe statement is that outer and inner agree within ~10–15%, and "the outer fields are fainter" is not robust. E023 is done: a joint inner+outer fit limits a large-core M31 halo at 20 kpc to a one-sided 95% UL of 0.46 (r_c=50) / 0.59 (r_c=100) / 0.90 (r_c=200) for \(\beta = 0.5\) — the Zhang nominal median profile 0.83 is excluded up to r_c \(\lesssim 125/175 \mathrm{kpc}\) (\(\beta = 0.4/0.5\)); after red team RT-02, "for all r_c \(\le 200\) at \(\beta = 0.6\)" is withdrawn (a 15% outer patchiness floor gives UL 0.90–1.06, so that end is marginal), and E023 will be redone in flux space on spectral surfaces; the EAGLE fiducial 0.51 up to r_c \(\lesssim 45/65/85 \mathrm{kpc}\) (\(\beta = 0.4/0.5/0.6\)). Red team RT-01 is triaged: E020's Model-C "−15%" is not an SP-shape systematic. Red team RT-06 (triaged): D49 stands as written — RASS excludes only a broad R45-bright excess, OMNI excludes only OMNI-traced high-flux SWCX, and the NW four's SP/QPB ratio is about 3\(\times\) the other ten (a posteriori) — the NW excess is still "unexplained" but leans towards soft protons. Red teams RT-09/RT-09-T1/RT-12/RT-14 (all triaged): the outer-field flux falls with observing year (\(\le2009 1.05\) \(\pm 0.08\) vs \(\ge2010 0.55\) \(\pm 0.09\), about 2\(\sigma\)) — "detector ageing suspected" is withdrawn, the 1.3 keV feature is a free 1.2–1.4 keV Gaussian residual, and the outer/inner ratio and the outer-based limits are marked "epoch-confounded". Red team RT-11 (blocking): E026's runner would fail all 28 fields, and its correction list passed on the pilot field (O1–O12 closed); the batch has been running on Virgo since 2026-10-17 05:22Z (first field 0080540101, 26 fields, 62–95 h), but red team RT-22 (09:4xZ, R015) found the custom flare filter is a no-op (the pilot exposure is 11.6 % above v19 and its GTIs match 2.filter), so D52 keeps the batch running while re-running the pilot from 2.1.filter — until that rerun passes E026's products are not compared with v19 and enter no fit. Triaged: RT-14-T1, RT-15 (under which E028 stage 0 is accepted (R022/D56, 2026-10-17 15:5xZ: after CORRECTION_02 the closure is 9/9 at or below A1d), the threshold is still uncalibrated, and stage 1 awaits the PI), RT-16, RT-22. E028: the pilot finished 40/40 profiles with a coverage curve first read as the expected shape (withdrawn, R016; 1.0 at A = 0 and A = A_UL, falling to 0.0 at larger truths); the closure test once fell short (0800731101's three nodes +0.34…+0.61, 0800730801's one node +0.059, caused by a 3-row warm start against A1d's 11-row kT chain), and R022/D56 of 2026-10-17 15:5xZ re-graded stage 0 as "accepted", but the threshold is still uncalibrated and stage 1 needs a PI decision (17–19 days per truth on 4 lang workers, 51–57 days for T0/T1/rc30); the pilot has only 2 fields and 1 toy per kind — a smoke test, not a calibration. D66 of 2026-10-18: the PI has approved stage 1 — the three truths (T0/T1/rc30) run in parallel on lang2, at least 200 simulations each, about 17–19 days after E029 ends because of the resource cap; until it finishes, every A2 upper limit keeps the label "threshold uncalibrated". Running (Pending): E028 stage 1 (PI-approved and running, the threshold still uncalibrated), E026's CORRECTION_02 rerun, E027 1b/1c, RT-09-T3, RT-21.
还有什么没解决？What is unsolved?
① 前景与 M31 谱形几乎相同，无法按距离分解——集中核（r_c \(\le 8 \mathrm{kpc}\)）已被 A2 基线（A1d 曲面）压到 0.049–0.16（见状态页；红队 RT-13 已分诊：算术与 \(\Delta \mathrm{stat} = 2.71\) 的边界用法成立，但上限主要靠一个视场（仅 v19 曲面）；红队 RT-16 已分诊：模型 C 曲面下该集中度基本消失，对外措辞「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」；红队 RT-08 已分诊：阈值尚未用能谱模拟校准，E028 stage 0 已被接受（R022/D56，2026-10-17 15:5xZ）但阈值仍未校准、stage 1 等 PI 决定；插值仍未收敛，线性/样条 0.77–0.82），大核（r_c \(\ge 50 \mathrm{kpc}\)）一度被 E023 限制（0.46–0.90 @20 kpc，\(\beta = 0.5\)；\(\beta = 0.6\) 大核端边际，「r_c \(\le 200\) 全部排除」已撤回）——但 R023 后这些上限被质疑、暂不引用（加年代项后 0.87/1.29/2.38，且偏好有 M31 项）；② 外 NW 四场偏亮：E014 排除宽的 R45 亮天空结构与 OMNI 追踪的高通量太阳风；红队 RT-01 分诊后 NW 超出仍是「未解释」——固定 SP 谱形的说法不成立（等权重下模型 C（模型 C = E020 的 SP 谱形与线能量自由变体，只作敏感性检验）内 NW/其余仍 1.34），D49 原文生效（0.25 未解释场级系统误差），D49 更新（−15% SP 谱形系统误差）已撤回；红队 RT-06 后「未解释」倾向软质子（NW 四场 SP/QPB 约为其余十场 3\(\times\)，事后 p \(\approx 0.005\)–0.008）；③ v19 拟合 QA：模型 C（SP 谱形 + 线能量自由）只是敏感性变体——−15% 的 IVW 移动一半来自 v19 重收敛（A 步 −7.65%，A→B −5.29%，B→C −2.87%），且模型 C 的 SP 谱形未经物理验证（13/14 场 SP 升、MWhalo 降）；④ PSF 翼泄漏实测 0.16（原值）／0.33（零点校正）vs E016 模型 0.12——落在 E016 的 0.12–0.33 括号内（红队 RT-05：高括号 0.33，不写「0.28 保守」）；⑤ Locatelli 中心预言 1.29 > 测量 0.867，计入模型自身误差（\(\sigma \approx 0.32\)）后差距只有约 1.3\(\sigma\)、2\(\sigma\) 内允许正的 M31 残差——写作「有条件中心值张力」，不写「高估」也不写「排除」（R027/RT-29，2026-10-17 18:3xZ；此前 RT-04 的「有条件紧张」措辞保留）；⑥ 300 kpc 图：外样本已扩到 26 个已拟合方向（23 个有双侧误差）（固定效应干净外/内 0.877 \(\pm 0.072\)，随机效应 0.94 \(\pm 0.11\)——稳妥说法「约 10–15% 内一致」），但全部在负银纬；12 个 P1 场与 10 个 P2 a+b 场的 v19 掩膜盖不住引文的展源圆（\(\le 0.066\)），需 Virgo 重提取（E026，28 场——批处理 2026-10-17 05:22Z 起在 Virgo 上运行（第 1 个场 0080540101，26 个场，62–95 h）；红队 RT-11（阻断级）发现 runner 会让全部 28 个场在审计这一步失败，修正单已在试点场上通过；红队 RT-22（09:4xZ，R015）又发现自定义耀斑过滤步骤是空操作——试点场曝光比 v19 多 11.6%、2.1.filter 与 2.filter 的 GTI 完全相同，D52 让批处理继续跑并同时从 2.1.filter 重跑试点，重跑通过前 E026 产品不与 v19 比较、不进任何拟合）；⑦ 外场流量随观测年代下降（红队 RT-09/RT-09-T1/RT-12/RT-14，均已分诊）：\(\le2009\) 的 10 个场 1.05 \(\pm 0.08\) vs \(\ge2010\) 的 3 个场 0.55 \(\pm 0.09\)（内场全是 2017–2018 年），年代分割约 2\(\sigma\)；「怀疑探测器老化」已撤回（公布的 MOS 校准只有几个百分点，实测 ARF 软端只变 +1.6%/−3.7%），1.3 keV 特征是自由的 1.2–1.4 keV 高斯残差而非仪器线，观测年份与 CCD 状态高度共线（方差膨胀因子 20–39）——原因未定，外/内比与基于外场的极限（E022/E023 大核上限）标为「受观测年代混杂」，等 RT-14-T1 与 RT-09-T3。为什么这叫「混杂」：内场全部采于 2017–2018 年，外场横跨 2001–2022 年，所以「外场比内场暗」这句话把两种可能混在一起——要么外场方向上的天空本来就更暗，要么晚期的观测整体上低估了流量。打个比方：用 2001 年的相机拍的一批照片和用 2022 年的相机拍的另一批照片比亮度，如果两批的曝光标定差了几个百分点，而你要找的信号只有 10%，那么「新照片更暗」既可能是物体真的暗了，也可能只是新相机的标定更保守。要分开它们，必须在同一套校准与同一套背景处理下重采一批早期和一批晚期外场；在那之前，任何依赖外场的内外比较都同时带着方向差异和年代差异，不能单独归因于任一个。
(i) The foreground and M31 have nearly identical spectra, so distance separation fails — the compact core (r_c \(\le 8 \mathrm{kpc}\)) is already pushed to 0.049–0.16 by the A2 baseline on the A1d surfaces (see the state page; red team RT-13 is triaged: the arithmetic and the \(\Delta \mathrm{stat} = 2.71\) boundary usage hold, but the limit leans on one field on the v19 surfaces only; red team RT-16 is triaged: under the Model C surfaces that concentration largely disappears, with the public wording "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)"; red team RT-08 is triaged: the threshold is not yet calibrated by spectral simulation — E028 stage 0 is accepted (R022/D56, 2026-10-17 15:5xZ) but the threshold is still uncalibrated and stage 1 awaits a PI decision; interpolation is still not converged, linear/spline 0.77–0.82), and the large core (r_c \(\ge 50 \mathrm{kpc}\)) was limited for a while by E023 (0.46–0.90 at 20 kpc, \(\beta = 0.5\); the \(\beta = 0.6\) large-core end is marginal, "for all r_c \(\le 200\)" withdrawn) — but after R023 those limits are Questioned and not quoted (with an epoch term they become 0.87/1.29/2.38 and an M31 term is preferred); (ii) the bright outer-NW four: E014 excluded a broad R45-bright sky structure and OMNI-traced high-flux solar wind; after red team RT-01's triage the NW excess is still "unexplained" — the fixed-SP-shape explanation does not hold (with equal weights NW/other stays 1.34 even under Model C — Model C is E020's free-SP-shape and free-line-energy variant, used only as a sensitivity test), D49 stands as written (the 0.25 unexplained field-level systematic), and the D49 update (a −15% SP-shape systematic) is withdrawn; after red team RT-06 "unexplained" leans towards soft protons (the NW four's SP/QPB is about 3\(\times\) the other ten, a posteriori p \(\approx 0.005\)–0.008); (iii) v19 fit QA: Model C (free SP shape and line energies) is only a sensitivity variant — half of the −15% IVW shift comes from re-converging the v19 fits (step A −7.65%, A→B −5.29%, B→C −2.87%), and Model C's SP shapes are not physically validated (SP up, MWhalo down in 13/14 fields); (iv) the PSF-wing leak is 0.16 (raw) / 0.33 (null-corrected) empirically vs E016's model 0.12 — inside E016's 0.12–0.33 bracket (red team RT-05: the high bracket is 0.33, not "0.28 conservative"); (v) Locatelli's central prediction 1.29 > the measured 0.867, but with the model's own error (\(\sigma \approx 0.32\)) the gap is only about 1.3\(\sigma\) and a positive M31 residual is allowed at 2\(\sigma\) — write "conditional central-value tension", neither "over-predicts" nor "excludes" (R027/RT-29, 2026-10-17 18:3xZ; RT-04's earlier "conditional tension" wording is retained); (vi) the 300-kpc map: the outer sample has grown to 26 fitted directions (23 with two-sided errors) (fixed-effect clean outer/inner 0.877 \(\pm 0.072\), random effects 0.94 \(\pm 0.11\) — the safe statement "agree within ~10–15%") but all at negative Galactic latitude; the v19 masks of the 12 P1 fields and 10 P2 a+b fields do not cover the cited extended-target circles (\(\le 0.066\)), so they need Virgo re-extraction (E026, 28 fields — the batch has been running on Virgo since 2026-10-17 05:22Z (first field 0080540101, 26 fields, 62–95 h); red team RT-11 (blocking) found the runner would fail all 28 fields at the audit step and its correction list passed on the pilot field; red team RT-22 (09:4xZ, R015) then found the custom flare-filter step is a no-op — the pilot field's exposure is 11.6 % above v19 and its 2.1.filter and 2.filter GTIs are identical, so D52 keeps the batch running while re-running the pilot from 2.1.filter, and until that rerun passes E026's products are not compared with v19 and enter no fit); (vii) the outer-field flux falls with observing year (red teams RT-09/RT-09-T1/RT-12/RT-14, all triaged): 10 fields \(\le2009\) give 1.05 \(\pm 0.08\) vs 3 fields \(\ge2010\) giving 0.55 \(\pm 0.09\) (the inner fields are all 2017–2018), an epoch split of about 2\(\sigma\); "detector ageing suspected" is withdrawn (the published MOS calibration is a few percent and the fitted ARFs' soft end changes only +1.6 %/−3.7 %), the 1.3 keV feature is a free 1.2–1.4 keV Gaussian residual not an instrumental line, and observation year is strongly collinear with CCD state (variance inflation 20–39) — the cause is unresolved, and the outer/inner ratio and the outer-based limits (E022/E023's large-core ULs) are marked "epoch-confounded" pending RT-14-T1 and RT-09-T3. Why this is called a "confound": the inner fields were all taken in 2017–2018 while the outer fields span 2001–2022, so the sentence "the outer fields are fainter than the inner ones" mixes two possibilities together — either the sky in the outer directions really is fainter, or the later observations systematically underestimate the flux. An analogy: comparing the brightness of one batch of photographs taken with a 2001 camera against another batch taken with a 2022 camera, if the two batches' exposure calibrations differ by a few percent while the signal you are hunting is only 10%, then "the new photographs are dimmer" could mean either that the object really got dimmer or that the new camera's calibration is simply more conservative. Separating them requires re-taking a batch of early and a batch of late outer fields under one calibration and one background treatment; until then every outer-based inner/outer comparison carries both a direction difference and an epoch difference and cannot be attributed to either alone.
下一步做什么？What do I do next?
按 编号计划：A1d 已验收，A2 基线已在它上面重算（临时，阈值未校准）→ 等 E028 用能谱模拟校准 \(\Delta \mathrm{stat} = 2.71\)（stage 0 已被接受（R022/D56）；stage 1 已由 PI 批准（D66）并在 lang2 上运行——三个真值并行、每个至少 200 次模拟，约需 17–19 天；在它完成前 A2 上限仍写「阈值未校准」）；在谱曲面上以流量空间重做 E023（加 MW 纬度项与外场系统项）；E026 的 28 场 Virgo 重提取批处理已在运行（2026-10-17 05:22Z 起，26 个场，62–95 h）——红队 RT-11 的修正单已在试点场上通过，但红队 RT-22（R015）发现自定义耀斑过滤是空操作（试点曝光比 v19 多 11.6%），D52 让批处理继续跑并同时从 2.1.filter 重跑试点，重跑通过前 E026 产品不与 v19 比较、不进任何拟合，之后才能做扩大后的内外样本联合拟合；红队 RT-14-T1（只用响应文件的 CCF 审计）与 RT-09-T3（早/晚配对重提取）决定外场年代混杂的原因。已完成：E013 A2 v1 + toy MC、A2 v2（敏感性变体）、E019、E020、E021、E022、E023、E017a、V01、E024、E025、A1c、A1d + A2 基线、RT-01…RT-14 分诊（RT-15/RT-16/RT-22 已分诊）。命令见 byhand。
Follow the numbered plan: A1d is accepted and the A2 baseline has been recomputed on it (provisional, threshold uncalibrated) → wait for E028 to calibrate \(\Delta \mathrm{stat} = 2.71\) with spectral simulation (stage 0 is accepted (R022/D56); stage 1 is PI-approved (D66) and running on lang2 — three truths in parallel, at least 200 simulations each, about 17–19 days; until it finishes every A2 upper limit keeps the label "threshold uncalibrated"); redo E023 in flux space on spectral surfaces (with an MW latitude term and an outer systematic); E026's 28-field Virgo re-extraction batch is already running (since 2026-10-17 05:22Z, 26 fields, 62–95 h) — red team RT-11's correction list passed on the pilot field, but red team RT-22 (R015) found the custom flare filter is a no-op (the pilot exposure is 11.6 % above v19), so D52 keeps the batch running while re-running the pilot from 2.1.filter, and until that rerun passes E026's products are not compared with v19 and enter no fit; only then comes the joint fit on the enlarged sample; red teams RT-14-T1 (a response-only CCF audit) and RT-09-T3 (an early/late paired re-extraction) decide the cause of the outer epoch confound. Done: E013 A2 v1 + toy MC, A2 v2 (a sensitivity variant), E019, E020, E021, E022, E023, E017a, V01, E024, E025, A1c, A1d + the A2 baseline, and the RT-01…RT-14 triages (RT-15/RT-16/RT-22 triaged). Commands at byhand.
最新骨架：流水线与各部分如何衔接Latest skeleton: the pipeline and how the parts connect
来源：流程依据 cognition-os/00_AI_BRIEFING.md 与 background_spectrum_fitting_2T_basedon_obsid_v19.py 的模型结构；数值见 cognition-os/insights/I033。
Source: the flow follows cognition-os/00_AI_BRIEFING.md and the model structure of background_spectrum_fitting_2T_basedon_obsid_v19.py; numbers from cognition-os/insights/I033.
目标表（O1–O5）Objectives (O1–O5)
| 目标Objective | 状态Status | 当前最好数值（估计器已命名）Best current number (estimator named) | 趋势Trend |
|---|---|---|---|
| O1 M31 内晕方向冷视线总和（CGMsum）的测量Total cool line-of-sight emission (CGMsum) toward the M31 inner halo | done | 引用总量（D65，2026-10-18 05:3xZ）：0.867 \(\pm 0.044\)（统计）\(+0.007/-0.003\)（NH，重拟合）\(\pm 0.027\)（本地热泡分配），吸收 0.5–2 keV，单位 1e-15 erg/cm\(^{2}\)/s/arcmin\(^{2}\)，来自 A1d 多元起始曲面 + 插值区间；共温 kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\)（v19）。这是毛总量（gross）：孔径内未分辨点源也计在里面（R033/D60，2026-10-18）；D49 的 0.25 逐场项没有加到这 14 个主场上（95% 上限 0.087）；从 0.965 降到 0.867 主要来自新的逐场权重，不是各场流量普遍下降。历史：v19 单起始值 0.965 \(\pm 0.044\)（D48 冻结，2026-10-14）已被 D57 取代（低 10%）；红队 RT-04（已分诊）当时指出它是局部单起始基线——同模型多元起始（E020 步 A）把均值移到 0.891（−7.7%），最终总量来自 A1d 剖面曲面。敏感性：模型 C（放开 SP 谱形）是敏感性检验，相对 A1d 为 −6 至 −12%（取决于加权方式，5/14 场反而变亮），AIC/BIC 在 14/14 场都更差（D58/R026）——不是单向系统误差。2026-10-18 R034/R035/D61 新增两项模型杠杆：放开 CXB 谱指数使 CGMsum 降 6.2%（13/14 场下降，AIC 更优）；金属丰度 Z = 0.3 也是一个流量杠杆（Z 0.1/1.0 时 −6 至 −9%）——E029 正按 D61 的预注册规则检验，结果为待定。吸收：把吸收柱密度换成 Willingale 总柱密度（比 HI4PI 高 19–31%）后重新拟合，吸收后的 CGM 流量只变 +0.8%，\(\pm10\%\) 只变不到 0.4%；因此 NH 项由 D65 从旧的固定归一化 \(\pm 0.034\) 改为 \(+0.007/-0.003\)（重拟合），光谱本身定不出柱密度。本地热泡分配：在 14 个场允许的本地热泡范围内总量变化 −0.7% 到 −7.0%，折合 \(\pm 0.027\)；−7% 那一端由模型里 0.08 keV 的温度下限造成（9/14 场卡在下限），把下限放到 0.05 keV 后两个场方向反转（R038）。\(\pm0.044\) 只是统计误差Quoted total (D65, 2026-10-18 05:3xZ): 0.867 \(\pm 0.044\) (stat) \(+0.007/-0.003\) (N_H, refit) \(\pm 0.027\) (LHB allocation), absorbed 0.5–2 keV, units 1e-15 erg/cm\(^{2}\)/s/arcmin\(^{2}\), from the A1d multi-start surfaces with interpolated intervals; common kT = 0.17531 \(\pm 0.00436 \mathrm{keV}\) (v19). This is the gross total: unresolved sources inside the apertures are counted in it (R033/D60, 2026-10-18); D49's 0.25 per-field term is not applied to these 14 primaries (95% upper limit 0.087), and the drop from 0.965 comes mostly from the new per-field weights rather than a uniform fall in every field's flux. History: the v19 single-start value 0.965 \(\pm 0.044\) (frozen by D48, 2026-10-14) is superseded by D57 (10 % lower); red team RT-04 (triaged) had already flagged it as a local single-start baseline — same-model multi-start (E020 step A) moves the mean to 0.891 (−7.7%), and the final total comes from the A1d profile surfaces. Sensitivity: Model C (free SP shape) is a sensitivity test, −6 to −12% versus A1d depending on the weighting (5 of 14 fields get brighter), and AIC/BIC are worse in 14/14 fields (D58/R026) — not a one-sided systematic. Two further model levers added on 2026-10-18 by R034/R035/D61: freeing the CXB photon index lowers the CGMsum by 6.2% (13/14 fields fall, AIC-preferred), and the metal abundance Z = 0.3 is also a flux lever (−6 to −9% at Z 0.1/1.0) — E029 is testing both under D61's pre-registered rule and its result is Pending. Absorption: refitting with the absorbing column replaced by the Willingale total column (19–31% above HI4PI) moves the absorbed CGM flux by only +0.8%, and \(\pm10\%\) by less than 0.4%; so D65 replaces the old fixed-normalisation \(\pm 0.034\) NH term with \(+0.007/-0.003\) (refit), and the spectra themselves do not determine the column. LHB allocation: over each field's allowed Local Hot Bubble range the total moves by −0.7% to −7.0%, i.e. \(\pm 0.027\); the −7% end is set by the model's 0.08 keV temperature floor (9 of 14 fields sit on it), and lowering the floor to 0.05 keV reverses the direction in two fields (R038). \(\pm0.044\) is statistical only | →已定（D65）fixed (D65) |
| O2a 由轮廓形状约束集中式 M31 热晕（小核）的大小/幅度Constrain the size/amplitude of a concentrated M31 hot halo (compact core) from the profile shape | provisional | A2 基线在 A1d 上（2026-10-16 13:2xZ，临时，阈值未校准）：更密的似然网格（14 场 \(\times 117\) 点 = 1638/1638，0 个拟合失败，65.7 worker-小时）已验收。同一估计器（MW 常数、kT 在最佳共温 \(\pm1\) 行内 profile、\(\Delta \mathrm{stat} = 2.71\)、插值）给出 20 kpc 处单侧 95% 上限：核半径 \(\le 8 \mathrm{kpc}\) 时 0.12–0.16（\(\beta 0.4\)）／0.075–0.10（\(\beta 0.5\)）／0.049–0.071（\(\beta 0.6\)）；r_c 30 为 0.30–0.56，r_c 50 为 0.68–1.26（单位 1e-15 erg/cm\(^{2}\)/s/arcmin\(^{2}\)）。插值收敛：线性/样条上限比 0.77–0.82（A1 上是 0.43–0.50），仍差于 10% 目标——取较松的样条值，线性值为插值系统误差（−18 至 −23%）。Zhang 名义中值 0.8285 / 上限（r_c \(\le 8\)）= 5.2–6.7（\(\beta 0.4\)）、8.1–11.1（\(\beta 0.5\)）、11.7–17.1（\(\beta 0.6\)），对外措辞 「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」（红队 RT-16 复核：v19 曲面下去掉 0800730601 放宽 +46…+62%，模型 C 曲面下只放宽 6–30% 且该场降到 14 场里第 4 位，故含模型 C 时引用 3.5–5.2 倍而非 \(\gtrsim4\) 倍）。红队 RT-13（已分诊）：算术与输入成立，A \(\ge 0\) 的最小值在全部 39 个配置都落在 A = 0，所以 \(\Delta \mathrm{stat} = 2.71\) 是「参数不能为负」边界情形下正确的 50:50 分位数；上限主要靠一个视场（0800730601）——去掉它放宽 +46…+62%（r_c 8、\(\beta 0.4\)：0.161 → 0.253，Zhang/上限 5.2 → 3.3）；报出的自然样条值是四种插值里最松的（去重/PCHIP/线性紧 2–23%），kT 一致的流量换算再紧约 \(\approx9.7\)%。红队 RT-08（已分诊）：「上限已校准良好」不成立——toy-MC 的位置族模型只在简化下有效（高斯抽样下覆盖率 0.78–0.88），\(\Delta \mathrm{stat} = 2.71\) 本身在理想高斯情形下仍正确，但阈值尚未用能谱模拟校准（E028 stage 0 已被接受（R022/D56，2026-10-17 15:5xZ：CORRECTION_02 后闭合 9/9 达到或低于 A1d），阈值仍未校准，stage 1 需 PI 决定——不是「阻断」）。RT-02 的插值警告仍在：A1 上敏感 2.0–2.3 倍，A1d 上降到 0.77–0.82。RT-07：只排除 Zhang 的名义中值剖面，不写「叠加预期」（Zhang 只发布 16/84% 边际、无协方差；v19 APEC 桥给 0.76–0.90）。历史值仅作记录：A2 v1（0.047–0.164，toy-MC 覆盖率 0.975–0.984）与敏感性变体 A2 v2（模型 C 曲面，\(\beta = 0.5\) 时 r_c = 4/8/15/30 = 0.111/0.138/0.215/0.484，覆盖率 0.918–0.933）The A2 baseline on A1d (2026-10-16 13:2xZ; provisional, threshold NOT calibrated): the denser likelihood grid (14 fields \(\times 117\) points = 1638/1638, 0 failed fits, 65.7 worker-hours) is accepted. The same estimator (MW constant, kT profiled over the best shared kT \(\pm1\) row, \(\Delta \mathrm{stat} = 2.71\), interpolated) gives one-sided 95% ULs at 20 kpc: for core radius \(\le 8 \mathrm{kpc}\), 0.12–0.16 (\(\beta 0.4) / 0.075\)–0.10 (\(\beta 0.5) / 0.049\)–0.071 (\(\beta 0.6\)); r_c 30 gives 0.30–0.56 and r_c 50 gives 0.68–1.26 (units 1e-15 erg/cm\(^{2}\)/s/arcmin\(^{2}\)). Interpolation convergence: the linear/spline UL ratio is 0.77–0.82 (0.43–0.50 on A1), still outside the 10 % target — the looser spline values are quoted and the linear ones are the interpolation systematic (−18 to −23 %). Zhang's nominal median 0.8285 / UL for r_c \(\le 8 = 5.2\)–6.7 (\(\beta 0.4\)), 8.1–11.1 (\(\beta 0.5\)), 11.7–17.1 (\(\beta 0.6\)); the public wording is "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)" (red team RT-16 re-checked it: dropping 0800730601 loosens by +46…+62 % on the v19 surfaces but only 6–30 % under Model C, where that field falls to 4th of 14, so 3.5–5.2\(\times\) is quoted once Model C is included, not \(\gtrsim4\)\(\times\)). Red team RT-13 (triaged): the arithmetic and inputs hold, the A \(\ge 0\) minimum sits at A = 0 in all 39 configurations, so \(\Delta \mathrm{stat} = 2.71\) is the correct 50:50 quantile for a parameter bounded at zero; the limit leans on one field (0800730601) — leaving it out loosens by +46…+62 % (r_c 8, \(\beta 0.4: 0.161\) → 0.253, Zhang/UL 5.2 → 3.3); the quoted natural-spline UL is the loosest of four interpolants (de-duplicated/PCHIP/linear are 2–23 % tighter) and a kT-consistent flux conversion another \(\approx9.7 %\) tighter. Red team RT-08 (triaged): "the limits are well calibrated" does not survive — the toy-MC location-family model holds only under its simplification (coverage 0.78–0.88 with a Gaussian sampler), and while the \(\Delta \mathrm{stat} = 2.71\) rule itself is still right in the ideal Gaussian case the threshold is not yet calibrated by spectral simulation (E028 stage 0 was ruled invalid as run by RT-15, the correction list CORRECTION_01_RT15 has been sent, and stage 1 is blocked). RT-02's interpolation warning still stands: 2.0–2.3\(\times\) sensitive on A1, down to 0.77–0.82 on A1d. RT-07: exclude Zhang's nominal median profile only, never "the stacked expectation" (Zhang publishes only 16/84 % marginals, no covariance; the v19 APEC bridge gives 0.76–0.90). Historical values, for the record only: A2 v1 (0.047–0.164, toy-MC coverage 0.975–0.984) and the sensitivity variant A2 v2 (Model-C surfaces, for \(\beta = 0.5\) at r_c = 4/8/15/30 = 0.111/0.138/0.215/0.484, coverage 0.918–0.933) | ↗ |
| O2b 约束大核 M31 热晕（r_c \(\ge 50 \mathrm{kpc}\)，A2 够不到的区间）Constrain a large-core M31 hot halo (r_c \(\ge 50 \mathrm{kpc}\), the regime A2 cannot reach) | provisional | E022（外场参考）+ E023（联合拟合，22 个 v19 norm，高斯似然）：\(\beta = 0.5\) 时 20 kpc 单侧 95% 上限 0.46（r_c=50）／0.59（r_c=100）／0.90（r_c=200）；内场单独拟合在 r_c \(\ge 50 \mathrm{kpc}\) 无约束力（r_c=100 时 3.61，r_c=200 时 13.8）。Zhang 名义中值 0.83 被排除到 r_c \(\lesssim 125/175 \mathrm{kpc}\)（\(\beta = 0.4/0.5\)）；红队 RT-02（已分诊）：「\(\beta = 0.6\) 时 r_c \(\le 200\) 全部排除」已撤回——15% 外场斑块底噪给 \(\beta = 0.6\)、r_c = 200 的上限 0.90–1.06，该端边际；E023 还因单一流量/norm 转换（外场跨度 1.40–3.22e-10）而将在流量空间于谱曲面上重做（加 MW 纬度项与外场系统项）。EAGLE fiducial 0.51 到 r_c \(\lesssim 45/65/85 \mathrm{kpc}\)（\(\beta = 0.4/0.5/0.6\)）。E022 的内禀 norm 外/内 = 0.80 \(\pm 0.13\)（两组样本统计均匀），平 M31 项 \(\lesssim 0.4\)–0.46（2\(\sigma\)，RT-03 后两种写法并列）。E025（D50 的 P1+P2）把外样本扩到 26 个已拟合方向（23 个有双侧误差）：固定效应干净外/内 = 0.877 \(\pm 0.072\)；红队 RT-03（已分诊）：随机效应 0.94 \(\pm 0.11\)，逐一剔除 0.83–0.99——稳妥说法「外场与内场在约 10–15% 内一致」，「外场更暗」不稳健E022 (the outer reference) + E023 (joint fit of the 22 v19 norms, Gaussian likelihood): for \(\beta = 0.5\) the one-sided 95% UL at 20 kpc is 0.46 (r_c=50) / 0.59 (r_c=100) / 0.90 (r_c=200); inner-only fits are uninformative at r_c \(\ge 50 \mathrm{kpc} (3.61\) at r_c=100, 13.8 at r_c=200). The Zhang nominal median 0.83 is excluded up to r_c \(\lesssim 125/175 \mathrm{kpc}\) (\(\beta = 0.4/0.5\)); red team RT-02 (triaged): "for all r_c \(\le 200\) at \(\beta = 0.6\)" is withdrawn — a 15% outer patchiness floor gives a \(\beta = 0.6, \mathrm{r_c} = 200\) limit of 0.90–1.06, so that end is marginal; E023 is also to be redone in flux space on spectral surfaces (with an MW latitude term and an outer systematic) because of its single flux/norm conversion (the outer fields span 1.40–3.22e-10). The EAGLE fiducial 0.51 up to r_c \(\lesssim 45/65/85 \mathrm{kpc}\) (\(\beta = 0.4/0.5/0.6\)). E022's intrinsic norm outer/inner = 0.80 \(\pm 0.13\) (both samples statistically homogeneous), and a flat M31 term is \(\lesssim 0.4\)–0.46 (2\(\sigma\), both wordings after RT-03). E025 (D50's P1+P2) has grown the outer sample to 26 fitted directions (23 with two-sided errors): fixed-effect clean outer/inner = 0.877 \(\pm 0.072\); red team RT-03 (triaged): random effects 0.94 \(\pm 0.11\), leave-one-out 0.83–0.99 — the safe statement is "outer and inner agree within ~10–15%", and "the outer fields are fainter" is not robust | ↗ |
| O3 外 NW 亮场的起源Origin of the bright outer-NW fields | E014 + E018a + E020 + RT-01 + RT-06 done | 红队 RT-01 分诊（2026-10-15，Claude 复核全部 6 项）：E020 模型 C 的「NW 超出大半是固定 SP 谱形伪迹」不成立——−15% 的 IVW 移动中 v19 重收敛（A 步）贡献 −7.65%、SP 谱形（A→B）−5.29%、线能量（B→C）−2.87%；用共同的模型 C 误差时 NW/其余比值 1.345（v19）→ 1.133（A）→ 1.134（C）——下降发生在重收敛，不在 SP 谱形；等权重下 1.381 → 1.274（A）→ 1.341（C），模型 C 内 NW 超出仍在。I033 第 5 条回到 Questioned；D49 原文生效（0.25 未解释场级系统误差）；D49 更新撤回；模型 C 只是敏感性变体。红队 RT-06 分诊（2026-10-16，Claude 复算 4/4）：D49 原文的措辞收窄——RASS 只排除宽的 R45 亮超出（足迹稀释：28% 足迹内超出稀释到 1.09–1.14，观测 0.91 \(\pm 0.12\)，\(\approx1.8\)\(\sigma\)），不排除足迹尺度结构或弱映射进 R45 的谱；OMNI 只排除 OMNI 追踪的高通量 SWCX（日球层 SWCX 未检验）；NW 四场的 SP/QPB 比值约为其余十场的 3\(\times\)（3.06\(\times\)，事后 p \(\approx 0.005\)–0.008）——「未解释」倾向软质子。E014（D49）：RASS 比值 0.91 \(\pm 0.12\) vs XMM 1.28；OMNI 无窗口超 4e8 cm\(^{-2}\) s\(^{-1}\)；I034（ParetoGTICal 交叉核对）：v19 用 2.1.filter GTI，比 2.filter 更严，NW 四场没有被欠过滤——剩余耀斑时间不是解释；稳态 SP 对 GTI 不可见，E018b 预期变化不大Red team RT-01 triaged (2026-10-15, Claude reproduced all 6 checks): E020's claim that "the NW excess is mostly a fixed-SP-shape artefact" does not hold — of the −15% IVW shift, the v19 re-convergence (step A) gives −7.65%, the SP shape (A→B) −5.29%, the line energies (B→C) −2.87%; with common Model-C errors the NW/other ratio goes 1.345 (v19) → 1.133 (A) → 1.134 (C) — the drop happens at re-convergence, not at the SP shape; with equal weights 1.381 → 1.274 (A) → 1.341 (C), so the NW excess survives under Model C. I033 point 5 is back to Questioned; D49 stands as written (the 0.25 unexplained field-level systematic); the D49 update is withdrawn; Model C is a sensitivity variant only. Red team RT-06 triaged (2026-10-16, Claude reproduced 4/4): D49's wording is narrowed — RASS excludes only a broad R45-bright excess (footprint dilution: a 28% excess inside the footprints dilutes to 1.09–1.14 against the observed 0.91 \(\pm 0.12\), \(\approx1.8\)\(\sigma\)), not footprint-scale structure or a spectrum that maps weakly into R45; OMNI excludes only OMNI-traced high-flux SWCX (heliospheric SWCX is untested); the NW four's SP/QPB ratio is about 3\(\times\) the other ten (3.06\(\times\), a posteriori p \(\approx 0.005\)–0.008) — "unexplained" leans towards soft protons. E014 (D49): RASS ratio 0.91 \(\pm 0.12\) vs XMM 1.28; no OMNI window above 4e8 cm\(^{-2}\) s\(^{-1}\); I034 (the ParetoGTICal cross-check): v19 uses the 2.1.filter GTIs, stricter than 2.filter, and the NW four are not under-filtered — residual flare time is not the explanation; a steady SP floor is invisible to GTIs, so E018b is expected to change little | ↗ |
| O4 银河系前景模型 + Local Hot Bridge 检验（300 kpc 图）MW foreground model + Local Hot Bridge test (300-kpc map) | E021/E022/E023/E024/E025 完成；E026 批处理运行中（产品暂不可比，RT-22/R015）E021/E022/E023/E024/E025 done; E026's batch is running (products not comparable yet, RT-22/R015) | E021 清点 + E022（8 个 class-A 外场拟合）之后，D50（PI，2026-10-15「同意第二个」）采纳 E024 建议：P1+P2 立即拟合（E025），6 个真正外围场上 Virgo 重提取，跳过 P4，弃 NGC 147（仅剩 22% 视场）。E025 结果：P1 12/12 未通过掩膜审计（现有 v19 谱的掩膜只盖住引文展源圆的 0–9%），全部转入重提取清单；P2 19 场拟了 18 个（1 个 dof=0 剔除），其中 3 个无有限通量误差。干净外样本（13 场，无未掩展源）IVW 0.846 \(\pm 0.059\)（\(\chi^{2} 27.3/12\)）vs 内场 0.965 \(\pm 0.044\)（v19 单起始值，D48 冻结；引用总量已由 D57 改为 0.867 \(\pm 0.044\)（统计），其 NH 项与本地热泡分配项见上表——旧的 \(\pm 0.034\) 固定归一化 NH 项已由 D65 撤回；此处保留 D57 当时的 \(\pm 0.034\)（NH）写法仅作历史记录，外/内比值仍基于 v19 逐场流量），固定效应外/内 = 0.877 \(\pm 0.072\)；全部 23 场（含未掩展源）0.831 \(\pm 0.045\)；与 b（\(\rho = 0.04\)/−0.10）、N_H（\(\rho\) = −0.13/−0.19）均无相关；未掩展源不抬高流量（b-only 0.865 \(\pm 0.133\) vs a+b 0.809 \(\pm 0.069\)）。红队 RT-03（已分诊）：干净 IVW 有 27% 权重集中在一个场（0884992601）且过度离散（\(\chi^{2}/\mathrm{dof} 2.27\)）——随机效应外/内 = 0.94 \(\pm 0.11\)（全部 23 场 0.89 \(\pm 0.08\)），逐一剔除 0.83–0.99；稳妥说法是「外场与内场在约 10–15% 内一致」，「外场更暗」不稳健；误差模型敏感（按剖面半宽重加权把 E022 均值移到 0.897，占干净差距的 46%）。correction 1：10 个 P2 a+b 场也未通过掩膜审计（MOS 最低 0–0.066）→ 重提取清单 = 6（D50）+ 12（P1）+ 10（a+b）= 28 场（E026，运行中，待定）。全部新增方向仍在 b < 0（−33.2 … −9.6），无正纬度覆盖。Locatelli 模型在 M31 足迹投影 1.290 vs 测量 0.867（残差 −0.42，有条件中心值张力——计入模型自身误差 \(\sigma \approx 0.32\) 后差距只有约 1.3\(\sigma\)、2\(\sigma\) 内允许正的 M31 残差，R027/RT-29 2026-10-17 18:3xZ；此前 RT-04 的「有条件紧张」基于 v19 单起始值 0.965）；Ueda 0.312、HaloSat 0.565。红队 RT-09/RT-09-T1/RT-12（已分诊）：外场流量随观测年代下降——干净 13 场 2009 年及以前（n = 10）1.05 \(\pm 0.08\) vs 2010 年以后（n = 3）0.55 \(\pm 0.09\)，而内场全部是 2017–2018 年；年代分割约 2\(\sigma\)（置换 p = 0.027），2001–2009 的下降可能部分是太阳周期 SWCX，2010 年后的偏低不是。红队 RT-14（luna + Hermes 臂均已分诊）：原因未定——「怀疑探测器老化」已撤回（公布的 MOS 校准只有几个百分点；拟合实际使用的 82 个 ARF 逐一核对哈希无误，其软端有效面积在 2010 年前后只变 +1.6%/−3.7%，离解释 43% 所需差 12–27 倍）；那条 1.3 keV 特征是自由的 1.2–1.4 keV 高斯残差（补偿 QPB 切除后残留的铝荧光，占该能段 17–38%），不是仪器荧光线示踪量；观测年份与探测器 CCD 状态高度共线（方差膨胀因子 20–39）。因此外/内比与基于外场的极限（E022/E023 大核上限）一律标为「受观测年代混杂」，直到 RT-14-T1（运行中）与 RT-09-T3（排队）出结果。红队 RT-11（阻断级）：E026 的 Virgo runner 会让全部 28 个场在审计这一步失败（3 处阻断级缺陷），CCF 也未锁定到 v19 集；修正单 CORRECTION_01_RT11.md 已备好并已在试点场上通过（O1–O12 关闭）。2026-10-17 05:22Z 批处理已在 Virgo 上启动（PI 授权；第 1 个场 0080540101，26 个场，预计 62–95 小时）。红队 RT-22（09:4xZ 分诊，R015）：runner 的自定义耀斑过滤步骤实际没起作用——试点场 mos1 目标谱曝光 11417.38 s vs v19 的 10228.74 s（+11.6%），而试点的 2.1.filter 与 2.filter 干净 GTI 完全相同（14 个区间、11630.0 s）；retained_fraction 拿 runner 自己的 2.filter 当基准，所以抓不到这一点，试点审计因此通过。D52：批处理继续跑（上游的校准、2.filter、edetect_stack 可复用），CORRECTION_02 已发（找回并恢复 v19 的自定义过滤器、从 2.1.filter 重跑试点、验收 = 目标谱曝光在 v19 2% 内且 GTI 重叠 \(\ge 98\)%）。重跑通过前不用 E026 的谱做任何拟合——E026 产品暂时不能与 v19 比较，所以外/内比与基于外场的极限继续标「受观测年代混杂」After the E021 inventory and E022 (8 class-A outer fields fitted), D50 (the PI, 2026-10-15, "agree with the second") adopts E024's suggestion: fit P1+P2 now (E025), re-extract 6 genuinely-outer fields on Virgo, skip P4, drop NGC 147 (only 22% of the field of view left). E025 results: all 12 P1 fields fail the mask audit (the existing v19 spectra's mask covers only 0–9% of the cited extended-target circle), all moving to the re-extraction list; 18 of 19 attempted P2 fits are accepted (1 rejected, dof = 0), 3 of them without a finite flux error. The clean outer sample (13 fields without an unmasked extended target) has IVW 0.846 \(\pm 0.059\) (\(\chi^{2} 27.3/12\)) vs the inner 0.965 \(\pm 0.044\) (the v19 single-start value frozen by D48; the quoted total is now 0.867 \(\pm 0.044\) (stat) under D57, with its N_H and LHB-allocation terms in the table above — the old \(\pm 0.034\) fixed-normalisation NH term is withdrawn by D65; D57's original \(\pm 0.034\) (NH) wording is kept here as a historical record only, and the outer/inner ratio still rests on the v19 per-field fluxes), fixed-effect outer/inner = 0.877 \(\pm 0.072\); all 23 fields (including unmasked targets) give 0.831 \(\pm 0.045\); no trend with b (\(\rho = 0.04\)/−0.10) or N_H (\(\rho\) = −0.13/−0.19); unmasked extended targets do not raise the flux (b-only 0.865 \(\pm 0.133\) vs a+b 0.809 \(\pm 0.069\)). Red team RT-03 (triaged): the clean IVW mean carries 27% of its weight in one field (0884992601) and is overdispersed (\(\chi^{2}/\mathrm{dof} 2.27\)) — random effects give outer/inner = 0.94 \(\pm 0.11\) (all 23 fields 0.89 \(\pm 0.08\)), leave-one-out 0.83–0.99; the safe statement is "outer and inner agree within ~10–15%", and "the outer fields are fainter" is not robust; the error model matters (reweighting by profile half-widths moves E022's mean to 0.897, 46% of the clean gap). Correction 1: the 10 P2 a+b fields also fail the mask audit (MOS minimum 0–0.066) → the re-extraction list = 6 (D50) + 12 (P1) + 10 (a+b) = 28 fields (E026, running, Pending). All added directions remain at b < 0 (−33.2 … −9.6); no positive-latitude coverage. The Locatelli model projected at the M31 footprint is 1.290 vs the measured 0.867 (residual −0.42, conditional central-value tension — with the model's own error \(\sigma \approx 0.32\) the gap is only about 1.3\(\sigma\) and a positive M31 residual is allowed at 2\(\sigma\), R027/RT-29 2026-10-17 18:3xZ; RT-04's earlier "conditional tension" was based on the v19 single-start value 0.965); Ueda 0.312, HaloSat 0.565. Red teams RT-09/RT-09-T1/RT-12 (triaged): the outer-field flux falls with observing year — in the clean 13, 10 fields up to 2009 give 1.05 \(\pm 0.08\) vs 3 fields from 2010 on giving 0.55 \(\pm 0.09\), while the inner fields are all from 2017–2018; the epoch split is about 2\(\sigma\) (permutation p = 0.027), part of the 2001–2009 decline may be solar-cycle SWCX, and the post-2010 deficit is not. Red team RT-14 (both the luna and Hermes arms triaged): the cause is unresolved — "detector ageing suspected" is withdrawn (the published MOS calibration is only a few percent; all 82 ARFs actually used in the fits are hash-verified and their soft effective area changes by only +1.6 %/−3.7 % across 2010, a factor 12–27 short of what a 43 % drop needs); the 1.3 keV feature is a free 1.2–1.4 keV Gaussian residual (absorbing aluminium fluorescence left after the QPB cut-out, 17–38 % of that band), not an instrumental-fluorescence tracer; observation year is strongly collinear with detector CCD state (variance inflation 20–39). Therefore the outer/inner ratio and the outer-based limits (E022/E023's large-core ULs) are all marked "epoch-confounded" until RT-14-T1 (running) and RT-09-T3 (queued) report. Red team RT-11 (blocking): E026's Virgo runner would fail all 28 fields at the audit step (3 blocking defects) and its CCF is not pinned to the v19 set; the correction list CORRECTION_01_RT11.md is ready and passed on the pilot field (O1–O12 closed). The batch launched on Virgo at 2026-10-17 05:22Z (PI-authorised; first field 0080540101, 26 fields, an expected 62–95 h). Red team RT-22 (triaged 09:4xZ, R015): the runner's custom flare-filter step did nothing — the pilot field's mos1 obj.pi exposure is 11417.38 s against v19's 10228.74 s (+11.6 %), while the pilot's 2.1.filter and 2.filter clean GTIs are identical (14 intervals, 11630.0 s); retained_fraction is measured against the runner's own 2.filter, so it cannot catch this, which is why the pilot audit passed. D52: the batch keeps running (the upstream calibration, 2.filter and edetect_stack steps are reusable) and CORRECTION_02 has been sent (find and restore v19's custom filter, re-run the pilot from 2.1.filter, acceptance = obj exposure within 2 % of v19 and GTI overlap \(\ge 98\) %). Until that rerun passes no fit uses any E026 spectrum — E026's products are not v19-comparable for now, so the outer/inner ratio and the outer-based limits stay marked "epoch-confounded" |
↗ |
| O5 New-ANGELS III 论文The New-ANGELS III manuscript | draft | 16 页内部稿完成（AASTeX，5 表 3 图）；提交门槛见 I013（径向+侧向联合似然、边界感知区间、系统误差网格等）A 16-page internal draft exists (AASTeX, 5 tables, 3 figures); submission gates in I013 (shared radial-plus-side likelihood, boundary-aware intervals, the systematic grid, ...) | ↗ |
来源：O1 cognition-os/insights/I012 与 I033，另加 2026-10-18 的毛总量与权重分解（R033/D60）、CXB 谱指数与 Z 杠杆（R034/R035/D61）、总柱密度/NH 项（R036/D62）、NH 重拟合与引用总量改写（R039/D65）、本地热泡分配项（R038）、放开谱指数后移走流量的空间平坦性与撤回（R037）；O2a cognition-os/experiments/E013_joint_spatial_profile_likelihood.md（「A1d accepted and A2 baseline on A1d」「Red-team RT-08 triage」「Red-team RT-13 triage」节；插值表 a2_ul_vs_rc_interp_A1d.csv 与 a2_ul_vs_rc_interp_A1d_linear.csv）、cognition-os/insights/I033 末两段、cognition-os/REDTEAM_QUEUE.md 第 15、26 行（RT-02 的 F3 发现仍在）；O2b cognition-os/experiments/E022（correction 1 节）、E023（limits.csv）与 E025（correction 1 节；correction1_ivw_table.csv）；O3 cognition-os/experiments/E020（「Red-team RT-01 triage」节）、E014、cognition-os/insights/I034、cognition-os/04_DECISION_LOG.md D49 及其 update 节（撤回）、cognition-os/REDTEAM_QUEUE.md；O4 cognition-os/experiments/E021（inventory.csv）、E024（policy_table.csv）、E025（RT-09、RT-09-T1、RT-12、RT-09-T2、RT-14 与 Hermes 臂修订节）、.agent/jobs/2026-10-15-e026-virgo-reextract-stage0/CORRECTION_01_RT11.md、cognition-os/04_DECISION_LOG.md D50、newangels3_discussion/E024_hermes/latitude_reach.csv、00_AI_BRIEFING.md（D42–D46 数值）与 02_CURRENT_QUESTION.md；O5 cognition-os/insights/I013 与 paper_apj_v19_cgmsum_draft/README_draft_status_and_submission_gaps.md。
Source: O1 cognition-os/insights/I012 and I033, plus the 2026-10-18 gross-total and weight decomposition (R033/D60), the CXB photon index and Z lever (R034/R035/D61), the total column / NH term (R036/D62), the NH refit and the rewritten quoted total (R039/D65), the LHB allocation term (R038), and the spatial flatness of the flux removed by freeing the photon index plus its withdrawal (R037); O2a the "A1d accepted and A2 baseline on A1d", "Red-team RT-08 triage" and "Red-team RT-13 triage" sections of cognition-os/experiments/E013_joint_spatial_profile_likelihood.md (the interpolated tables a2_ul_vs_rc_interp_A1d.csv and a2_ul_vs_rc_interp_A1d_linear.csv), the last two paragraphs of cognition-os/insights/I033, and rows 15 and 26 of cognition-os/REDTEAM_QUEUE.md (RT-02's F3 finding still stands); O2b cognition-os/experiments/E022 (the correction-1 section), E023 (limits.csv) and E025 (the correction-1 section; correction1_ivw_table.csv); O3 cognition-os/experiments/E020 (the "Red-team RT-01 triage" section), E014, cognition-os/insights/I034, cognition-os/04_DECISION_LOG.md D49 with its update section (withdrawn), cognition-os/REDTEAM_QUEUE.md; O4 cognition-os/experiments/E021 (inventory.csv), E024 (policy_table.csv), E025 (the RT-09, RT-09-T1, RT-12, RT-09-T2 and RT-14 sections and the Hermes-arm amendment), .agent/jobs/2026-10-15-e026-virgo-reextract-stage0/CORRECTION_01_RT11.md, cognition-os/04_DECISION_LOG.md D50, newangels3_discussion/E024_hermes/latitude_reach.csv, cognition-os/00_AI_BRIEFING.md (the D42–D46 numbers) and 02_CURRENT_QUESTION.md; O5 cognition-os/insights/I013 and paper_apj_v19_cgmsum_draft/README_draft_status_and_submission_gaps.md.
未解问题表Open problems
| 问题Problem | 为什么重要Why it matters | 当前最佳假设Best current hypothesis | 下一个检验Next test | 负责Owner | 状态Status | |
|---|---|---|---|---|---|---|
| MW 与 M31 气体同温同形，EPIC 谱无法按距离分解（I006）MW and M31 gas share temperature and spectral shape; EPIC spectra cannot separate them by distance (I006) | 决定论文能声明什么：只能测总和 CGMsum，任何分量归属都需要外部假设Determines what the paper can claim: only the sum CGMsum is measured; any component split needs external assumptions | 分离只能来自空间形状（半径衰减、南北对称）与外部前景约束Separation can only come from spatial shape (radial decline, N/S symmetry) and external foreground constraints | A2 基线已在 A1d 上重算（见状态页插值表）：集中核（r_c \(\le 8 \mathrm{kpc}\)）20 kpc 上限 0.12–0.16（\(\beta 0.4\)）／0.075–0.10（\(\beta 0.5\)）／0.049–0.071（\(\beta 0.6\)）；红队 RT-13 已分诊——算术与输入成立、\(\Delta \mathrm{stat} = 2.71\) 是「参数不能为负」边界情形下正确的分位数，但上限主要靠一个视场（仅 v19 曲面；红队 RT-16 复核：模型 C 曲面下该集中度基本消失），对外措辞「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」；红队 RT-08 已分诊——toy-MC 校准只在简化模型下有效，阈值尚未用能谱模拟校准（E028 stage 0 已被接受（R022/D56，2026-10-17 15:5xZ），但阈值仍未校准，stage 1 需 PI 决定）；插值仍未收敛（线性/样条 0.77–0.82，RT-02 在 A1 上的 2.0–2.3 倍警告仍在）The A2 baseline has been recomputed on A1d (see the interpolated table on the state page): the compact-core (r_c \(\le 8 \mathrm{kpc}\)) limit at 20 kpc is 0.12–0.16 (\(\beta 0.4) / 0.075\)–0.10 (\(\beta 0.5) / 0.049\)–0.071 (\(\beta 0.6\)); red team RT-13 is triaged — the arithmetic and inputs hold and \(\Delta \mathrm{stat} = 2.71\) is the correct quantile for a parameter bounded at zero, but the limit leans on one field on the v19 surfaces only (red team RT-16 re-checked it: under the Model C surfaces that concentration largely disappears), with the public wording "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)"; red team RT-08 is triaged — the toy-MC calibration holds only under its simplified model, so the threshold is not yet calibrated by spectral simulation (E028 stage 0 is accepted (R022/D56, 2026-10-17 15:5xZ), but the threshold is still uncalibrated and stage 1 needs a PI decision); interpolation is still not converged (linear/spline 0.77–0.82, RT-02's 2.0–2.3\(\times\) warning on A1 still stands) | E013 | A1d/A2 基线已出（临时）；RT-08/RT-13 已分诊；E028 校准运行中A1d/A2 baseline delivered (provisional); RT-08/RT-13 triaged; E028's calibration running | |
| 轮廓平（\(\chi^{2}=11.99/13\)；E020 模型 C 下同样平，斜率 −0.0005 \(\pm 0.006 /\mathrm{kpc}\)）：M31 集中晕（r_c \(\le 8 \mathrm{kpc}\)）在 20 kpc \(\le\)0.049–0.16（A2 基线在 A1d 上，插值版，\(\beta 0.6\)…0.4；阈值未校准），远低于 Zhang 名义中值剖面 0.83The profile is flat (\(\chi^{2}=11.99/13\); still flat under E020's Model C, slope −0.0005 \(\pm 0.006 /\mathrm{kpc}\)): a concentrated M31 halo (r_c \(\le 8 \mathrm{kpc}\)) is \(\le\)0.049–0.16 at 20 kpc (the A2 baseline on A1d, interpolated, \(\beta 0.6\)…0.4; threshold uncalibrated), far below the Zhang nominal median profile 0.83 | M31 内晕热气要么比 Zhang 名义中值剖面暗几倍，要么在 10–30 kpc 内分布很平（核 \(\gtrsim30 \mathrm{kpc}\)）M31's inner-halo hot gas is either several times fainter than the Zhang nominal median profile, or flat over 10–30 kpc (core \(\gtrsim30 \mathrm{kpc}\)) | 「平/大核」这条退路一度被限制，但该上限现在被质疑（R023，2026-10-17 15:5xZ）：E023 用 8 个 82–253 kpc 外场把大核晕压到 0.46（r_c=50）／0.59（r_c=100）／0.90（r_c=200）@20 kpc（\(\beta=0.5\)），但外场受观测年代混杂——加一个年代项后上限变成 0.87/1.29/2.38，且数据反而偏好有 M31 项；只用 2010 年后的外场也类似。在匹配年代的外场重提取之前，这些上限与「Zhang 名义中值被排除到 r_c 125–200 kpc」一律不引用（小核端上限不受影响）。「\(\beta = 0.6\) 时 r_c \(\le 200\) 全部排除」已被红队 RT-02 撤回（15% 外场斑块底噪给 \(\beta = 0.6\)、r_c = 200 的上限 0.90–1.06，该端边际）；E025 再把外样本扩到 26 个已拟合方向（23 个有双侧误差）（干净外/内 0.877 \(\pm 0.072\)；RT-03 后稳妥说法是「约 10–15% 内一致」）；r_c = 50 时上限 0.318–0.601（A2 v1，内场单独）已被联合拟合取代the "flat/large-core" escape route was limited for a while, but those limits are now Questioned (R023, 2026-10-17 15:5xZ): using the 8 outer fields at 82–253 kpc, E023 pushed the large-core halo to 0.46 (r_c=50) / 0.59 (r_c=100) / 0.90 (r_c=200) at 20 kpc (\(\beta=0.5\)), but the outer fields are epoch-confounded — adding an epoch term turns the limits into 0.87/1.29/2.38 and the data then prefer an M31 term; post-2010 outer fields alone behave similarly. Until matched-epoch outer fields are re-extracted, neither these limits nor "the Zhang nominal median is excluded to r_c 125–200 kpc" is quoted (the compact-core limits are unaffected). "For all r_c \(\le 200\) at \(\beta = 0.6\)" was withdrawn by red team RT-02 (a 15% outer patchiness floor gives a \(\beta = 0.6, \mathrm{r_c} = 200\) limit of 0.90–1.06, so that end is marginal); E025 has since grown the outer sample to 26 fitted directions (23 with two-sided errors) (clean outer/inner 0.877 \(\pm 0.072\); after RT-03 the safe wording is "agree within ~10–15%"); the A2 v1 inner-only limit at r_c = 50 (0.318–0.601) is superseded by the joint fit | A2 基线（A1d 曲面）已复核集中核上限（RT-13 已分诊）；E028 用能谱模拟校准 \(\Delta \mathrm{stat} = 2.71\) 阈值（stage 0 已被接受（R022/D56，2026-10-17 15:5xZ：CORRECTION_02 后闭合 9/9 达到或低于 A1d），但阈值仍未校准；stage 1 需要 PI 决定——4 个 lang worker 每个真值 17–19 天，T0/T1/rc30 三个真值 51–57 天，不是「阻断」）；E023 在匹配年代的谱曲面上重做大核上限（加 MW 纬度项）；E016 的未解源项空间上平，不削弱形状上限the A2 baseline (on the A1d surfaces) has re-checked the compact-core limit (RT-13 triaged); E028 calibrates the \(\Delta \mathrm{stat} = 2.71\) threshold with spectral simulation (stage 0 is accepted (R022/D56, 2026-10-17 15:5xZ: after CORRECTION_02 the closure is 9/9 at or below A1d), but the threshold is still uncalibrated; stage 1 needs a PI decision — 17–19 days per truth on 4 lang workers, 51–57 days for T0/T1/rc30, not "blocked"); E023 is redone on matched-epoch spectral surfaces for the large-core limit (with an MW latitude term); E016's unresolved-source terms are spatially flat, so they do not weaken the shape limit | E013 | provisional | |
| 外 NW 四场偏亮（v19 外四/其余十 = 1.28）The four outer-NW fields are bright (v19 outer-four/other-ten = 1.28) | 若为银河系纬度结构，会挤占 M31 的解释空间；若为 SWCX 则是时间效应；若为仪器/软质子则影响系统误差预算If it is Galactic-latitude structure it eats into the M31 budget; if SWCX it is a time effect; if instrumental/soft-proton it changes the systematic budget | 红队 RT-01 分诊后回到「未解释」：−15% 的 IVW 移动一半来自 v19 重收敛（0800730301 通量 0.975 → 0.559，\(\Delta \mathrm{stat}\) 仅 −0.04），固定 SP 谱形只贡献 −5.29%；用共同模型 C 误差时 NW/其余 1.345 → 1.133（A）→ 1.134（C）——下降在重收敛，不在 SP 谱形；等权重下模型 C 内仍 1.34。E014 已排除大尺度天空（RASS 0.91 \(\pm 0.12\) vs XMM 1.28）与 OMNI 太阳风；I034：NW 四场没有被 GTI 欠过滤（2.1.filter 比 2.filter 严，NW 四场仅损失 0–28% 时间），稳态 SP 对时间筛选不可见。D49 原文生效（0.25 未解释场级系统误差）；D49 更新（−15% SP 谱形系统误差）已撤回；模型 C 只是敏感性变体back to "unexplained" after red team RT-01's triage: half of the −15% IVW shift comes from re-converging the v19 fits (0800730301's flux 0.975 → 0.559 at \(\Delta \mathrm{stat}\) only −0.04), the fixed SP shape contributes just −5.29%; with common Model-C errors NW/other goes 1.345 → 1.133 (A) → 1.134 (C) — the drop is at re-convergence, not the SP shape; with equal weights it stays 1.34 under Model C. E014 excluded large-scale sky (RASS 0.91 \(\pm 0.12\) vs XMM 1.28) and OMNI solar wind; I034: the NW four are not under-filtered by the GTIs (2.1.filter is stricter than 2.filter; the NW four lose only 0–28% of live time), and a steady SP floor is invisible to time filtering. D49 stands as written (the 0.25 unexplained field-level systematic); the D49 update (a −15% SP-shape systematic) is withdrawn; Model C is a sensitivity variant only | A1d 曲面重跑 A2 基线；红队 RT-02/RT-03 分诊；E018b（Virgo 更严耀斑过滤）预期变化不大（I034），保持可选rerun the A2 baseline on the A1d surfaces; triage red teams RT-02/RT-03; E018b (a stricter Virgo flare filter) is expected to change little (I034) and stays optional | E020/RT-01/PI | 回到未解释；红队进行中back to unexplained; red team running | |
| v19 拟合 QA：SP 谱形过刚、仪器线能量钉边（E018a/E013）——E020 已量化v19 fit QA: the SP shape was too rigid and line energies pinned (E018a/E013) — now quantified by E020 | 决定系统误差网格里必须有哪一行（I013 提交门槛 3）；不修会以场级伪迹污染科学结论Decides which row the systematic grid must contain (I013 submission gate 3); unfixed, it contaminates science with field-level artefacts | E020（20 场，多元起始）：SP 谱形自由在 0401/0501/0901/1101/2101 改善 8–16；线能量自由在 0401 改善 8.5、1001 改善 15。模型 C 下 14 场逆方差均值 0.964 → 0.819 \(\pm 0.041\)。红队 RT-01 分解：−15.04% 中重收敛（A）−7.65%、SP 谱形（A→B）−5.29%、线能量（B→C）−2.87%——不是一个 SP 谱形系统误差；模型 C 的 SP 谱形未经物理验证（13/14 场 SP 升、MWhalo 降，log-ratio 相关 −0.49；0800730901 的 SP \(\times1150\) 且 PhoIndx2 钉在极限）。模型 C 的剖面误差在固定 kT 下算得（低估），4 场半宽仅 0.15–0.27 \(\sigma_\mathrm{v19}\)；残差散布 \(\chi^{2} 24.5/13\)（p = 0.026），单个 −15% 标量不能替代场级项。2026-10-18 R034/R035/D61 新增两个流量杠杆：放开 CXB 谱指数使 CGMsum 降 6.2%（13/14 场下降，AIC 更优）；金属丰度 Z = 0.3 也使流量随 Z 0.1/1.0 变化 −6 至 −9%。放开谱指数后被移走的那部分流量在空间上是平的（外场 −7.3%、内场 −7.9%，与恒星质量分布无关，量级约为 M31 未分辨 X 射线双星预算的 360 倍）——「AGN 冒充热气体约 15–25%」的说法已由 R037 撤回，归属仍开放；下降的方向（13/14 场）成立，大小只有约 0.6\(\sigma\)，对总量的影响由 E029 判定。判定规则已在 E029 出任何结果前收紧（D63）：参考值 = A1d 曲面的连续最小值；同一组流量换不同权重会给 0.817 或 0.850，所以必须同时报告「同口径加权」的总量，两者都超过 0.044 才改引用值；模型比较只在原有参数边界内进行；若变化只来自少数几个场，则作为这些场的系统误差处理；新增的幂律成分是 CXB 还是 M31 未分辨 X 射线双星仍是开放问题。「仅气体 0.82」已撤回（R034/D61），不再作为名义值引用；E029 结果为待定E020 (20 fields, multi-start): freeing the SP shape improves 0401/0501/0901/1101/2101 by 8–16; freeing line energies improves 0401 by 8.5 and 1001 by 15. Under Model C the 14-field inverse-variance mean is 0.964 → 0.819 \(\pm 0.041\). Red team RT-01's decomposition: of −15.04%, re-convergence (A) gives −7.65%, the SP shape (A→B) −5.29%, the line energies (B→C) −2.87% — not a single SP-shape systematic; Model C's SP shapes are not physically validated (SP up, MWhalo down in 13/14 fields, log-ratio correlation −0.49; 0800730901's SP \(\times1150\) with PhoIndx2 at its limit). Model C's profile errors were computed at fixed kT (an underestimate); 4 fields have half-widths only 0.15–0.27 \(\sigma_\mathrm{v19}\); the residual scatter is \(\chi^{2} 24.5/13\) (p = 0.026), so a single −15% scalar cannot replace the field-to-field term. Two further flux levers added on 2026-10-18 by R034/R035/D61: freeing the CXB photon index lowers the CGMsum by 6.2% (13/14 fields fall, AIC-preferred), and the metal abundance Z = 0.3 moves the flux by −6 to −9% between Z 0.1 and 1.0. The flux removed by freeing the photon index is spatially flat (outer −7.3%, inner −7.9%, unrelated to the stellar-mass distribution, and about 360 times the unresolved M31 X-ray-binary budget) — the "15–25% AGN mimic" statement is withdrawn by R037 and the attribution stays open; the direction of the drop (13/14 fields) holds, its size is only about 0.6\(\sigma\), and E029 decides the effect on the total. The decision rule was tightened before E029 produced any result (D63): the reference is the continuous minimum of the A1d surfaces; the same fluxes give 0.817 or 0.850 depending on the weights, so a like-for-like weighted total must also be reported, and the quote moves only if both differ from it by more than 0.044; models are compared within the original parameter bounds; a shift coming from only a few fields is treated as a systematic of those fields; whether the extra power-law term is the CXB or unresolved M31 X-ray binaries remains open. "Gas-only 0.82" is withdrawn (R034/D61) and is no longer quoted as a nominal; E029's result is Pending | A1d（v19 模型、每点多元起始）曲面成为 A2 基线（运行中）；模型 C 曲面（A1c）降级为敏感性变体；D49 更新撤回——系统误差预算继续携带场级未解释项the A1d (v19-model, multi-start at every point) surfaces become the A2 baseline (running); the Model-C surfaces (A1c) are demoted to a sensitivity variant; the D49 update is withdrawn — the systematic budget keeps carrying the unexplained field-level term | E020 | E020 完成E020 done | |
| PSF 翼泄漏：掩膜亮源的翼落在提取区里（E016）PSF-wing leak: the wings of masked bright sources fall inside the extraction region (E016) | E016 里它是最大的单一未解源项（模型估计 \(\approx0.12 \mathrm{F}\) [最大 0.32]，主要是幂律；E019 里记的「高括号 0.22」已被 0.32 取代）；若实测偏大，会压低「仅气体」的总和并改变系统误差预算In E016 it is the single largest unresolved-source term (model estimate \(\approx0.12 \mathrm{F} [max 0.32\)], mostly power law; E019's "high 0.22" is superseded by 0.32); if the empirical value is larger it lowers the gas-only total and changes the systematic budget | E019（已完成）：123 个源相机测量里的 81 个孤立类点源叠加——只有最亮 bin（中位 1.2e-14 cgs）可用：f_leak = 10.3 \(\pm 8.6%\)（原值）／20.4 \(\pm 7.8%\)（零点校正）；King 模型对同样椭圆给 14.3%。14 主场的逐场泄漏 SB：0.16（原值）／0.33（零点校正） vs E016 模型 0.12（高括号 0.22）；共享标度误差 \(\approx\) \(\pm0.14\)。泄漏是幂律型、空间上平——不改变 M31 形状上限；对总量是又一个向下项（主要落在幂律/CXB 项，不在 APEC）E019 (done): a stack of 81 isolated point-like sources out of 123 source-camera measurements — only the brightest bin (median 1.2e-14 cgs) is usable: f_leak = 10.3 \(\pm 8.6%\) raw / 20.4 \(\pm 7.8%\) null-corrected; the King model gives 14.3% for the same ellipses. Per-field leak SB over the 14 primaries: 0.16 raw / 0.33 null-corrected vs E016's model 0.12 (high bracket 0.22); the shared-scale error is \(\approx\) \(\pm0.14\). The leak is power-law-like and spatially flat — it does not change the M31 shape limit; for the total it is one more downward term (mostly in the power-law/CXB term, not the APEC) | 暗 bin 低于零点底噪（椭圆外净计数为负），不可用；2801（E016 异常掩膜标记）在 1.06 > F，0101 在 0.50 \(\approx \mathrm{F}\)the faint bins sit below the background floor (net counts out of the ellipse are negative), unusable; 2801 (an anomalous mask, E016 flag) is at 1.06 > F and 0101 at 0.50 \(\approx \mathrm{F}\) | E019 | E019 完成E019 done | |
| Locatelli MW 模型在 M31 方向的中心值 1.29 > 测量 0.867（R027，2026-10-17 18:3xZ）The Locatelli MW model's central value 1.29 > the measured 0.867 toward M31 (R027, 2026-10-17 18:3xZ) | 有条件中心值张力（R027/RT-29）：计入模型自身误差（\(\sigma \approx 0.32\)）后差距只有约 1.3\(\sigma\)，2\(\sigma\) 内允许正的 M31 残差——不是排除。写作「该参考模型的中心值不给 M31 留正的余量」。待定：LHB 与 CGMsum 之间的分配数据定不住Conditional central-value tension (R027/RT-29): with the model's own error (\(\sigma \approx 0.32\)) included the gap is only about 1.3\(\sigma\), and a positive M31 residual is allowed at 2\(\sigma\) — not an exclusion. Write "the reference model's central value leaves no positive M31 residual". Pending: the data do not pin down the LHB/CGMsum split | 模型高估、线-宽带转换失配、或 v19 总和偏低（三种解释并列）Model over-prediction, line-to-band conversion mismatch, or a low-biased v19 sum (three live explanations) | 300 kpc RASS 图建立本地 MW 模型（O4）A 300-kpc RASS map to build a local MW model (O4) | 计划planned | 计划planned | |
| 吸收模型：总 X 射线柱密度比 HI4PI 高 19–31%；引用总量的 NH 项已由 D65 改为 \(+0.007/-0.003\)（重拟合），旧的 \(\pm 0.034\) 是固定归一化模型响应、已撤回（R036/D62 → R039/D65，2026-10-18）Absorption model: the total X-ray column is 19–31% above HI4PI; the quoted total's NH term is now \(+0.007/-0.003\) (refit) under D65, and the old \(\pm 0.034\) fixed-normalisation response is withdrawn (R036/D62 → R039/D65, 2026-10-18) | 柱密度直接缩放发射流量——若真实柱密度高于 HI4PI，固有流量与 MW 前景份额都会变，Locatelli 紧张的含义也跟着变The column scales the emitted flux directly — if the true column is above HI4PI, both the intrinsic flux and the MW foreground share change, and so does the meaning of the Locatelli tension | 拟合用的是 HI4PI 的原子氢柱密度；真正吸收 X 射线的总柱密度（按 Willingale et al. 2013 加上分子氢）在这 14 个场高 19–31%。重拟合已完成（RT-39-T1，R039/D65）：把柱密度换成 Willingale 总柱密度后重新拟合，吸收后的 CGM 流量只变 +0.8%（\(\pm10\%\) 时不到 0.4%）；旧的 \(\pm 0.034\) 是固定归一化、不重新拟合算出的响应，已撤回；按 D62 的预定规则，NH 项改为 \(+0.007/-0.003\)。光谱本身定不出柱密度（自由柱密度拟合落在或低于 HI4PI 的 0.5 倍下界）——这是吸收流量对柱密度不敏感，不是柱密度测量The fits use HI4PI's atomic-hydrogen column; the total column that actually absorbs X-rays (adding molecular hydrogen per Willingale et al. 2013) is 19–31% higher in these 14 fields. The refit is done (RT-39-T1, R039/D65): refitting with the Willingale total column moves the absorbed CGM flux by only +0.8% (less than 0.4% for \(\pm10\%\)); the old \(\pm 0.034\) was a fixed-normalisation response computed without a refit and is withdrawn; under D62's pre-registered rule the NH term becomes \(+0.007/-0.003\). The spectra themselves do not determine the column (free-column fits sit at or below the 0.5 \(\times\) HI4PI bound) — this is insensitivity of the absorbed flux, not a column measurement | 无（RT-39-T1 已按 D62 规则落地为 D65 的 NH 项）；后续若 E029 采纳新的模型基准，NH 项随引用总量一起重算None (RT-39-T1 has landed in D65's NH term under the D62 rule); if E029 adopts a new model baseline, the NH term is recomputed with the quoted total | RT-39-T1/D65RT-39-T1/D65 | 已完成（D65）done (D65) | |
| 300 kpc 图：30 kpc 以外的档案样本很薄（E021）The 300-kpc map: the archival sample beyond 30 kpc is thin (E021) | O4（本地 MW 模型 + Local Hot Bridge 检验）需要 30 kpc 以外的视线；样本太薄会限制图的能力O4 (a local MW model + the Local Hot Bridge test) needs sightlines beyond 30 kpc; a thin sample limits what the map can do | lang 本地磁盘上的 145 个 ObsID 里 class-A 只有 22 个，30 kpc 以外仅 8 个（30–60 kpc 为零）；全部在 b = −5° 到 −33°，无正纬度场。红队 RT-18（R013）：该清单只覆盖本地磁盘，不是 XMM 档案——22.5° 锥的 XSA 查询列 563 个指向，其中 418 个本地没有（395 个带公开科学标记，344 个在 30 kpc 以外，305 个计划时长 >10 ks）；「没有遗漏可用观测」的说法已撤回。E022 已完成这 8 个场的拟合（外 IVW 0.842 \(\pm 0.065\)；内禀 norm 外/内 0.80 \(\pm 0.13\)）。D50 已定（2026-10-15）：P1+P2 立即拟合、6 场 Virgo 重提取、跳过 P4、弃 NGC 147。E025 已跑：P1 12/12 掩膜审计未过（掩膜仅盖展源圆 0–9%）→ 全部转重提取；P2 18/19 接受；干净外样本（13 场）0.846 \(\pm 0.059\)，外/内 0.877 \(\pm 0.072\)；与 b、N_H 无相关。correction 1 后重提取清单 = 6 + 12 + 10 = 28 场（E026 批处理 2026-10-17 05:22Z 起在 Virgo 上运行，产品暂不可比，结果待定）；4 个引文半径是估值，任何提升结果前需从图像重推 | Of the 145 ObsIDs on lang's local disk only 22 are class A, and just 8 lie beyond 30 kpc (zero at 30–60 kpc); all at b = −5° to −33°, no positive-latitude field. Red team RT-18 (R013): that inventory covers the local disk only, not the XMM archive — a 22.5° XSA cone lists 563 pointings, of which 418 are absent locally (395 carry a public science flag, 344 lie beyond 30 kpc, 305 have a scheduled duration >10 ks); the claim that no usable observation is missed is withdrawn. E022 has fitted those 8 fields (outer IVW 0.842 \(\pm 0.065\); intrinsic norm outer/inner 0.80 \(\pm 0.13\)). D50 is decided (2026-10-15): fit P1+P2 now, re-extract 6 fields on Virgo, skip P4, drop NGC 147. E025 has run: all 12 P1 fields fail the mask audit (the mask covers only 0–9% of the extended-target circle) → all move to re-extraction; 18/19 P2 fits accepted; the clean outer sample (13 fields) gives 0.846 \(\pm 0.059\), outer/inner 0.877 \(\pm 0.072\); no trend with b or N_H. After correction 1 the re-extraction list = 6 + 12 + 10 = 28 fields (E026's batch has been running on Virgo since 2026-10-17 05:22Z, its products not yet comparable, results Pending); 4 cited radii are estimates, to be re-derived from images before any promoted result | E026（28 场 Virgo 重提取，D50 条件：复现 v19 的 2.1.filter GTI 约定并逐相机报告相对 2.filter 的保留时间）→ 重提取场按 v19 规则拟合 → 在扩大样本上重做 E023 式联合拟合（加 MW 纬度项）；红队 RT-03 审查外/内比较。红队 RT-11（阻断级，已分诊）：runner 会让全部 28 个场在审计这一步失败（3 处阻断级缺陷）且 CCF 未锁定——修正单 CORRECTION_01_RT11.md 已备好并已在试点场上通过（O1–O12 关闭）。批处理 2026-10-17 05:22Z 已在 Virgo 上运行（第 1 个场 0080540101，26 个场，62–95 h）。红队 RT-22（09:4xZ，R015）：自定义耀斑过滤是空操作——试点场目标谱曝光比 v19 多 11.6%、2.1.filter 与 2.filter 的 GTI 完全相同，而 retained_fraction 以 runner 自己的 2.filter 为基准所以抓不到它。D52：批处理继续跑，CORRECTION_02 从 2.1.filter 重跑试点（验收 = 曝光在 v19 2% 内、GTI 重叠 \(\ge 98\)%）；重跑通过前 E026 产品不与 v19 比较、不进任何拟合。D64/D67 再修订（2026-10-18）：E026 产品是「v19 约定、重新掩膜」——为遮住 2.9 角分的目标，开孔面积小了 22–44%，而且这批场里没有一个有 v19 对应场，所以不能说「与 v19 可比」；与 v19 的措辞改为「与同一观测已有的 4.background 产品相比」。决定性的年份检验 RT-09-T3 只在 E026 早期与 E026 晚期之间比较，两边用同一套流程（同一 runner 版本、引擎哈希、CCF 与 HEASoft 构建；每场一个流量来源，取 E026 不取 E025；NGC 315 是内建的配对；每边至少 3 个匹配场，否则判不可行）——所以「外/内比」与基于外场的极限继续标「受观测年代混杂」。E027 的 1b Virgo 提取也排在其后。外场年代混杂（RT-09/RT-14）使重提取成为决定性检验：E026 清单里有 15 个 2010 年后的外场，配上锁定校准与冻结污染模型的早/晚配对重提取（RT-09-T3）才能分开年代与 CCD 状态E026 (28-field Virgo re-extraction, D50 conditions: reproduce the v19 2.1.filter GTI convention and report retained time vs 2.filter per camera) → the re-extracted fields are fitted under v19 rules → the E023-style joint fit is redone on the enlarged sample (with an MW latitude term); red team RT-03 reviews the outer/inner comparison. Red team RT-11 (blocking, triaged): the runner would fail all 28 fields at the audit step (3 blocking defects) and its CCF is unpinned — the correction list CORRECTION_01_RT11.md is ready and passed on the pilot field (O1–O12 closed). The batch has been running on Virgo since 2026-10-17 05:22Z (first field 0080540101, 26 fields, 62–95 h). Red team RT-22 (09:4xZ, R015): the custom flare filter is a no-op — the pilot field's obj exposure is 11.6 % above v19 and its 2.1.filter and 2.filter GTIs are identical, while retained_fraction is measured against the runner's own 2.filter so it cannot catch that. D52: the batch keeps running and CORRECTION_02 re-runs the pilot from 2.1.filter (acceptance = exposure within 2 % of v19, GTI overlap \(\ge 98\) %); until that rerun passes E026's products are not compared with v19 and enter no fit. Amended again by D64/D67 (2026-10-18): E026 products are "v19-convention, re-masked" — the aperture is 22–44% smaller because the 2.9 arcmin target is now masked, and no field in this batch has a v19 counterpart, so they cannot be called "comparable to v19"; "vs v19" becomes "vs the pre-existing 4.background product of the same observation". The decisive year test RT-09-T3 compares E026-early with E026-late only, both sides from the same pipeline (the same runner revision, engine hashes, CCF set and HEASoft build; one flux source per field, E026 not E025; NGC 315 is the built-in matched pair; at least 3 matched fields per side or the test is declared infeasible) — so the outer/inner ratio and the outer-based limits stay marked "epoch-confounded". E027's 1b Virgo extraction is queued behind it too. The outer epoch confound (RT-09/RT-14) makes the re-extraction the decisive test: the E026 list holds 15 post-2010 outer fields, and only a pinned-calibration, frozen-contamination early/late paired re-extraction (RT-09-T3) can separate year from CCD state |
E026/RT-03/RT-11/RT-22 | E021/E022/E024/E025 完成；E026 批处理运行中（产品为「v19 约定、重新掩膜」，不与 v19 可比——D67/R041）E021/E022/E024/E025 done; E026's batch running (products are "v19-convention, re-masked", not comparable to v19 — D67/R041) |
| 源掩膜的出处缺口（I013 提交门槛 5）：v19 掩膜规则不可复原（经验上是 2–4.5 keV 硬带掩膜）The source-mask provenance gap (I013 submission gate 5): the v19 mask rule is unrecoverable (empirically a hard-band 2–4.5 keV mask) | 审稿人会问「掩膜怎么来的、漏了多少」；E017a 现在给出实测答案a referee will ask "where did the mask come from and what does it miss"; E017a now answers it empirically | E017a（E017a = Paper I 星表与 v19 源掩膜的交叉核对，已完成）：v19 掩膜按 0.5–2 keV 流量掩住 Paper I 亮源（>1e-14）的 95.5 %（233/244；11 个未掩，6 个主场亮源完备度低于 0.95）；在探测器上的 Paper I 亮源掩膜完备度 99.4 %（160/161）；11 个未掩亮源里 10 个在该场 v19 观测的 MOS1 探测器之外（实测净计数 1.8 + 10，经验泄漏 \(\approx 0\)）。残余点源总预算应取 E016/E019 的 0.12–0.33 F（含 PSF 翼泄漏与星表下源）——「+0.05 残余偏差」已由红队 RT-17 撤回。注意引用总量 0.867 是毛总量（含未分辨点源，R033/D60，2026-10-18）：D49 的 0.25 逐场项没有加到 14 个主场上（95% 上限 0.087），所以这个 0.12–0.33 F 的预算是「已计入哪些、还剩多少」的说明，不是从 0.867 里再减一次。掩膜椭圆 8784 个里 87 % 没有 10″ 内的 Paper I 源——v19 掩膜来自更深的逐观测探测表（E016：有效 ML \(\approx 0.55\)），切掉更多（多为暗源/伪源），花的是天空面积不是偏差E017a (E017a = the Paper I catalogue vs the v19 source-mask cross-match; done): by 0.5–2 keV flux the v19 mask covers 95.5 % of the bright Paper I sources (>1e-14; 233/244; 11 unmasked, 6 primary fields below 0.95 bright completeness); the on-detector Paper I bright completeness is 99.4 % (160/161); 10 of the 11 unmasked bright sources lie off the MOS1 detector in their field's v19 observation (measured net counts 1.8 + 10, empirical leak \(\approx 0\)). The total residual-source budget is E016/E019's 0.12–0.33 F (including PSF-wing leakage and sub-catalogue sources) — the "+0.05 residual bias" is withdrawn by red team RT-17. Note that the quoted total 0.867 is the gross total (it includes unresolved sources, R033/D60, 2026-10-18): D49's 0.25 per-field term is not applied to the 14 primaries (95% upper limit 0.087), so this 0.12–0.33 F budget says "what is already counted and what remains", not a second subtraction from 0.867. Of the 8784 mask ellipses, 87 % have no Paper I source within 10″ — the v19 mask comes from a much deeper per-observation detection list (E016: effective ML \(\approx 0.55\)), so it cuts many more (mostly faint or spurious) regions; it costs sky area, not bias | E017b（低优先，除非审稿人要）：只用 Paper I 星表重建掩膜（需 Virgo 重提取），对比 v19 的 skyarea 与通量E017b (low priority unless a referee asks): rebuild the mask from the Paper I catalogue only (needs re-extraction on Virgo), compare skyarea and flux with v19 | E017a | E017a 完成E017a done | |
| 时间过滤（GTI）与场间底噪：v19 的 GTI 是历史 2.1.filter，场间散布的来源是什么？（I034）Time filtering (GTIs) and the between-field floor: v19's GTIs are the historical 2.1.filter — what causes the field-to-field scatter? (I034) | 若散布是残余耀斑，更严的 GTI 能压低它（E018b）；若是稳态 SP/未遮光的 QPB，GTI 无效，系统误差要靠场级模型项携带if the scatter were residual flares, a stricter GTI would suppress it (E018b); if it is a steady SP or unvignetted QPB, GTIs are powerless and the systematic must ride as a field-level model term | I034（ParetoGTICal 交叉核对）：v19 的诊断产品与 2.1.filter 逐字节一致（sha256 核对三个场）；对全部 14 主场 2.1.filter 保留的时间少于 2.filter（更严）；NW 四场损失 0–28%，而 1101/2301/2801 损失 17–25% 却不亮——残余耀斑不是 NW 超出的解释；稳态 SP 对 GTI 不可见；ParetoGTICal E020 给出同 z 处 9–12% 的场间 rms 底噪（未遮光 QPB/SP 之差），与谱拟合的场间散布同量级I034 (the ParetoGTICal cross-check): v19's diagnostic products are byte-identical to 2.1.filter (sha256-checked on three fields); on all 14 primaries 2.1.filter keeps less time than 2.filter (stricter); the NW four lose 0–28% while 1101/2301/2801 lose 17–25% without being bright — residual flares do not explain the NW excess; a steady SP floor is invisible to GTIs; ParetoGTICal E020 gives a 9–12% rms between-field floor at the same z (unvignetted QPB/SP differences), the same scale as the spectral fits' field-to-field scatter | E018b 预期变化不大（保持可选）；E027 的谱级 stage 1（31 个观测 \(\times\) 泳道）用预注册的 S1–S4 指标直接回答「GTI 分支有多大」——RT-04-T1（2.filter 分支测试）已并入；D50 的 Virgo 重提取必须复现 2.1.filter 约定并报告相对 2.filter 的保留时间；系统误差预算继续携带场级项E018b is expected to change little (stays optional); E027's spectral stage 1 (31 observations \(\times\) lanes) answers "how big is the GTI branch" directly with the pre-registered S1–S4 metrics — RT-04-T1 (the 2.filter branch test) is merged into it; D50's Virgo re-extractions must reproduce the 2.1.filter convention and report retained time vs 2.filter; the systematic budget keeps the field-level term | I034/E027 | I034 完成I034 done | |
| GTI 政策（ParetoGTICal 的 2.filter/FLR_z3/MS_z3.5_core/GATE）能为我们保留多少有价值时间、降低多少误差？（E027，PI 2026-10-16）How much valuable time do the GTI policies (ParetoGTICal's 2.filter/FLR_z3/MS_z3.5_core/GATE) keep for us, and how much error do they reduce? (E027, the PI, 2026-10-16) | v19 用历史 2.1.filter；若别的策略能在同一样本上给出更小的谱误差，或把更多观测变成可用场，样本与系统误差预算都会变；同时要给 ParetoGTI 的标准设计反馈v19 uses the historical 2.1.filter; if another policy gives smaller spectral errors on the same sample, or turns more observations into usable fields, both the sample and the systematic budget change; and ParetoGTI's criterion design needs the feedback | Stage 0/0b（灯光曲线级，已完成）：GATE 不可用（丢 5–6 个主场，含 NW 四场中的 2–3 个，与 ParetoGTICal 不采用它一致）；FLR_z3 与 MS_z3.5_core 清洗效果相同（NW 四场总水平约 −2%、其余约 −0.5%），代价不同（MS 损失 2–3% 曝光，FLR 约 10%）；对 NW 超出的预期效应小（NW−其他差变 −1.4 至 −2.3% 的总水平，最多约 CGMsum 中 ~25% NW 超出的四分之一）；场间 rms 无改善。Stage 0b（31 个观测）：总量掩盖再分配——各策略保留总量相近（none 的 0.61–0.65）但落在不同观测上，主要价值是覆盖（红队 RT-19 复核后：按 live K 计 3 个有用非 v19 指向，按 GTI 总和计 4 个，2501 边缘；另有若干浅但仍耀发的场；31 个已打分观测里有 6 个是策略的调参集，须单独报告）。红队 RT-10 分诊 + 预注册修订 v2（任何谱数据之前）：S2 场间 \(\tau\) 只作描述（功效太低且误差模型敏感），策略间不排座次；泳道不嵌套 → S4 改为集合差 + 公共核分解；S1 的 CGMsum 与 SP 简并 → 主通道改为软质子归一化与不依赖模型的软波段面亮度。Stage 1a 完成（203 个 GTI 文件，kept live 与 ParetoGTICal 的 K 逐秒相等）；谱级结论待定Stage 0/0b (light-curve level, done): GATE is unusable (it drops 5–6 primaries, including 2–3 of the NW four, agreeing with ParetoGTICal not adopting it); FLR_z3 and MS_z3.5_core clean identically (about −2% on the NW four's gross level, about −0.5% on the others) at different cost (MS costs 2–3% of the exposure, FLR_z3 about 10%); the expected effect on the NW excess is small (the NW−other difference moves by −1.4 to −2.3% of the gross level, at most about a quarter of CGMsum's ~25% NW excess); no improvement in the between-field rms. Stage 0b (31 observations): totals hide the redistribution — the policies keep similar totals (0.61–0.65 of none) but on different observations, so the main value is coverage (after red team RT-19's review: 3 useful non-v19 pointings by live K, 4 by the GTI sum, with 2501 borderline, plus several shallow but still flaring fields; 6 of the 31 scored observations are the policies' tuning set and must be reported separately). Red team RT-10 triaged + pre-registration amendment v2 (before any spectral data): S2's field-to-field \(\tau\) becomes descriptive only (too little power and error-model dependent), so it does not rank policies; the lanes are not nested → S4 becomes the set difference plus a common-core decomposition; S1 on CGMsum is degenerate with the SP term → the primary channel becomes the soft-proton norm and a model-independent soft-band surface brightness. Stage 1a done (203 GTI files, kept live equal to ParetoGTICal's K to the second); the spectral verdict is Pending |
E027 stage 1：1a GTI 导出（lang，06:26Z → ~07:10Z）已完成并验收（203 个 GTI 文件）→ 1b Virgo 逐观测逐泳道提取（等 E026 修正后的 runner）→ 1c lang 拟合 → 1d 反馈备忘录 FEEDBACK_ParetoGTI.md。RT-04-T1（2.filter GTI 分支测试）已并入 stage 1E027 stage 1: 1a GTI export (lang, 06:26Z → ~07:10Z) done and accepted (203 GTI files) → 1b per-observation per-lane extraction on Virgo (waits for E026's corrected runner) → 1c fits on lang → 1d the feedback memo FEEDBACK_ParetoGTI.md. RT-04-T1 (the 2.filter GTI-branch test) is merged into stage 1 |
E027 | stage 0/0b + 1a 完成；1b/1c 待定stages 0/0b + 1a done; 1b/1c Pending | |
| 南北差异 0.212 \(\pm 0.210\)（1.01\(\sigma\)）——一致性，不是非对称探测North-minus-South 0.212 \(\pm 0.210 (1.01\)\(\sigma\)) — consistency, not an asymmetry detection | 非对称上限是论文的直接结果之一；两侧半径分布不匹配（中位 19.28 vs 14.97 kpc）The asymmetry limit is one of the paper's direct results; the two sides have unmatched radius distributions (medians 19.28 vs 14.97 kpc) | 总和在两侧一致；任何分量级非对称结论都需要 MW 均匀或 M31 对称假设The sum is consistent between sides; any component-level asymmetry claim needs MW-uniformity or M31-symmetry assumptions | 共享径向项 + 侧向偏移的联合拟合（I013 提交门槛 1）A joint fit with a shared radial term plus a side offset (I013 submission gate 1) | 计划planned | 计划planned |
来源：cognition-os/02_CURRENT_QUESTION.md（2026-10-14 D48 块）；cognition-os/insights/I033（E022 + E023 段；第 5 条经 RT-01 回到 Questioned）、I006、I012、I017、I034；cognition-os/experiments/E013（A2 v1、toy-MC 与「A2 v2 final」节）、E014、E016、E018a、E019、E020（「Red-team RT-01 triage」节）、E021、E022（correction 1 节）、E023、E024、E025（correction 1 节）、E017a；cognition-os/04_DECISION_LOG.md D49 及其 update 节（撤回）、D50 与 D60–D62；cognition-os/07_REVISION_LOG.md R029–R036；handbook/live_updates.jsonl L024–L029；cognition-os/REDTEAM_QUEUE.md（RT-02 的 F3 发现）；.agent/LOOP_STATE.md（A1d/RT-02/RT-03/E026 运行与队列）；03_TRUST_TABLE.md 末行。吸收行：R036/D62（总柱密度 19–31%、NH 项为固定归一化模型响应、RT-39-T1 待定）；模型敏感性行：R034/R035/D61（CXB 谱指数 −6.2%、Z 杠杆、E029 待定、「仅气体 0.82」撤回）；点源预算行：R033/D60（毛总量、D49 的 0.25 不用于 14 主场）。
Source: cognition-os/02_CURRENT_QUESTION.md (2026-10-14 D48 block); cognition-os/insights/I033 (the E022 + E023 paragraph; point 5 back to Questioned after RT-01), I006, I012, I017, I034; cognition-os/experiments/E013 (the A2 v1, toy-MC and "A2 v2 final" sections), E014, E016, E018a, E019, E020 (the "Red-team RT-01 triage" section), E021, E022 (the correction-1 section), E023, E024, E025 (the correction-1 section), E017a; cognition-os/04_DECISION_LOG.md D49 with its update section (withdrawn), D50 and D60–D62; cognition-os/07_REVISION_LOG.md R029–R036; handbook/live_updates.jsonl L024–L029; cognition-os/REDTEAM_QUEUE.md (RT-02's F3 finding); .agent/LOOP_STATE.md (the A1d/RT-02/RT-03/E026 runs and the queue); the last rows of 03_TRUST_TABLE.md. Absorption row: R036/D62 (the total column is 19–31% higher, the NH term is a fixed-normalisation model response, RT-39-T1 Pending); model-sensitivity row: R034/R035/D61 (the CXB photon index −6.2%, the Z lever, E029 Pending, "gas-only 0.82" withdrawn); point-source-budget row: R033/D60 (the gross total; D49's 0.25 is not applied to the 14 primaries).
未来 1–4 周计划（依据 D48 与 02_CURRENT_QUESTION）Plan for the next 1–4 weeks (from D48 and 02_CURRENT_QUESTION)
数据：newangels3_discussion/E013_A2_claude/a2_ul_vs_rc_interp_A1d.csv（A2 基线，Claude；RT-13 复核 39 行全部一致）。本页图形于 2026-10-18 由 handbook/hermes_figscripts/make_fig_a2_A1d.py 从该 CSV 重新绘制（只画图、不做任何拟合），总量线取自毛总量 0.867（D57/D60）并标注「gross total, D57」；旧生成器 a2_fig_ul_vs_rc.py 图中 0.965 的灰线已不再绘制（B033/R032）。红队 RT-08/RT-13 见 E013 对应节。
Data: newangels3_discussion/E013_A2_claude/a2_ul_vs_rc_interp_A1d.csv (the A2 baseline, Claude; RT-13 reproduced all 39 rows). The figure on this page was redrawn on 2026-10-18 by handbook/hermes_figscripts/make_fig_a2_A1d.py from that CSV (plotting only, no fitting), with the total line taken from the gross total 0.867 (D57/D60) and labelled "gross total, D57"; the old generator a2_fig_ul_vs_rc.py drew a grey line at 0.965, which is no longer drawn (B033/R032). Red teams RT-08/RT-13: the matching sections of E013.
数据：newangels3_discussion/E018a_luna/fits_per_field.csv（E018a，luna，Claude 已核）；比值 1.295/1.168/1.130 由本站用 v19 权重从该 CSV 复算。图由 handbook/hermes_figscripts/make_fig_e018a.py 生成。
Data: newangels3_discussion/E018a_luna/fits_per_field.csv (E018a, luna, Claude-checked); the 1.295/1.168/1.130 ratios recomputed by this site from that CSV with v19 weights. Figure made by handbook/hermes_figscripts/make_fig_e018a.py.
data/ 的 145 个 ObsID，不是 XMM 档案——22.5° 锥的 XSA 查询列 563 个指向，418 个本地没有（红队 RT-18，R013）。lang 的 data/ 有 145 个 ObsID：class-A（v19 可用）22 个、class-B 114 个、class-C 9 个。class-A 按到 M31 的距离：0–30 kpc 14 个（New-ANGELS 核心样本）、30–60 kpc 0 个、60–100 kpc 3 个、100–150 kpc 1 个、150–200 kpc 3 个、200–300 kpc 1 个——30 kpc 以外只有 8 个，且全部在负银纬。
Figure: the sky coverage of E021's archival inventory (viewed by the reviewer). Note: this is the 145 ObsIDs on lang's local data/, not the XMM archive — a 22.5° XSA cone lists 563 pointings, 418 of them absent locally (red team RT-18, R013). lang's data/ holds 145 ObsIDs: 22 class A (v19-ready), 114 class B, 9 class C. Class A by distance from M31: 14 at 0–30 kpc (the New-ANGELS core), 0 at 30–60 kpc, 3 at 60–100, 1 at 100–150, 3 at 150–200, 1 at 200–300 — only 8 beyond 30 kpc, all at negative Galactic latitude.
数据：newangels3_discussion/E021_hermes/inventory.csv 与 class_counts.csv（E021，Hermes，已接受）。图来自 newangels3_discussion/E021_hermes/fig_e021_sky_coverage.png（评审已查看），复制入 assets/。
Data: newangels3_discussion/E021_hermes/inventory.csv and class_counts.csv (E021, Hermes, accepted). Figure from newangels3_discussion/E021_hermes/fig_e021_sky_coverage.png (viewed by the reviewer), copied into assets/.
-
第一版联合似然 + 玩具蒙特卡洛（已完成）：谱联合似然在 14 主场的 v19 曲面上插值，对 N_MW、梯度、kT（0.185 keV）做 profile；无 M31 分量被偏好（\(\Delta \mathrm{stat}\) \(\le 1.07\)）。单侧 95% 上限（\(\Delta \mathrm{stat} = 2.71\)，插值版）见状态页表；玩具蒙特卡洛覆盖率 0.975–0.984（略保守，但只在反射似然位置族玩具下成立——红队 RT-08：高斯抽样只给 0.87–0.88），误报率 0.069–0.086，观测上限不等于玩具中位数（v1 比它低 0.8–10.8%，v2C 比它高 24–58%）。产出
newangels3_discussion/E013_A2_claude/（a2_ul_vs_rc_interp.csv、a2_toymc_threshold.csv、a2_ul_vs_rc.png）。 The first joint-likelihood run + toy Monte Carlo (done): the spectral joint likelihood interpolated on the 14 primary fields' v19 surfaces, profiling over N_MW, the gradient and kT (0.185 keV); no M31 component is preferred (\(\Delta \mathrm{stat}\) \(\le 1.07\)). One-sided 95% UL (\(\Delta \mathrm{stat} = 2.71\), interpolated) in the state-page table; toy-MC coverage 0.975–0.984 (slightly conservative, but only under the reflected-likelihood location-family toy — red team RT-08: a Gaussian sampler gives 0.87–0.88), false-detection rate 0.069–0.086, and the observed limits are not the toy median (v1 sits 0.8–10.8 % below it, v2C 24–58 % above it). Outputsnewangels3_discussion/E013_A2_claude/(a2_ul_vs_rc_interp.csv, a2_toymc_threshold.csv, a2_ul_vs_rc.png). -
样本级重收敛（已完成；重收敛分诊已做）：样本级重收敛（20 场，模型 A/B/C，多元起始）。模型 C 下 14 场均值 0.964 → 0.819 \(\pm 0.041\)。重收敛分诊（Claude 复核）：−15.04% 中 v19 重收敛（A）−7.65%、SP 谱形（A→B）−5.29%、线能量（B→C）−2.87%——「SP 谱形单侧 −15% 系统误差」不成立；模型 C 的 SP 谱形未经物理验证；D49 更新撤回。模型 C 曲面（A1c）降级为敏感性变体；多元起始剖面曲面（v19 模型）成为联合似然基线。产出
newangels3_discussion/E020_luna/。 The sample-wide re-convergence (done; its triage is done): the sample-wide re-convergence (20 fields, Models A/B/C, multi-start). Under Model C the 14-field mean moves 0.964 → 0.819 \(\pm 0.041\). The re-convergence triage (Claude-reproduced): of −15.04%, the v19 re-convergence (A) gives −7.65%, the SP shape (A→B) −5.29%, the line energies (B→C) −2.87% — the "one-sided −15% SP-shape systematic" does not hold; Model C's SP shapes are not physically validated; the D49 update is withdrawn. The Model-C surfaces (A1c) are demoted to a sensitivity variant; the multi-start profile surfaces (v19 model) become the joint-likelihood baseline. Outputsnewangels3_discussion/E020_luna/. -
实测 PSF 翼（已完成）：PSF 翼泄漏的实测——81 个孤立类点源叠加，最亮 bin f_leak 10.3 \(\pm 8.6%\)（原值）／20.4 \(\pm 7.8%\)（零点校正）；14 主场泄漏 SB 0.16（原值）／0.33（零点校正）vs 未解点源预算的模型 0.12。产出
newangels3_discussion/E019_codexnim/。 The empirical PSF wing (done): the empirical PSF-wing leak — a stack of 81 isolated point-like sources, brightest-bin f_leak 10.3 \(\pm 8.6%\) raw / 20.4 \(\pm 7.8%\) null-corrected; per-field leak SB over the 14 primaries 0.16 raw / 0.33 null-corrected vs the unresolved-source budget's model 0.12. Outputsnewangels3_discussion/E019_codexnim/. -
档案清点（已完成）：档案清点与 300 kpc 图设计（无拟合）——lang 磁盘上 145 个 ObsID，class-A 22 个，30 kpc 以外仅 8 个（红队 RT-18/R013：只覆盖本地磁盘，XMM 档案 22.5° 锥有 563 个指向、418 个本地没有）；产出
newangels3_discussion/E021_hermes/（inventory.csv、DESIGN.md、天区覆盖图）。 The archival inventory (done): the archival inventory and the 300-kpc map design (no fitting) — 145 ObsIDs on lang's disk, 22 class A, only 8 beyond 30 kpc (red team RT-18/R013: disk-scoped only; the XMM archive's 22.5° cone lists 563 pointings, 418 absent locally); outputsnewangels3_discussion/E021_hermes/(inventory.csv, DESIGN.md, the sky-coverage figure). -
八个外场拟合（已完成）：8 个外 class-A 档案场（82–253 kpc）的 v19 + 模型 C 拟合（N_H 从 HI4PI 重建，与 v19 差 0.05–0.31 %）。v19 结果被接受为外场参考：外 IVW 0.842 \(\pm 0.065\)；内禀 norm 外/内 0.80 \(\pm 0.13\)；模型 C 仅作系统变体（其外场剖面误差 \(\chi^{2} 168/7\)，不能用于加权平均）。产出
newangels3_discussion/E022_luna/。 The eight outer-field fits (done): v19 + Model-C fits of the 8 outer class-A archival fields (82–253 kpc; N_H rebuilt from HI4PI, agreeing with v19 to 0.05–0.31 %). The v19 results are accepted as the outer reference: outer IVW 0.842 \(\pm 0.065\); intrinsic norm outer/inner 0.80 \(\pm 0.13\); Model C is a systematic variant only (its outer profile errors give \(\chi^{2} 168/7\), unusable for a weighted mean). Outputsnewangels3_discussion/E022_luna/. -
大核联合拟合（已完成）：22 个 v19 norm（14 内 + 8 外）的联合 MW 常数 + M31 \(\beta\) 拟合（高斯似然，局部协方差误差）——大核晕在 20 kpc 的单侧 95% 上限 0.46（r_c=50）／0.59（r_c=100）／0.90（r_c=200）（\(\beta = 0.5\)）；Zhang 0.83 被排除到 r_c \(\approx 125\)–175 kpc (\(\beta = 0.4/0.5\))；无 M31 分量被偏好（\(\Delta \chi^{2}(\mathrm{A}=0)\) \(\le 1.3\)）。Claude 已独立复算 r_c = 50/100 的上限，数值一致。产出
newangels3_discussion/E023_luna/limits.csv。下一步：在谱曲面上重做。 The large-core joint fit (done): a joint MW-constant + M31-beta fit of the 22 v19 norms (14 inner + 8 outer; Gaussian likelihood on local-covariance errors) — the large-core halo's one-sided 95% UL at 20 kpc is 0.46 (r_c=50) / 0.59 (r_c=100) / 0.90 (r_c=200) for \(\beta = 0.5\); Zhang 0.83 is excluded up to r_c \(\approx 125\)–175 kpc (\(\beta = 0.4/0.5\)); no M31 term is preferred (\(\Delta \chi^{2}(\mathrm{A}=0)\) \(\le 1.3\)). Claude independently recomputed the r_c = 50/100 limits and got the same values. Outputsnewangels3_discussion/E023_luna/limits.csv. Next: redo it on spectral surfaces. -
掩膜出处交叉核对（已完成）：Paper I 星表与 v19 掩膜的交叉匹配（14 主场，MOS1）——亮源（>1e-14）掩住 95.5 %（233/244，11 个未掩）；8784 个掩膜椭圆里 87 % 无 10″ 内的 Paper I 源（掩膜来自更深的逐观测探测表）。掩膜出处缺口就此闭合。产出
newangels3_discussion/E017a_luna/。 The mask-provenance cross-check (done): the Paper I catalogue cross-matched against the v19 masks (14 primary fields, MOS1) — 95.5 % of bright sources (>1e-14) masked (233/244, 11 unmasked); of the 8784 mask ellipses 87 % have no Paper I source within 10″ (the masks come from a deeper per-observation detection list). The mask-provenance gap is thereby closed. Outputsnewangels3_discussion/E017a_luna/. -
数值独立复核（已完成）：样本级重收敛与八个外场拟合数值的独立复核（luna，直接从四个 CSV 复算，不用任何先前报告）——20 个标量比较 18 个 PASS；2 个「FAIL」只是样本级重收敛加权径向斜率的舍入差（复算 −0.0005348 vs 声称 −0.0005；0.006051 vs 0.0061；四位小数下一致，超出 0.5 % 严格容差）。产出
newangels3_discussion/V01_luna/verify_table.csv。 The independent verification (done): an independent verification of the re-convergence and outer-field-fit numbers (luna, recomputed directly from the four CSVs, no prior reports as inputs) — 18 of 20 scalar comparisons PASS; the 2 "FAILs" are rounding-only differences in the re-convergence's weighted radial slope (recomputed −0.0005348 vs claimed −0.0005; 0.006051 vs 0.0061; identical at four decimals, outside the strict 0.5 % tolerance). Outputsnewangels3_discussion/V01_luna/verify_table.csv. -
模型 C 曲面完成 + 联合似然第二版终稿（敏感性变体，2026-10-15 18:26Z）：模型 C 曲面（codexnim）14/14 场、加密后 117 点/场，门检全过；联合似然第二版终稿给出插值上限（\(\beta = 0.5\)，r_c = 4/8/15/30 = 0.111/0.138/0.215/0.484；\(\beta = 0.4/0.6\) 在 r_c = 4 为 0.176/0.075），玩具蒙特卡洛 1000 次覆盖率 0.918–0.933（低于 0.95）——\(\Delta \mathrm{stat} = 2.71\) 在模型 C 曲面上略不保守，校准阈值待做（Claude TODO）；Zhang/上限（r_c \(\le 8\)）= 3.9–11.0 倍。产出
newangels3_discussion/E013_A2_claude/（a2_ul_vs_rc_interp_v2C.csv、a2_ul_vs_rc_v2C.png）。重收敛分诊后定位：敏感性变体，不是基线。 The Model-C surfaces done + the second joint-likelihood final (a sensitivity variant, 2026-10-15 18:26Z): the Model-C surfaces (codexnim) 14/14 fields, 117 points per field after densification, all gates exact; the second joint-likelihood final gives interpolated ULs (\(\beta = 0.5, \mathrm{r_c} = 4/8/15/30 = 0.111/0.138/0.215/0.484\); \(\beta = 0.4/0.6\) at r_c = 4: 0.176/0.075), with a 1000-toy MC coverage of 0.918–0.933 (below 0.95) — the \(\Delta \mathrm{stat} = 2.71\) rule is slightly anti-conservative on the Model-C surfaces, a calibrated threshold is TODO (Claude); Zhang/UL for r_c \(\le 8 = 3.9\)–11.0x. Outputsnewangels3_discussion/E013_A2_claude/(a2_ul_vs_rc_interp_v2C.csv, a2_ul_vs_rc_v2C.png). After the re-convergence triage its status is: a sensitivity variant, not the baseline. -
外 B 场政策完成（2026-10-15 ~13:2xZ）+ 政策决定 + 干净外样本完成：政策选项表（Hermes）给出 47 个外 B 场的四种政策：P1（遮展源）+12 个方向、P2（放宽到 8–10 ks）累计 +31、P3（加 9 个 Virgo 重提取）+40、P4（加单相机）+43；全部在 b < 0；NGC 147 不可用（仅剩 22% 视场）。PI 决定（2026-10-15）：P1+P2 立即拟合、6 个真正外围场 Virgo 重提取、跳过 P4、弃 NGC 147。干净外样本（luna）照此执行：P1 12/12 掩膜审计未过 → 全部转重提取；P2 18/19 接受；干净外样本（13 场）0.846 \(\pm 0.059\)，外/内 = 0.877 \(\pm 0.072\)；correction 1 把 10 个 P2 a+b 场也判为掩膜不足 → 重提取清单 6 + 12 + 10 = 28 场。产出
newangels3_discussion/E024_hermes/、newangels3_discussion/E025_luna/。 The outer class-B policy done (2026-10-15 ~13:2xZ) + the policy decision + the clean outer sample done: the policy-options table (Hermes) laid out four policies for the 47 outer class-B fields: P1 (mask extended targets) +12 directions, P2 (relax to 8–10 ks) +31 cumulative, P3 (add 9 Virgo re-extractions) +40, P4 (add one-camera fields) +43; all at b < 0; NGC 147 unusable (22% of the field of view left). The PI decided (2026-10-15): fit P1+P2 now, re-extract 6 genuinely-outer fields on Virgo, skip P4, drop NGC 147. The clean outer sample (luna) executed it: all 12 P1 fields fail the mask audit → all to re-extraction; 18/19 P2 fits accepted; the clean outer sample (13 fields) 0.846 \(\pm 0.059\), outer/inner = 0.877 \(\pm 0.072\); correction 1 found the 10 P2 a+b fields also under-masked → the re-extraction list is 6 + 12 + 10 = 28 fields. Outputsnewangels3_discussion/E024_hermes/,newangels3_discussion/E025_luna/. -
多元起始剖面曲面完成 + 联合似然基线在它上面重算（2026-10-16 13:2xZ；临时，阈值未校准）：多元起始剖面曲面（codexnim，18:50Z → ~11:20Z，65.7 worker-小时）14/14 场 \(\times 117\) 点 = 1638/1638，0 个拟合失败，v19 与 Model-A 恢复门检 14/14。联合似然基线（同一估计器：MW 常数、kT 在最佳共温 \(\pm1\) 行内 profile、\(\Delta \mathrm{stat} = 2.71\)、插值）给出 20 kpc 处单侧 95% 上限：核半径 \(\le 8 \mathrm{kpc}\) 时 0.12–0.16（\(\beta 0.4\)）／0.075–0.10（\(\beta 0.5\)）／0.049–0.071（\(\beta 0.6\)）；r_c 30 为 0.30–0.56，r_c 50 为 0.68–1.26。插值收敛（插值警告的 O9）：线性/样条上限比 0.77–0.82（A1 上 0.43–0.50），仍差于 10% 目标，取较松的样条值。Zhang 名义中值 0.8285 / 上限（r_c \(\le 8\)）= 5.2–6.7（\(\beta 0.4\)）、8.1–11.1（\(\beta 0.5\)）、11.7–17.1（\(\beta 0.6\)）。晕算术分诊已做：算术与输入成立、2.71 是「参数不能为负」边界情形下正确的分位数；上限主要靠一个视场（0800730601）——v19 曲面上去掉它放宽 +46…+62%（模型 C 曲面复核：模型 C 曲面下只放宽 6–30%，该场降到 14 场里第 4 位），对外措辞改为「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」；报出的自然样条值是四种插值里最松的（其余紧 2–23%）。玩具中位数撤回已做：玩具蒙特卡洛的位置族校准只在简化模型下有效，「上限已校准良好」不成立——阈值改用能谱级模拟校准（能谱模拟阈值校准）。产出
newangels3_discussion/E013_A2_claude/（a2_ul_vs_rc_interp_A1d.csv、a2_ul_vs_rc_interp_A1d_linear.csv、a2_ul_vs_rc_A1d.png）。 The multi-start profile surfaces done + the joint-likelihood baseline recomputed on them (2026-10-16 13:2xZ; provisional, threshold NOT calibrated): the multi-start profile surfaces (codexnim, 18:50Z → ~11:20Z, 65.7 worker-hours) give 14/14 fields \(\times 117\) points = 1638/1638, 0 failed fits, with the v19 and Model-A restore gates 14/14. The joint-likelihood baseline (the same estimator: MW constant, kT profiled over the best shared kT \(\pm1\) row, \(\Delta \mathrm{stat} = 2.71\), interpolated) gives one-sided 95% ULs at 20 kpc: for core radius \(\le 8 \mathrm{kpc}\), 0.12–0.16 (\(\beta 0.4) / 0.075\)–0.10 (\(\beta 0.5) / 0.049\)–0.071 (\(\beta 0.6\)); r_c 30 gives 0.30–0.56 and r_c 50 gives 0.68–1.26. Interpolation convergence (the interpolation warning's O9): the linear/spline UL ratio is 0.77–0.82 (0.43–0.50 on A1), still outside the 10 % target, so the looser spline values are quoted. Zhang's nominal median 0.8285 / UL for r_c \(\le 8 = 5.2\)–6.7 (\(\beta 0.4\)), 8.1–11.1 (\(\beta 0.5\)), 11.7–17.1 (\(\beta 0.6\)). The halo-arithmetic triage is done: the arithmetic and inputs hold and 2.71 is the correct quantile for a parameter bounded at zero; the limit leans on one field (0800730601) — leaving it out loosens by +46…+62 % on the v19 surfaces (re-checked under the Model C surfaces: only 6–30 % there, where that field falls to 4th of 14), so the public wording becomes "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)"; the quoted natural-spline UL is the loosest of four interpolants (the others 2–23 % tighter). The toy-median withdrawal is done: the toy-MC location-family calibration holds only under its simplified model, so "the limits are well calibrated" does not survive — the threshold is to be calibrated by spectrum-level simulation (the spectral-simulation threshold calibration). Outputsnewangels3_discussion/E013_A2_claude/(a2_ul_vs_rc_interp_A1d.csv, a2_ul_vs_rc_interp_A1d_linear.csv, a2_ul_vs_rc_A1d.png). - GTI 政策筛选 stage 0/0b + 1a 完成（2026-10-16 06:0xZ–07:10Z）；1b/1c 待定：PI 2026-10-16 两次要求——先问 GATE/FLR_z3/MS_z3.5_core 对我们结果的改进，随后把范围改为全部 31 个 LP 080073 观测（保留多少有价值时间、降低多少误差、如何反馈 ParetoGTI 标准设计）。Stage 0/0b（灯光曲线级）：GATE 不可用（丢 5–6 个主场）；FLR_z3 与 MS_z3.5_core 清洗相同、代价不同（2–3% vs ~10% 曝光）；主要价值是覆盖（覆盖复核后：按 live K 计 3 个有用非 v19 指向，按 GTI 总和计 4 个，2501 边缘；6/31 已打分观测是调参集）。指标分诊 + 预注册修订 v2（任何谱数据之前）：S2 场间 \(\tau\) 只作描述；泳道不嵌套 → S4 改为集合差 + 公共核分解；S1 的 CGMsum 与 SP 简并 → 主通道改为软质子归一化与不依赖模型的软波段面亮度。Stage 1a 完成并验收（codexnim，06:26Z → ~07:10Z）：203 个 GTI FITS 带 sha256 manifest，kept live 与 ParetoGTICal per_obs K 逐秒相等。1b Virgo 提取等重提取批处理修正后的 runner；1c 拟合；1d 反馈备忘录。2.filter 分支测试已并入。谱级结果待定。 The GTI-policy screen's stages 0/0b + 1a done (2026-10-16 06:0xZ–07:10Z); 1b/1c Pending: the PI's two 2026-10-16 requests — first about GATE/FLR_z3/MS_z3.5_core for the improvement to our results, then rescoped to all 31 LP 080073 observations (how much valuable time is kept, how much error is reduced, and how to feed back into ParetoGTI criterion design). Stages 0/0b (light-curve level): GATE is unusable (it drops 5–6 primaries); FLR_z3 and MS_z3.5_core clean identically at different cost (2–3% vs ~10% of the exposure); the main value is coverage (after the coverage review: 3 useful non-v19 pointings by live K, 4 by the GTI sum, 2501 borderline, plus several shallow but still flaring fields; 6 of the 31 scored observations are the tuning set). The metrics triage + pre-registration amendment v2 (before any spectral data): S2's field-to-field \(\tau\) becomes descriptive only; the lanes are not nested → S4 becomes the set difference plus a common-core decomposition; S1 on CGMsum is degenerate with the SP term → the primary channel becomes the soft-proton norm and a model-independent soft-band surface brightness. Stage 1a done and accepted (codexnim, 06:26Z → ~07:10Z): 203 GTI FITS with a sha256 manifest, kept live equal to ParetoGTICal's per_obs K to the second. Stage 1b's Virgo extraction waits for the re-extraction batch's corrected runner; then 1c fits and 1d the feedback memo. The 2.filter branch test is merged in. Spectral results Pending.
-
已分诊完毕：响应文件审计（只用响应文件的 CCF 审计——ARF 相关 CCF 逐字节相同，响应不能解释年代效应）、第 0 阶段验收分诊（审阈值校准的 stage 0——判为运行本身无效：fake_pha 丢分组、只有一个 kT 行、闭合 5/9）、模型 C 曲面复核（0601 场的集中度是 v19 拓扑依赖的，模型 C 下基本消失）、运行器配置复核（审重提取批处理 runner——自定义耀斑过滤是空操作，试点曝光比 v19 多 11.6%，产品暂不可比，R015）；运行中，待定：能谱模拟阈值校准（用能谱模拟校准 \(\Delta \mathrm{stat} = 2.71\) 阈值——试点 40/40 个配置已跑完，覆盖曲线形状一度被读作符合预期（已撤回，R016），但 stage 1 仍被阻断：闭合检验未达标，0800731101 三个节点比多元起始曲面高 +0.34…+0.61、0800730801 一个节点 +0.059，原因是闭合只用 3 行热启动而多元起始曲面用 11 行 kT 链；要求用 11 行链闭合或把 stage 1 限制在能闭合的场上；试点只有 2 个场、每类 1 个玩具，是冒烟测试不是校准）；Virgo 重提取批处理（28 场 Virgo 重提取——批处理 2026-10-17 05:22Z 起在 Virgo 上运行，第 1 个场 0080540101，26 个场，62–95 h；红队 RT-11 的修正单已在试点场上通过，但运行器配置复核发现自定义过滤器没生效，批处理继续决定让批处理继续跑并同时发
CORRECTION_02从 2.1.filter 重跑试点（验收 = 目标谱曝光在 v19 2% 内、GTI 重叠 \(\ge 98\)%），重跑通过前重提取批处理产品不与 v19 比较、不进任何拟合）；GTI 政策筛选 stage 1b/1c（谱级提取与拟合，等重提取批处理修正后的 runner）；早/晚配对重提取（排队）；闭合原因更正（审阈值校准闭合/试点，报告仍在写）。红队 RT-01…RT-16、RT-22 已分诊完毕（见组会页）。结果出来前不引用任何数字。 Triaged: the response-file audit (a response-only CCF audit — the ARF-relevant CCF is byte-identical, so the response cannot explain the epoch effect), the stage-0 acceptance triage (reviewing the threshold calibration's stage 0 — ruled invalid as run: fake_pha ungroups, one kT row, closure 5/9), the Model-C surfaces re-check (the 0601 concentration is v19-topology dependent and largely disappears under Model C) and the runner-config review (reviewing the re-extraction batch's runner — the custom flare filter is a no-op, the pilot exposure is 11.6 % above v19, products not comparable yet, R015); Running, Pending: the spectral-simulation threshold calibration (calibrating the \(\Delta \mathrm{stat} = 2.71\) threshold with spectral simulation — the pilot's 40/40 profiles are done and the coverage curve was first read as the expected shape (withdrawn, R016), but stage 1 is still blocked: the closure test is not met, with 0800731101's three nodes +0.34…+0.61 above the multi-start surface and 0800730801's one node +0.059, because the closure warm-starts from only 3 rows while the multi-start surface uses an 11-row kT chain; acceptance requires closure with the 11-row chain or restricting stage 1 to the fields that close; the pilot has only 2 fields and 1 toy per kind, so it is a smoke test, not a calibration); the Virgo re-extraction batch (28-field Virgo re-extraction — the batch has been running on Virgo since 2026-10-17 05:22Z, first field 0080540101, 26 fields, 62–95 h; red team RT-11's correction list passed on the pilot field, but the runner-config review found the custom filter never took effect, so the batch-continue decision keeps the batch running while sendingCORRECTION_02to re-run the pilot from 2.1.filter (acceptance = obj exposure within 2 % of v19, GTI overlap \(\ge 98\) %), and until that rerun passes the re-extraction batch's products are not compared with v19 and enter no fit); the GTI-policy screen's stage 1b/1c (spectral extraction and fits, waiting for the re-extraction batch's corrected runner); the early/late paired re-extraction (queued); the closure-cause correction (reviewing the threshold calibration's closure/pilot, report still being written). Red teams RT-01…RT-16 and RT-22 are triaged (see the meeting page). No numbers quoted until results land. - 其后：等能谱模拟阈值校准用能谱模拟校准 \(\Delta \mathrm{stat} = 2.71\) 阈值，再决定插值终值（PCHIP 或显式节点合并）并引用校准后的收敛上限；在谱曲面上以流量空间重做大核联合拟合（加 MW 纬度项与外场系统项）；重提取批处理的 CORRECTION_02 重跑通过后（目标谱曝光在 v19 2% 内、GTI 重叠 \(\ge 98\)%），28 个重提取场才可按 v19 规则拟合并进入扩大的内外样本联合拟合；GTI 政策筛选 stage 1 的谱结果与 ParetoGTI 反馈备忘录；响应文件审计与早/晚配对重提取判定外场年代混杂的原因；RASS 25° 原生计数图（排队）；更严耀斑过滤保持可选（GTI 交叉核对：预期变化不大）。 After that: wait for the spectral-simulation threshold calibration to calibrate the \(\Delta \mathrm{stat} = 2.71\) threshold with spectral simulation, then settle the interpolant (PCHIP or an explicit node-merge rule) and quote the calibrated, converged UL; redo the large-core joint fit in flux space on spectral surfaces (with an MW latitude term and an outer systematic); once the re-extraction batch's correction passes (obj exposure within 2 % of v19, GTI overlap \(\ge 98\) %), batch the 28 fields, fit them under v19 rules and run the joint fit on the enlarged inner+outer sample; the GTI-policy screen's stage-1 spectral results and the ParetoGTI feedback memo; the response-file audit and the early/late paired re-extraction decide the cause of the outer epoch confound; the 25° native-count RASS map (queued); the stricter flare filter stays optional (the GTI cross-check: expected to change little).
- RASS 图（排队）：M31 周围 25° 的 RASS 原生计数图（MW 纬度模型 + M31 中心项 + LHB 模板），为本地 MW 模型与 Local Hot Bridge 检验（O4）铺路。 The RASS map (queued): a 25° native-count RASS map around M31 (MW latitude model + an M31-centred term + an LHB template), paving the way for the local MW model and the Local Hot Bridge test (O4).
- 掩膜重建（低优先，排队）：只用 Paper I 星表重建源掩膜（需 Virgo 重提取），对比 v19 的 skyarea 与通量——掩膜出处交叉核对已把掩膜出处缺口闭合，此项仅在审稿人要求时做。 The mask rebuild (low priority, queued): rebuild the source mask from the Paper I catalogue only (needs re-extraction on Virgo) and compare skyarea and flux with v19 — the mask-provenance cross-check has closed the mask-provenance gap, so this runs only if a referee asks.
- 论文讨论：把 I033 的科学线（总和、轮廓形状——含大核限制、NW 起源）与未解点源预算、实测 PSF 翼、样本级重收敛（重收敛分诊后作敏感性变体）、八个外场拟合、大核联合拟合、干净外样本、掩膜出处交叉核对的新结果写进 New-ANGELS III 草稿的讨论部分；提交门槛仍按 I013。 Paper discussion: write the I033 science lines (total; profile shape — now including the large-core limit; NW origin) plus the new results of the unresolved-source budget, the empirical PSF wing, the sample-wide re-convergence (a sensitivity variant after its triage), the eight outer-field fits, the large-core joint fit, the clean outer sample and the mask-provenance cross-check into the New-ANGELS III draft discussion; submission gates remain those of I013.
来源：cognition-os/00_AI_BRIEFING.md（2026-10-14 D48 与 2026-10-15 13Z update）；cognition-os/experiments/E013（A2 v1、toy-MC 与「A2 v2 final」节）、E019、E020（「Red-team RT-01 triage」节）、E021、E014、E016、E018a、E022（correction 1 节）、E023、E024、E025（correction 1 节）、E017a；cognition-os/insights/I013、I034；.agent/jobs/2026-10-15-v01-verify-e020-e022/REPORT.luna.md；cognition-os/04_DECISION_LOG.md D49（update 节已撤回）与 D50；cognition-os/REDTEAM_QUEUE.md；.agent/LOOP_STATE.md（A1d/RT-02/RT-03/E026 运行与队列）。
Source: cognition-os/00_AI_BRIEFING.md (2026-10-14 D48 and the 2026-10-15 13Z update); cognition-os/experiments/E013 (the A2 v1, toy-MC and "A2 v2 final" sections), E019, E020 (the "Red-team RT-01 triage" section), E021, E014, E016, E018a, E022 (the correction-1 section), E023, E024, E025 (the correction-1 section), E017a; cognition-os/insights/I013, I034; .agent/jobs/2026-10-15-v01-verify-e020-e022/REPORT.luna.md; cognition-os/04_DECISION_LOG.md D49 (its update section withdrawn) and D50; cognition-os/REDTEAM_QUEUE.md; .agent/LOOP_STATE.md (the A1d/RT-02/RT-03/E026 runs and the queue).
本次更新What's new
- 2026-10-18 07:4xZ（刷新 10；来源 R037–R041、D63–D67、live_updates L030–L036）：(1) E029 的预注册判定规则在出任何结果前被收紧（D63，红队事前审查 RT-40，8 条意见全部采纳）——参考值改用 A1d 曲面的连续最小值（不是网格最小值）；同一组流量换不同权重会得到 0.817 或 0.850，所以必须同时报告「同口径加权」的总量，两者都偏离超过 0.044 且同号才改用新总量；模型比较只在原有参数边界内进行（E029 的宽边界只作敏感性报告）；若变化只来自少数几个场，按影响场检查记为这些场的系统误差。E029 仍为待定，不预期任何数值。(2) 放开 CXB 谱指数后被移走的流量在空间上是平的（外场 −7.3%、内场 −7.9%，与恒星质量无关，约为 M31 未分辨 X 射线双星预算的 360 倍；拟合改善与流量变化反相关）——「AGN 冒充热气体约 15–25%」已由 R037 撤回，归属开放；下降方向（13/14 场）成立，大小约 0.6\(\sigma\)。(3) E026 产品定为「v19 约定、重新掩膜」，不与 v19 可比（D64/D67，红队 RT-43）：为遮住 2.9 角分目标，开孔面积小 22–44%，且这批场无一有 v19 对应场；决定性年份检验 RT-09-T3 只比较 E026 早期与 E026 晚期，两边同一套流程，每边至少 3 个匹配场否则判不可行。(4) 本地热泡（LHB）分配项实测约 \(\pm 0.027\)（R038）：在 14 场允许的 LHB 范围内总量变化 −0.7% 到 −7.0%；−7% 一端由模型里 0.08 keV 的 LHB 温度下限造成（9/14 场卡在下限），下限放到 0.05 keV 后两个场方向反转；原先「把 LHB 冻在最小值会让总量升到 1.22」不在 1\(\sigma\) 范围内。(5) 引用总量改为 0.867 \(\pm 0.044\)（统计）\(+0.007/-0.003\)（NH，重拟合）\(\pm 0.027\)（LHB 分配），毛总量（D65/R039）：把柱密度换成 Willingale 总柱（比 HI4PI 高 19–31%）后重拟合，吸收流量只变 +0.8%，\(\pm10\%\) 只变不到 0.4%——旧的 \(\pm 0.034\) 固定归一化 NH 项已撤回；模型敏感性（CXB 指数、丰度）等 E029（待定）。(6) MOS 1.4–2 keV 排除对 0.5–2 keV 流量只移 0.37%（R040，红队 RT-42）：被排除频段只占流量的 0.37%；但该频段里很强的 Al/Si 仪器线的「响应翼」会漏进保留的 0.5–1.4 keV，上界相当于 CGM 流量的 12–21%、方向偏高、尚未量化——记为开放的软带系统误差，不进误差预算，总量数字不变。(7) PI 批准 E028 stage 1 与论文修订（D66）：stage 1 在 lang2 上三个真值并行、每个至少 200 次模拟（约 17–19 天），结束前 A2 上限仍写「阈值未校准」；论文修订（含把图 3 与 explorer 的硬编码 0.965 改为从同一文件读取）已启动。(8) 本站术语密度：byhand、history、index 三页超过 15.44/千字上限，已按「用平实词语替换标签」降到上限以下。2026-10-18 07:4xZ (refresh 10; sources R037–R041, D63–D67, live_updates L030–L036): (1) E029's pre-registered decision rule was tightened before any result (D63, red-team pre-mortem RT-40, all eight objections adopted) — the reference is now the continuous minimum of the A1d surfaces (not the grid minimum); because the same fluxes give 0.817 or 0.850 depending on the weights, a like-for-like weighted total must also be reported, and the quote moves only if both differ by more than 0.044 with the same sign; models are compared only within the original parameter bounds (E029's wider bounds are a sensitivity report only); a shift coming from a few fields is recorded, via an influence check, as a systematic of those fields. E029 stays Pending; no number is anticipated. (2) The flux removed by freeing the CXB photon index is spatially flat (outer −7.3%, inner −7.9%, unrelated to stellar mass, about 360 times the unresolved M31 X-ray-binary budget; fit improvement and flux change are anti-correlated) — the "15–25% AGN mimic" statement is withdrawn by R037 and the attribution is open; the direction of the drop (13/14 fields) holds, its size about 0.6\(\sigma\). (3) E026 products are "v19-convention, re-masked", not comparable to v19 (D64/D67, red team RT-43): the aperture is 22–44% smaller because the 2.9 arcmin target is masked, and no field in the batch has a v19 counterpart; the decisive year test RT-09-T3 compares E026-early with E026-late only, both sides from the same pipeline, at least 3 matched fields per side or the test is infeasible. (4) The Local Hot Bubble (LHB) allocation term is measured at about \(\pm 0.027\) (R038): over the 14 fields' allowed LHB ranges the total moves by −0.7% to −7.0%; the −7% end is set by the model's 0.08 keV LHB temperature floor (9/14 fields sit on it), and lowering the floor to 0.05 keV reverses the direction in two fields; the earlier "freezing the LHB at its minimum raises the total to 1.22" is outside the 1\(\sigma\) range. (5) The quoted total becomes 0.867 \(\pm 0.044\) (stat) \(+0.007/-0.003\) (N_H, refit) \(\pm 0.027\) (LHB allocation), gross (D65/R039): refitting with the Willingale total column (19–31% above HI4PI) moves the absorbed flux by only +0.8%, and \(\pm10\%\) by less than 0.4% — the old \(\pm 0.034\) fixed-normalisation NH term is withdrawn; model sensitivity (CXB index, abundance) awaits E029 (Pending). (6) The MOS 1.4–2 keV exclusion moves the 0.5–2 keV flux by only 0.37% (R040, red team RT-42): the excluded band carries 0.37% of the flux; but the strong Al/Si instrumental lines in it leak response wings into the kept 0.5–1.4 keV, with an upper bound of 12–21% of the CGM flux level, direction upward, not yet quantified — recorded as an open soft-band systematic, not in the error budget, and the total is unchanged. (7) The PI approved E028 stage 1 and the paper revision (D66): stage 1 runs the three truths in parallel on lang2, at least 200 simulations each (about 17–19 days), and until it ends the A2 limits keep the label "threshold uncalibrated"; the paper revision (which also makes Figure 3 and the explorer read the total from one pinned file instead of the hard-coded 0.965) has started. (8) This site's jargon density: byhand, history and index were above the 15.44 per-1k cap and have been brought to or below it by replacing labels with plain words.
- 2026-10-18 凌晨（刷新 9；来源 R029–R036、D60–D62、live_updates L024–L029）：(1) 外场变暗不是 2010 年的突变——全部 23 个外场随观测年份连续下降，斜率 −0.032 \(\pm 0.007\) 每年（t −4.4，对银纬、半径、黄纬与深度稳健），约 88% 的降幅在 2001–2009，最佳 2010 切分的显著性在多次切割校正后不成立（min-p 0.14）；用全部外场看变暗是真的、不是挑样本造成的，原因仍未解决（R029/L024，红队 RT-33）。(2) 能带换算依赖 kT：0.4–1.25 keV ↔ 0.5–2 keV 的换算在 14 场实际为 1.10–1.15（kT 0.10 时 1.49、0.30 时 1.06）；与 Locatelli 参考比较要乘 1.18，而 CGM 自身是 1.137（R031/L025）。(3) 图 3 的 MW 比较残差按总量 0.867 重算：Locatelli −0.42、Ueda 0.56、eFEDS 0.58/0.46、Henley–Shelton 0.48，没有符号变化；图 3 生成器与 explorer 仍用 0.965（B033），到论文那一轮再改（R032/L026）。(4) 0.867 是毛总量（gross）：含孔径内未分辨点源；D49 的 0.25 逐场项没有加到 14 个主场上（95% 上限 0.087）；从 0.965 降到 0.867 主要来自新的逐场权重（R033/D60/L027）。(5) 放开 CXB 谱指数使 CGMsum 降 6.2%（13/14 场，AIC 更优）；AGN-mimic 占比实测 15–25% 而非假设的 9%；「仅气体 0.82」撤回、不再作为名义值；Z = 0.3 是一个流量杠杆（Z 0.1/1.0 时 −6 至 −9%）；E029 正按预注册规则运行（D61），其结果为待定（R034/R035/D61/L028）。(6) 总 X 射线柱密度比 HI4PI 高 19–31%；\(\pm 0.034\) 的 NH 项是固定归一化的模型响应；真实效应等重拟合 RT-39-T1（待定）（R036/D62/L029）。(7) A2 图重画：0.867 的粗线标注「gross total, D57」，0.965 的灰线不再绘制（B033/R032）。2026-10-18 early UTC (refresh 9; sources R029–R036, D60–D62, live_updates L024–L029): (1) the outer-field dimming is not a 2010 step — across all 23 outer fields it is a continuous decline with observing year at −0.032 \(\pm 0.007\) per year (t −4.4, robust to Galactic latitude, radius, ecliptic latitude and depth), about 88% of the drop falls in 2001–2009, and the best 2010 cut's significance does not survive multiple-cut correction (min-p 0.14); the dimming is real on the full outer sample and is not a selection artefact, and the cause is unresolved (R029/L024, red team RT-33). (2) the band conversion depends on kT: the 0.4–1.25 keV ↔ 0.5–2 keV conversion is 1.10–1.15 over the 14 fields (1.49 at kT 0.10, 1.06 at kT 0.30); against the Locatelli reference the factor is 1.18, while the CGM itself is 1.137 (R031/L025). (3) the Figure-3 MW comparator residuals are recomputed at the total 0.867: Locatelli −0.42, Ueda 0.56, eFEDS 0.58/0.46, Henley–Shelton 0.48, with no sign change; the Figure-3 generator and explorer still use 0.965 (B033) until the paper pass (R032/L026). (4) 0.867 is the gross total: it includes unresolved sources inside the apertures; D49's 0.25 per-field term is not applied to the 14 primaries (95% upper limit 0.087), and the drop from 0.965 comes mostly from the new per-field weights (R033/D60/L027). (5) freeing the CXB photon index lowers the CGMsum by 6.2% (13/14 fields, AIC-preferred); the AGN-mimic share is measured at 15–25% rather than the assumed 9%; "gas-only 0.82" is withdrawn and no longer quoted as a nominal; Z = 0.3 is a flux lever (−6 to −9% at Z 0.1/1.0); E029 is running under a pre-registered rule (D61) and its result is Pending (R034/R035/D61/L028). (6) the total X-ray column is 19–31% above HI4PI; the \(\pm 0.034\) NH term is a fixed-normalisation model response, and the real effect awaits the refit RT-39-T1 (Pending) (R036/D62/L029). (7) the A2 figure is redrawn: the 0.867 line is labelled "gross total, D57" and the 0.965 grey line is no longer drawn (B033/R032).
- 2026-10-17 晚（刷新 8，红队 RT-32 传播审计后修订）：引用总量改为 0.867 \(\pm 0.044\)（统计）\(\pm 0.034\)（NH 变 \(\pm10\%\)，各场相关；该 \(\pm 0.034\) 是旧的固定归一化项，当日为最新值，现已由 D65 撤回）（D57/R024，A1d 多元起始 + 插值区间），取代 v19 单起始值 0.965 \(\pm 0.044\)（低 10%）——0.965 只作为历史值保留。模型 C 改为「敏感性检验」（D58/R026）：放开 SP 谱形后相对 A1d 为 −6 至 −12%（取决于加权方式，5/14 场反而变亮），AIC/BIC 在 14/14 场都更差——不是单向系统误差。E023 大核上限改为「被质疑」（R023/RT-28）：外场受观测年代混杂，加年代项后上限 0.87/1.29/2.38 且数据偏好有 M31 项；匹配年代外场重提取前，这些上限与「Zhang 名义中值被排除到 r_c 125–200 kpc」一律不引用。Locatelli 改为「有条件中心值张力」（R027/RT-29）：计入模型自身误差 \(\sigma \approx 0.32\) 后差距只有约 1.3\(\sigma\)，2\(\sigma\) 内允许正的 M31 残差；LHB 与 CGMsum 的分配仍待定。点源预算 0.12–0.33 F 不是上限（R020/RT-27）；「仅气体 0.82」已于 2026-10-18 R034/D61 撤回、不再作为名义值引用（放开 CXB 谱指数后 13/14 场流量下降、AGN-mimic 实测 15–25% 而非假设的 9%）。E028 stage 0 已被接受（R022/D56：CORRECTION_02 后闭合 9/9 达到或低于 A1d），阈值仍未校准，stage 1 需要 PI 决定（4 个 lang worker 每个真值 17–19 天）——不再写「阻断」。NW 外场改为「组间对比」而非「银纬梯度」（R025/RT-30）。A2 图与状态页、物理页的 0.965 一并更新；Zhang 线按 D54 表述为总体平均名义中值，不写「排除」。论文草稿（7 月冻结）仍带 0.965，修订方案已提交 PI（D59）。2026-10-17 evening (refresh 8, revised after red team RT-32's propagation audit): the quoted total becomes 0.867 \(\pm 0.044\) (stat) \(\pm 0.034\) (NH varied by \(\pm10\%\), correlated across fields; that \(\pm 0.034\) is the old fixed-normalisation term, current as of that day and since withdrawn by D65) (D57/R024, A1d multi-start plus interpolated intervals), superseding the v19 single-start 0.965 \(\pm 0.044\) (10 % lower) — 0.965 survives only as a historical value. Model C is reworded as "a sensitivity test" (D58/R026): with the SP shape free it is −6 to −12 % versus A1d depending on the weighting (5 of 14 fields get brighter), and AIC/BIC are worse in 14/14 fields — not a one-sided systematic. E023's large-core limits become "Questioned" (R023/RT-28): the outer fields are epoch-confounded, and with an epoch term the limits become 0.87/1.29/2.38 with an M31 term preferred; until matched-epoch outer fields are re-extracted, neither those limits nor "the Zhang nominal median is excluded to r_c 125–200 kpc" is quoted. Locatelli becomes "conditional central-value tension" (R027/RT-29): with the model's own error \(\sigma \approx 0.32\) the gap is only about 1.3\(\sigma\), and a positive M31 residual is allowed at 2\(\sigma\); the LHB/CGMsum split is still open. The point-source budget 0.12–0.33 F is not a ceiling (R020/RT-27); "gas-only 0.82" was withdrawn on 2026-10-18 by R034/D61 and is no longer quoted as a nominal (freeing the CXB photon index lowers the flux in 13/14 fields, and the AGN-mimic share is measured at 15–25 % rather than the assumed 9 %). E028 stage 0 is accepted (R022/D56: after CORRECTION_02 the closure is 9/9 at or below A1d), the threshold is still uncalibrated, and stage 1 needs a PI decision (17–19 days per truth on 4 lang workers) — no longer written as "blocked". The outer-NW fields are re-read as a group contrast, not a latitude gradient (R025/RT-30). The 0.965 in the A2 figure and on the state and physics pages is updated with them; the Zhang line is worded per D54 as a population-mean nominal median, never an "exclusion". The paper draft (frozen in July) still carries 0.965; a revision proposal has gone to the PI (D59).
- 2026-10-16 深夜（刷新 6 续，B 阶段，红队 RT-20 全站审计后修订）：RT-18/R013——宽场清单只覆盖本地磁盘：E021 的 145 个 ObsID 是 lang 磁盘上的，不是 XMM 档案；22.5° 锥的 XSA 查询列 563 个指向，418 个本地没有（395 个带公开科学标记，344 个在 30 kpc 以外，305 个计划时长 >10 ks），「没有遗漏可用观测」的说法撤回。E028 stage 0 闭合尚未达到：修正 1 后 stage-0 会话 22:56Z 退出，pilot 驱动仍在跑（19/40 csv），闭合未达标——0800731101 比 A1d 高 +0.34…+0.61 stat、0800730801 有一个节点 +0.059（原因：3 行热启动 vs A1d 的 11 行链）；不接纳进 stage 1，要求用 11 行链闭合（或把 stage 1 限制在能闭合的场上）。RT-20 审计修订（O1–O12）：Zhang 排除倍数改为「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」（RT-16：v19 曲面上去掉 0601 放宽 +46…+62%，模型 C 曲面下只放宽 6–30%）；E017a 残余点源预算取 0.12–0.33 F（「+0.05」撤回，探测器上亮源完备度 99.4%）；「外场更暗」改标受观测年代混杂（\(\le2009 1.049\) \(\pm 0.076\) vs \(\ge2010 0.551\) \(\pm 0.092\)，内场全部 2017–2018）；E027 覆盖改为「按 live K 计 3 个有用（按 GTI 总和计 4 个，2501 边缘；6/31 已打分观测是调参集）」；「观测上限 = toy 中位数」撤回；E028 stage 0 改标运行本身无效、stage 1 阻断；E016 高括号统一为 0.32；Zhang/上限（r_c \(\le 8\)）改为 3.9–11.0 倍。2026-10-16 late (refresh 6 continued, stage B, revised after red team RT-20's site-wide audit): RT-18/R013 — the wide-field inventory is disk-scoped: E021's 145 ObsIDs are the ones on lang's disk, not the XMM archive; a 22.5° XSA cone lists 563 pointings, of which 418 are absent locally (395 carry a public science flag, 344 lie beyond 30 kpc, 305 have a scheduled duration >10 ks), so the claim that no usable observation is missed is withdrawn. E028 stage 0's closure is not yet met: after correction 1 the stage-0 session exited at 22:56Z with the pilot drivers still running (19/40 csv) and the closure not met — 0800731101 sits +0.34…+0.61 stat above A1d and 0800730801 has one node +0.059 (cause: a 3-row warm start vs A1d's 11-row chain); it is not accepted for stage 1, and the required fix is closure with the 11-row chain (or restricting stage 1 to the fields that close). RT-20 audit revisions (O1–O12): the Zhang exclusion factor becomes "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)" (RT-16: dropping 0601 loosens by +46…+62 % on the v19 surfaces but only 6–30 % under Model C); E017a's residual point-source budget is 0.12–0.33 F ("+0.05" withdrawn, on-detector bright completeness 99.4 %); "the outer fields are fainter" is re-marked epoch-confounded (\(\le2009 1.049\) \(\pm 0.076\) vs \(\ge2010 0.551\) \(\pm 0.092\), the inner fields all 2017–2018); E027's coverage becomes "3 useful by live K (4 by the GTI sum, 2501 borderline; 6 of the 31 scored observations are the tuning set)"; "the observed limits = the toy median" is withdrawn; E028 stage 0 is re-marked invalid as run with stage 1 blocked; E016's high bracket is standardised on 0.32; and Zhang/UL (r_c \(\le 8\)) becomes 3.9–11.0\(\times\).
- 2026-10-16 晚（刷新 6 续，A 阶段）：A1d 已验收，A2 基线在它上面重算——集中核（r_c \(\le 8 \mathrm{kpc}\)）20 kpc 上限 0.12–0.16（\(\beta 0.4\)）／0.075–0.10（\(\beta 0.5\)）／0.049–0.071（\(\beta 0.6\)），Zhang 名义中值 / 上限（r_c 8，\(\beta 0.4\)）3.5–5.2 倍；红队 RT-13 复核通过但指出上限主要靠一个视场（仅 v19 曲面），红队 RT-16 复核：模型 C 曲面下该集中度基本消失，措辞改为「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」，报出的值偏保守（插值与流量换算再紧 2–23%）；红队 RT-08 判定「上限已校准良好」不成立，\(\Delta \mathrm{stat} = 2.71\) 形式正确但尚未用能谱模拟校准（E028 运行中）。外场流量随观测年代下降（\(\le2009 1.05\) \(\pm 0.08\) vs \(\ge2010 0.55\) \(\pm 0.09\)，约 2\(\sigma\)）——「怀疑探测器老化」撤回，1.3 keV 特征是自由的 1.2–1.4 keV 高斯残差而非仪器线，外/内比与基于外场的极限改标「受观测年代混杂」。红队 RT-11（阻断级）：E026 的 runner 会让全部 28 个场失败，修正通过前不批处理。E027：预注册修订 v2 + 覆盖收益改为「按 live K 计 3 个有用非 v19 指向（按 GTI 总和计 4 个，2501 边缘；6/31 已打分观测是调参集）」+ 203 个 GTI 文件验收。新图 fig_a2_ul_vs_rc_A1d.png。已分诊：RT-14-T1、RT-15（判 E028 stage 0 运行本身无效）、RT-16。运行中（待定）：E028 stage 0 修正 1 与 stage 1（阻断）、E026 修正、E027 1b/1c、RT-09-T3。2026-10-16 evening (refresh 6 continued, stage A): A1d is accepted and the A2 baseline is recomputed on it — the compact-core (r_c \(\le 8 \mathrm{kpc}\)) limit at 20 kpc is 0.12–0.16 (\(\beta 0.4) / 0.075\)–0.10 (\(\beta 0.5) / 0.049\)–0.071 (\(\beta 0.6\)), Zhang's nominal median / UL (r_c 8, \(\beta 0.4\)) is 3.5–5.2\(\times\); red team RT-13's re-check survives but finds the limit leans on one field on the v19 surfaces only, and red team RT-16 re-checked it: under the Model C surfaces that concentration largely disappears, so the wording becomes "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)", and the quoted values are conservative (the interpolant and flux conversion would tighten them by 2–23 %); red team RT-08 rules that "the limits are well calibrated" does not survive — \(\Delta \mathrm{stat} = 2.71\) is right in form but not yet calibrated by spectral simulation (E028 running). The outer-field flux falls with observing year (\(\le2009 1.05\) \(\pm 0.08\) vs \(\ge2010 0.55\) \(\pm 0.09\), about 2\(\sigma\)) — "detector ageing suspected" is withdrawn, the 1.3 keV feature is a free 1.2–1.4 keV Gaussian residual not an instrumental line, and the outer/inner ratio and outer-based limits are marked "epoch-confounded". Red team RT-11 (blocking): E026's runner would fail all 28 fields, and no batch starts before the correction passes. E027: pre-registration amendment v2 + the coverage gain reworded to "3 useful non-v19 pointings by live K (4 by the GTI sum, 2501 borderline; 6 of the 31 scored observations are the tuning set)" + 203 GTI files accepted. New figure fig_a2_ul_vs_rc_A1d.png. Triaged: RT-14-T1, RT-15 (which ruled E028 stage 0 invalid as run), RT-16. Running (Pending): E028 stage 0's correction 1 and stage 1 (blocked), E026's corrections, E027 1b/1c, RT-09-T3.
- 2026-10-16（第 6 次刷新，组会优先）：红队 RT-02/RT-03/RT-04/RT-05/RT-06/RT-07 全部分诊完毕——集中核上限对插值方式敏感 2.0–2.3 倍（样条比线性松；r_c 4、\(\beta 0.5\)：样条 0.061、线性 0.027、修掉卡点后 0.044），收敛值待 A1d；Zhang 比值当时写作「\(\gtrsim4\)\(\times\)」（RT-07：排除名义中值剖面，不写「排除叠加预期」——Zhang 只发布 16/84% 边际、无协方差；v19 APEC 桥给 0.76–0.90；该比值后来经红队 RT-16 修订为「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」，见刷新 6 续条目）；E023 的「\(\beta = 0.6\) 时 r_c \(\le 200\) 全部排除」撤回（该端上限 0.90–1.06，边际），E023 将在流量空间于谱曲面上重做；外/内稳妥说法改为「约 10–15% 内一致」（随机效应 0.94 \(\pm 0.11\)，逐一剔除 0.83–0.99）；v19 总量确认为局部单起始基线（多元起始移到 0.891，−7.7%），最终总量来自 A1d 剖面曲面；Locatelli 改为「有条件紧张」；未解源括号 0.12–0.33 F（不写「0.28 保守」）；D49 措辞收窄（RASS 只排除宽的 R45 亮超出；OMNI 只排除 OMNI 追踪的高通量 SWCX），NW 四场 SP/QPB 约为其余十场 3\(\times\)（事后）——「未解释」倾向软质子。E027（PI 两次要求，范围改为全部 31 个 LP 观测）stage 0/0b 完成：GATE 不可用（丢 5–6 个主场）；FLR_z3 与 MS_z3.5_core 清洗相同、代价 2–3% vs ~10% 曝光；主要价值是覆盖（当时写「多达约 11 个非 v19 指向可能变成可用场」；同日 09:0xZ 红队 RT-10 复核后改为「约 4 个有用 + 5 个边缘」；当天 21:56Z 红队 RT-19 再改为「按 live K 计 3 个有用（按 GTI 总和计 4 个，2501 边缘；6/31 已打分观测是调参集）」）；谱级 S1–S4 已预注册，stage 1 的 1a GTI 导出 06:26Z 派发。新增组会页 meetings.html 与首页组会卡（PI 2026-10-16 要求）。运行中（待定）：A1d（~1240/1638，ETA ~09:30Z）、E026（Virgo ssh 06:16Z–~08:50Z 中断，已自行恢复）、红队 RT-08（报告已交）与 RT-09（后来均已分诊，见上条）。2026-10-16 (refresh 6, the group meeting first): red teams RT-02/RT-03/RT-04/RT-05/RT-06/RT-07 are all triaged — the compact-core UL is interpolation-sensitive by 2.0–2.3\(\times\) (spline looser than linear; r_c 4, \(\beta 0.5\): spline 0.061, linear 0.027, 0.044 with the stuck point fixed), the converged value awaits A1d; the Zhang ratio was then written "\(\gtrsim4\)\(\times\)" (RT-07: exclude the nominal median profile, never "the stacked expectation" — Zhang publishes only 16/84 % marginals, no covariance; the v19 APEC bridge gives 0.76–0.90; that ratio was later revised by red team RT-16 to "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)" — see the refresh-6-continued entry); E023's "for all r_c \(\le 200\) at \(\beta = 0.6\)" is withdrawn (that end's limit is 0.90–1.06, marginal) and E023 will be redone in flux space on spectral surfaces; the outer/inner safe wording becomes "agree within about 10–15%" (random effects 0.94 \(\pm 0.11\), leave-one-out 0.83–0.99); the v19 total is confirmed as a local single-start baseline (multi-start moves it to 0.891, −7.7%), the final total coming from the A1d profile surfaces; Locatelli becomes "conditional tension"; the unresolved-source bracket is 0.12–0.33 F (not "0.28 conservative"); D49's wording is narrowed (RASS excludes only a broad R45-bright excess; OMNI excludes only OMNI-traced high-flux SWCX) and the NW four's SP/QPB is about 3\(\times\) the other ten (a posteriori) — "unexplained" leans towards soft protons. E027 (the PI's two requests, rescoped to all 31 LP observations) stages 0/0b done: GATE is unusable (it drops 5–6 primaries); FLR_z3 and MS_z3.5_core clean identically at 2–3% vs ~10% of the exposure; the main value is coverage (then written as "up to ~11 non-v19 pointings may become usable fields"; the same day at 09:0xZ red team RT-10's review reworded it to "about 4 useful and 5 marginal", and at 21:56Z red team RT-19 re-reworded it to "3 useful by live K (4 by the GTI sum, 2501 borderline; 6 of the 31 scored observations are the tuning set)"); the spectral S1–S4 metrics are pre-registered and stage 1's 1a GTI export was dispatched at 06:26Z. New: the meetings.html group-meeting page and the home group-meeting card (the PI's 2026-10-16 request). Running (Pending): A1d (~1240/1638, ETA ~09:30Z), E026 (the Virgo ssh outage 06:16Z–~08:50Z, it came back by itself), red teams RT-08 (report filed) and RT-09 (both triaged later, see the item above).
- 2026-10-15（第 5 次刷新）：红队 RT-01 分诊——E020 模型 C 的「−15% SP 谱形系统误差」不成立：−15.04% 中 v19 重收敛（A）−7.65%、SP 谱形 −5.29%、线能量 −2.87%；NW/其余比值（共同误差）1.345 → 1.133（A）→ 1.134（C）——下降在重收敛；I033 第 5 条回到 Questioned，D49 原文生效，D49 更新撤回，模型 C 降级为敏感性变体；A1d（v19 多元起始曲面）派发为 A2 新基线。D50（PI）：外 B 场政策 = P1+P2 立即拟合 + 6 场 Virgo 重提取，跳过 P4，弃 NGC 147。E025 完成：P1 12/12 掩膜审计未过；P2 18/19 接受；干净外样本（13 场）0.846 \(\pm 0.059\) vs 内场 0.965 \(\pm 0.044\)，外/内 = 0.877 \(\pm 0.072\)；全部 23 场 0.831 \(\pm 0.045\)；无 b/N_H 趋势；correction 1 后重提取清单 = 28 场（E026 运行中）。A2 v2 final（敏感性变体）：\(\beta = 0.5\) 时 r_c = 4/8/15/30 = 0.111/0.138/0.215/0.484；toy-MC 覆盖率 0.918–0.933（< 0.95，需校准阈值）；Zhang/上限（r_c \(\le 8\)）= 3.9–11.0 倍。I034（ParetoGTICal 交叉核对）：v19 用 2.1.filter GTI（比 2.filter 严）；NW 四场没有被欠过滤；稳态 SP 对 GTI 不可见；~10% 场间底噪。评审更正已并入：Zhang 0.83 排除到 r_c \(\lesssim 125/175 \mathrm{kpc}\)（\(\beta = 0.4/0.5\)），\(\beta = 0.6\) 时 r_c \(\le 200\) 全部排除；EAGLE 0.51 到 r_c \(\lesssim 45/65/85 \mathrm{kpc}\)；E022/E023 两图已重绘。运行中（待定）：A1d、RT-02（已发现 A1 曲面优化卡点）、RT-03、E026。 2026-10-15 (refresh 5): red team RT-01 triaged — E020's "−15% SP-shape systematic" does not hold: of −15.04%, the v19 re-convergence (A) gives −7.65%, the SP shape −5.29%, the line energies −2.87%; the NW/other ratio (common errors) goes 1.345 → 1.133 (A) → 1.134 (C) — the drop is at re-convergence; I033 point 5 is back to Questioned, D49 stands as written, the D49 update is withdrawn, Model C is demoted to a sensitivity variant; A1d (v19 multi-start surfaces) dispatched as the new A2 baseline. D50 (the PI): the outer class-B policy = fit P1+P2 now + 6 Virgo re-extractions, skip P4, drop NGC 147. E025 done: all 12 P1 fields fail the mask audit; 18/19 P2 fits accepted; the clean outer sample (13 fields) is 0.846 \(\pm 0.059\) vs the inner 0.965 \(\pm 0.044\), outer/inner = 0.877 \(\pm 0.072\); all 23 fields 0.831 \(\pm 0.045\); no b/N_H trend; after correction 1 the re-extraction list is 28 fields (E026 running). A2 v2 final (a sensitivity variant): for \(\beta = 0.5\) at r_c = 4/8/15/30 = 0.111/0.138/0.215/0.484; toy-MC coverage 0.918–0.933 (< 0.95, a calibrated threshold needed); Zhang/UL (r_c \(\le 8) = 3.9\)–11.0x. I034 (the ParetoGTICal cross-check): v19 uses the 2.1.filter GTIs (stricter than 2.filter); the NW four are not under-filtered; a steady SP is invisible to GTIs; a ~10% between-field floor. The review correction is folded in: Zhang 0.83 excluded up to r_c \(\lesssim 125/175 \mathrm{kpc}\) (\(\beta = 0.4/0.5\)), and for all r_c \(\le 200\) at \(\beta = 0.6\); EAGLE 0.51 up to r_c \(\lesssim 45/65/85 \mathrm{kpc}\); the E022/E023 figures redrawn. Running (Pending): A1d, RT-02 (stuck points already found in the A1 surfaces), RT-03, E026.
- 2026-10-14（第 3 次刷新）：E013 A2 v1 + toy MC 完成——无 M31 分量被偏好（\(\Delta \mathrm{stat}\) \(\le 1.07\)）；集中式 M31 \(\beta\) 晕（r_c \(\le 8 \mathrm{kpc}\)，MW 常数）在 20 kpc 的插值单侧 95% 上限 0.047–0.164，Zhang 叠加 0.83 被排除 \(\gtrsim5\) 倍（各 \(\beta\) 为 5–18 倍；第 6 次刷新时红队 RT-02 把该比值降到 \(\gtrsim4\)\(\times\)，刷新 6 续时红队 RT-16 再修订为「Zhang/上限 >~3.5x（r_c 8，beta 0.4，incl. Model C）」）；toy-MC 覆盖率 0.975–0.984（略保守；后来红队 RT-08 撤回「观测上限 = toy 中位数」——v1 比中位数低 0.8–10.8%）；新图 fig_a2_ul_vs_rc.png。E020 完成：模型 C 下 14 场均值 0.964 → 0.819（当时记为单侧 −15%；第 5 次刷新时红队 RT-01 推翻该解读），径向斜率 0，当时解读为 b 梯度与 NW 超出大半是固定 SP 谱形的伪迹（现已被 RT-01 推翻：见第 5 次刷新）。E019 完成：PSF 翼泄漏 0.16（原值）／0.33（零点校正）vs 模型 0.12。E021 完成：30 kpc 以外仅 8 个 class-A 档案场；新图 fig_e021_sky_coverage.png。当时提出 D49 更新提案（改记 −15% 系统误差）；第 5 次刷新时已撤回。A2 v2（E020 模型 C 曲面）与 E022 在跑。 2026-10-14 (refresh 3): E013 A2 v1 + toy MC done — no M31 component is preferred (\(\Delta \mathrm{stat}\) \(\le 1.07\)); the interpolated one-sided 95% UL on a concentrated M31 beta halo (r_c \(\le 8 \mathrm{kpc}\), MW constant) at 20 kpc is 0.047–0.164, excluding the Zhang stack 0.83 by \(\gtrsim5\)x (5–18x across \(\beta\); at refresh 6 red team RT-02 lowered that ratio to \(\gtrsim4\)\(\times\), and in refresh 6 continued red team RT-16 revised it to "Zhang/UL >~3.5x (r_c 8, beta 0.4, incl. Model C)"); toy-MC coverage 0.975–0.984 (slightly conservative; red team RT-08 later withdrew "the observed limits sit at the toy median" — v1 sits 0.8–10.8 % below it); new figure fig_a2_ul_vs_rc.png. E020 done: under Model C the 14-field mean moves 0.964 → 0.819 (then read as a one-sided −15%; that reading was overturned by red team RT-01 at refresh 5), the radial slope is 0, and the b gradient and NW excess were then read as mostly a fixed-SP-shape artefact (now overturned by RT-01: see refresh 5). E019 done: the PSF-wing leak is 0.16 raw / 0.33 null-corrected vs the model 0.12. E021 done: only 8 class-A archival fields beyond 30 kpc; new figure fig_e021_sky_coverage.png. The D49-update proposal (re-carrying the −15% systematic) was raised then; it was withdrawn at refresh 5. A2 v2 (on the E020 Model-C surfaces) and E022 are running.
- 2026-10-14（第 2 次刷新）：E018a 完成——外 NW 超出大半是拟合伪迹：重收敛 + SP 谱形自由后外四/对照四比值 1.295 → 1.168 → 1.130（\(\sigma\approx0.13\)，\(\approx1\)\(\sigma\)，与对照持平）；0601 的 \(\Delta \mathrm{stat}\) −14.7 来自 MOS1 InstLine3.LineE 钉在 1.25 keV；D49 的 0.25 系统误差可望缩到 \(\approx0.1\)。新图 fig_e018a_fluxes.png。 2026-10-14 (refresh 2): E018a done — much of the outer-NW excess is a fitting artefact: after re-convergence and a free SP shape the outer-four/control-four ratio goes 1.295 → 1.168 → 1.130 (\(\sigma\approx0.13\), \(\approx1\)\(\sigma\), no significant difference from the controls); 0601's \(\Delta \mathrm{stat}\) −14.7 comes from MOS1 InstLine3.LineE pinned at 1.25 keV; D49's 0.25 systematic can shrink to \(\approx0.1\). New figure fig_e018a_fluxes.png.
- 2026-10-14：E016 完成——未解点源的软热分量 \(\approx0.15 \mathrm{F}\)（高括号 0.28），空间上平（不削弱形状上限），仅气体的 CGMsum 总和 \(\approx0.82\)（0.69）——该「仅气体 0.82」已于 2026-10-18 R034/D61 撤回、不再作为名义值引用（放开 CXB 谱指数后 13/14 场流量下降、AGN-mimic 实测 15–25% 而非假设的 9%），当时的 Locatelli 紧张更强；最大的单一不确定项是 PSF 翼泄漏（\(\approx0.12 \mathrm{F}\)，模型估计）→ E019 排队。同日 E013 A1 完成——20/20 场的 (kT, norm) \(\Delta \mathrm{stat}\) 曲面（恢复回归 20/20、0 失败拟合；0601 的网格极小值比 v19 低 12.2，再次证实未收敛），A2（权威 M31 \(\beta\) 上限）随后运行并待定。2026-10-14: E016 done — the unresolved-source soft-thermal term is \(\approx0.15 \mathrm{F} (0.28\) at the high bracket), spatially flat (it does not weaken the shape limit); the gas-only CGMsum total is \(\approx0.82 (0.69\)) — that "gas-only 0.82" was withdrawn on 2026-10-18 by R034/D61 and is no longer quoted as a nominal (freeing the CXB photon index lowers the flux in 13/14 fields, and the AGN-mimic share is measured at 15–25 % rather than the assumed 9 %), strengthening the Locatelli tension at the time; the single largest uncertain term is the PSF-wing leak (\(\approx0.12 \mathrm{F}\), a model estimate) → E019 queued. The same day E013 A1 finished — (kT, norm) \(\Delta \mathrm{stat}\) surfaces for 20/20 fields (restore regression 20/20, 0 failed fits; 0601's grid minimum beats the v19 fit by 12.2, confirming the non-convergence), and A2 (the authoritative M31 beta limit) then ran to a Pending result.
- 2026-10-14：E014 完成，D49 落笔：外 NW 超出不是大尺度天空（RASS 0.91 \(\pm 0.12\) vs XMM 1.28），不是 OMNI 太阳风（无窗口超 4e8），软质子相关性未确立（Spearman +0.35，p=0.22）；四场记为未解释场级系统误差（~0.25）。D48 同日冻结 v19 测量；论文定名 New-ANGELS III；本工作台首次建成。 2026-10-14: E014 finished and D49 recorded: the outer-NW excess is not large-scale sky (RASS 0.91 \(\pm 0.12\) vs XMM 1.28), not OMNI-traced solar wind (no window above 4e8), and the soft-proton correlation is not established (Spearman +0.35, p=0.22); the four fields are carried as an unexplained field-level systematic (~0.25). D48 froze the v19 measurement the same day; the paper was named New-ANGELS III; this workbench was first built.
来源：2026-10-18 07:4xZ 条目（刷新 10）：cognition-os/07_REVISION_LOG.md R037–R041、cognition-os/04_DECISION_LOG.md D63–D67、handbook/live_updates.jsonl L030–L036，以及各条对应实验记录（E029 判定规则收紧见 D63 与红队 RT-40；放开谱指数后移走流量的空间平坦性与撤回见 R037 与 experiments/E016；E026 验收政策与 RT-09-T3 设计见 D64/D67 与 R041、experiments/E026；本地热泡分配项见 R038 与 insights/I017；NH 重拟合与引用总量见 D65 与 R039（RT-39-T1）；1.4–2 keV 排除与软带响应翼见 R040 与 D67；PI 批准 E028 stage 1 与论文修订见 D66）。2026-10-18 凌晨条目（刷新 9）：cognition-os/07_REVISION_LOG.md R029–R036、cognition-os/04_DECISION_LOG.md D60–D62、handbook/live_updates.jsonl L024–L029，以及各条对应实验记录（外场年代斜率见 cognition-os/experiments/E025「RT-33」节；能带换算见 R031 与 physics.html §5；图 3 残差重算见 R032/B033；毛总量与权重分解见 R033/D60；CXB 谱指数与 Z 杠杆见 R034/R035/D61；总柱密度与 NH 项见 R036/D62）；更早条目：cognition-os/experiments/E013（「A1d accepted and A2 baseline on A1d」「Red-team RT-08 triage」「Red-team RT-13 triage」节；历史条目的 A2 v1、toy-MC 与「A2 v2 final」节）、E019、E020（「Red-team RT-01 triage」节）、E021、E018a、E016、E022（correction 1 节）、E023、E024、E025（correction 1 节、RT-09/RT-09-T1/RT-12/RT-09-T2/RT-14 与 Hermes 臂修订节）、E027（stage 1a、RT-10 修订 v2 节）、E028、E017a；.agent/jobs/2026-10-15-e026-virgo-reextract-stage0/CORRECTION_01_RT11.md；.agent/jobs/2026-10-15-v01-verify-e020-e022/REPORT.luna.md；cognition-os/04_DECISION_LOG.md D48、D49（update 节已撤回）与 D50；cognition-os/insights/I033、I034；cognition-os/REDTEAM_QUEUE.md；cognition-os/06_AGENT_LOG.md 181–196 行；handbook/live_updates.jsonl L001–L008。
Source: the 2026-10-18 07:4xZ entry (refresh 10): cognition-os/07_REVISION_LOG.md R037–R041, cognition-os/04_DECISION_LOG.md D63–D67, handbook/live_updates.jsonl L030–L036, plus the matching experiment records (the tightened E029 decision rule from D63 and red team RT-40; the spatial flatness of the flux removed by freeing the photon index and its withdrawal from R037 and experiments/E016; the E026 acceptance policy and the RT-09-T3 design from D64/D67 and R041 plus experiments/E026; the LHB allocation term from R038 and insights/I017; the NH refit and the quoted total from D65 and R039 (RT-39-T1); the 1.4–2 keV exclusion and the soft-band response wings from R040 and D67; the PI's approval of E028 stage 1 and the paper revision from D66). The 2026-10-18 early-UTC entry (refresh 9): cognition-os/07_REVISION_LOG.md R029–R036, cognition-os/04_DECISION_LOG.md D60–D62, handbook/live_updates.jsonl L024–L029, plus the matching experiment records (the outer-year slope is from the "RT-33" section of cognition-os/experiments/E025; the band conversion from R031 and physics.html §5; the Figure-3 residual recomputation from R032/B033; the gross total and weight decomposition from R033/D60; the CXB photon index and the Z lever from R034/R035/D61; the total column and NH term from R036/D62); earlier entries: the "A1d accepted and A2 baseline on A1d", "Red-team RT-08 triage" and "Red-team RT-13 triage" sections of cognition-os/experiments/E013 (plus the A2 v1, toy-MC and "A2 v2 final" sections for the historical entries), E019, E020 (the "Red-team RT-01 triage" section), E021, E018a, E016, E022 (the correction-1 section), E023, E024, E025 (the correction-1 section and the RT-09/RT-09-T1/RT-12/RT-09-T2/RT-14 sections plus the Hermes-arm amendment), E027 (the stage-1a and RT-10-amendment-v2 sections), E028, E017a; .agent/jobs/2026-10-15-e026-virgo-reextract-stage0/CORRECTION_01_RT11.md; .agent/jobs/2026-10-15-v01-verify-e020-e022/REPORT.luna.md; cognition-os/04_DECISION_LOG.md D48, D49 (its update section withdrawn) and D50; cognition-os/insights/I033, I034; cognition-os/REDTEAM_QUEUE.md; rows 181–196 of cognition-os/06_AGENT_LOG.md; handbook/live_updates.jsonl L001–L008.
数据截至Data as of
2026-10-18T07:4xZ（最新读到的来源：cognition-os/07_REVISION_LOG.md R037–R041；cognition-os/04_DECISION_LOG.md D63–D67；handbook/live_updates.jsonl L030–L036；cognition-os/experiments/E016（R037 的移走流量分析）；cognition-os/insights/I017（本地热泡分配，R038）；newangels3_discussion/RT39T1_codexnim/per_field.csv（NH 重拟合，D65/R039）；cognition-os/experiments/E026（验收政策，D64/D67/R041）；handbook/live_updates.jsonl L030–L036；此前条目沿用 cognition-os/07_REVISION_LOG.md R029–R036 与 04_DECISION_LOG.md D60–D62newest source read: rows R037–R041 of cognition-os/07_REVISION_LOG.md; D63–D67 of cognition-os/04_DECISION_LOG.md; handbook/live_updates.jsonl L030–L036; cognition-os/experiments/E016 (the removed-flux analysis of R037); cognition-os/insights/I017 (the LHB allocation, R038); newangels3_discussion/RT39T1_codexnim/per_field.csv (the NH refit, D65/R039); cognition-os/experiments/E026 (the acceptance policy, D64/D67/R041); earlier entries still rest on rows R029–R036 of cognition-os/07_REVISION_LOG.md and D60–D62 of 04_DECISION_LOG.md）