组会：2026-10-21（UTC）
Group meeting: 2026-10-21 (UTC)
每次刷新把"组会卡片"放在首页最上面：像代理每天向 PI 汇报一次。本页显示最新一天（UTC 日期）的完整内容；更早的日子见组会存档（2026-10-20、2026-10-19）。
Every refresh opens the home page with the group-meeting card: as if the agent reported to the PI once a day. This page shows the latest day (UTC date) in full; earlier days are in the meeting archive (2026-10-20, 2026-10-19).
成果 / Results
Results
- 手册刷新 3 启动（Hermes，本页）：从与线上逐字节相同的树出发，把内容推进到
cognition-os/08_RUNNING.mdrevision 40（2026-10-20 22:55 EDT）。已确立Handbook refresh 3 launched (Hermes, this page): starting from a tree byte-identical to the live site, the content is brought up tocognition-os/08_RUNNING.mdrevision 40 (2026-10-20 22:55 EDT). established - 线上站点的 62 处 LaTeX 定界符不可渲染（本刷新自查发现）：上一轮刷新留下的双反斜杠不是 KaTeX auto-render 认识的定界符，所以组会存档页上的公式在浏览器里显示成字面文本；
check_math.py仍报 PASS，因为它的"未渲染定界符"只数单反斜杠残留。已修正并用无头 Chromium 复核：14 个渲染出的公式、0 个渲染错误。已确立62 LaTeX delimiters on the live site did not render (found by this refresh's self-check): the doubled backslashes left by the previous refresh are not a delimiter KaTeX's auto-render recognises, so the formulas on the meeting-archive page showed as literal text in the browser;check_math.pystill printed PASS because its "unrendered delimiter" test counts only single-backslash leftovers. Fixed and re-verified in headless Chromium: 14 rendered formulas, 0 render errors. established - 曝光图定标降级为"未定"，且步长对响应残差非常敏感：红队检查（RT-M119）指出预登记联合规则在当前样本量下不可能通过（零假设下预期 2.8 个单元失败、实际 2 个；弥漫臂 \(\beta=0.20\) 的移动 \(-7.3\) 个百分点小于其 \(\sigma\approx7.6\)），所以"vignetting 残差不是主因"不是判决，\(\textit{R75 为未定}\)；功效重放显示 \(\beta=0.10\) 就把同相机步长移 \(-6.86\) 个百分点（MSC-R77）。ProvisionalThe exposure-map verdict is downgraded to inconclusive, and the step is very sensitive to a response residual: a red-team check (RT-M119) shows the pre-registered joint rule could not be passed at this sample size (2 fails against 2.8 expected under the null; the diffuse arm's move at \(\beta=0.20\), \(-7.3\) percentage points, is smaller than its \(\sigma\approx7.6\)), so \"vignetting residual not the main cause\" is not a verdict and \(\textit{R75 is inconclusive}\); the power replay shows \(\beta=0.10\) alone moves the same-camera step by \(-6.86\) percentage points (MSC-R77). Provisional
- 把 M31 步长归零需要多大的响应残差，实测又有多大：归零所需 \(\beta\)（对数响应斜率，\(x=(\theta/10')^2\)）为 MOS1 \(0.21\pm0.24\)、MOS2 \(0.043\pm0.065\)、合并 \(0.084\pm0.066\)（MSC-R80）；即 \(\beta\approx0.08\)、9′ 到 13′ 约 7 % 的残差就能抹掉步长。PI 提出的 4XMM 目录重复探测率曝光图（MSC-R81/R81b）经选择感知校正后给 MOS1 \(+0.042/+0.066\)、MOS2 \(+0.051/+0.051\)、pn \(+0.012/+0.006\)（0.5–1 / 1–2 keV）——MOS-only 残差约为所需的一半到全部，但它随 S/N 切割上升、在 30 % 变源模型下减半，仍为 Provisional，正由事件文件孔径测光独立复核（MSC-R82）；同天空弥漫重叠（MSC-R79）给 MOS2 \(+0.018\pm0.022\)（SP 先验）或 \(+0.040\pm0.011\)（自由 SP）、MOS1 \(+0.050\pm0.075\)。ProvisionalHow large a response residual would zero the M31 step, and how large is measured: the needed \(\beta\) (log-response slope in \(x=(\theta/10')^2\)) is MOS1 \(0.21\pm0.24\), MOS2 \(0.043\pm0.065\), combined \(0.084\pm0.066\) (MSC-R80); i.e. a residual of \(\beta\approx0.08\), about 7 per cent between 9′ and 13′, would erase the step. The PI's 4XMM repeat-detection exposure map (MSC-R81/R81b), after the selection-aware correction, gives MOS1 \(+0.042/+0.066\), MOS2 \(+0.051/+0.051\), pn \(+0.012/+0.006\) (0.5–1 / 1–2 keV) — a MOS-only residual of about half to the full size needed, but it rises with the S/N cut and halves under a 30 per cent variability model, so it stays Provisional and is being checked by aperture photometry from event files (MSC-R82); same-sky diffuse overlaps (MSC-R79) give MOS2 \(+0.018\pm0.022\) (SP prior) or \(+0.040\pm0.011\) (free SP), MOS1 \(+0.050\pm0.075\). Provisional
- 专家顾问的问题成为新任务主线：我们是否有足够的独立信息来判别数据要求哪一种天空，还是只是在软质子、粒子背景、响应和天空之间挪动同一批光子，直到图看起来平了？→ MSC-R83（冻结流程的盲预测 + 信号注入）；同日采纳 P0-A/P0-B（→ MSC-R82）/P1-C/P1-D 提案。ProvisionalThe expert advisor's question becomes the lead of the new tasks: do we have enough independent information to say which sky the data require, or are we moving the same photons between soft protons, particle background, response and sky until the map looks flat? → MSC-R83 (frozen-pipeline blind prediction with signal injection); the P0-A/P0-B (→ MSC-R82)/P1-C/P1-D proposals adopted the same day. Provisional
- 盲预测失败了，而失败的方式排除了离轴响应这条解释（rev 45–46）：冻结流程的盲预测里，M31 的一对近重复观测 MOS2 差 \(z=4.03\)、联合 \(\chi^2=26.74/6\)（\(p=1.6\times10^{-4}\)），MOS1 同一对却一致（\(z=0.10\)）。把 MOS 离轴残余曲线乘回去不能闭合它（MSC-R84：\(\chi^2\) 只从 26.74/6 动到 25.74–26.21/6，\(p<3\times10^{-4}\)；近重复 \(z\) 4.03 → 3.95）——因为失败的那一对 0109270301/0402561501 在天空中同一个位置（相距 \(0.33'\)，重放几何确认），离轴曲线对它两个观测几乎相同。所以这是一个观测之间的差异（约 24 %），两条观测的 SP（8.60 对 9.96e-5）与 QPB（1.02 对 1.68e-4）状态都不同。逐项分解（MSC-R85）显示：原始计数差在 MOS1/MOS2 里几乎相等，MOS2 的失败来自减掉了更大的 QPB（+0.240e-6）与 SP（+0.160e-6）差；同一 CCD 子集 \(z=3.92\)，共同天空上 MOS2−MOS1 只 1.22\(\sigma\)。原因未识别，下一步按能段分离 QPB/SP/SWCX（MSC-R85b）。未解决The blind prediction failed, and the way it failed rules out the off-axis response (rev 45–46): in the frozen-pipeline blind prediction one M31 near-repeat pair differs in MOS2 at \(z=4.03\), joint \(\chi^2=26.74/6\) (\(p=1.6\times10^{-4}\)), while MOS1 of the same pair agrees (\(z=0.10\)). Multiplying the MOS off-axis residual curve back in does not close it (MSC-R84: \(\chi^2\) only moves from 26.74/6 to 25.74–26.21/6, \(p<3\times10^{-4}\); the near-repeat \(z\) goes 4.03 → 3.95) — because the failing pair 0109270301/0402561501 is at the same position in the sky (\(0.33'\) apart, replayed geometry), so the off-axis curve is nearly identical for its two observations. So this is a between-observation difference (about 24 per cent) with different SP (8.60 versus 9.96e-5) and QPB (1.02 versus 1.68e-4) states. The per-term decomposition (MSC-R85) shows the raw count differences are nearly equal in MOS1 and MOS2, and the MOS2 failure comes from subtracting larger QPB (+0.240e-6) and SP (+0.160e-6) differences; the same-CCD subset gives \(z=3.92\) and on common sky MOS2−MOS1 is only 1.22\(\sigma\). The cause is not identified; the next step separates QPB/SP/SWCX by band (MSC-R85b). open
- 目录法测到的 MOS 残余被红队降为"条件性"（rev 45–46）：RT-M124 指出 MSC-R81b 的 MOS 残差依赖访问散射模型——在它自己的 30 % 散射变体下 \(z=0.42\)–\(1.23\)，且没有任何单一 \(\beta\) 能同时归零两台相机的步长（需要的比值 5.74，实测 0.84–2.78），因此 R81b 的残余判为 Questioned。独立的孔径测光复核（MSC-R82）为未定：事件文件孔径与目录计数给同一条斜率（MOS1 \(\Delta\beta+0.0146\pm0.139\)、MOS2 \(+0.0066\pm0.107\)），每相机只有 5–7 个配对，既不能确认也不能排除 \(\beta\approx0.04\)–\(0.07\)。RT-M123b 另确认 \(\beta\approx0\) 在原生 RAWX/RAWY 坐标里成立，不是坐标/roll/基线假象，但其量级随权重变（约 40 %），年代下降项 Questioned。ProvisionalThe catalogue method's MOS residual is downgraded to conditional by a red team (rev 45–46): RT-M124 points out that MSC-R81b's MOS residual depends on the visit-scatter model — under its own 30 per cent scatter variant \(z=0.42\)–\(1.23\), and no single \(\beta\) zeroes both cameras' steps (the needed ratio is 5.74 against 0.84–2.78 measured), so R81b's residual is recorded as Questioned. The independent aperture-photometry re-check (MSC-R82) is inconclusive: event-file apertures and catalogue counting give the same slope (MOS1 \(\Delta\beta+0.0146\pm0.139\), MOS2 \(+0.0066\pm0.107\)), with only 5–7 pairs per camera, so it can neither confirm nor exclude \(\beta\approx0.04\)–\(0.07\). RT-M123b separately confirms \(\beta\approx0\) in native RAWX/RAWY coordinates, i.e. not a coordinate/roll/baseline artefact, but its magnitude is weight-dependent (about 40 per cent) and its epoch-decline terms are Questioned. Provisional
来源：Source: cognition-os/06_AGENT_LOG.md rev 40 行，the rev 40 row, cognition-os/08_RUNNING.md revision 40 头部revision 40 header
工作 / Work
Work
- 02:54 UTC（2026-10-20 22:54 EDT）派出手册刷新 3（Hermes，本任务，job
2026-10-20-handbook-refresh-3，基线 = 与线上逐字节相同的树）。02:54 UTC (2026-10-20 22:54 EDT) dispatched handbook refresh 3 (Hermes, this job,2026-10-20-handbook-refresh-3, base = a tree byte-identical to the live site). - 在飞（08 §3，rev 46）：MSC-R85b（近重复对的逐能段 QPB/SP/SWCX，sol）、MSC-R86（太阳风 SWCX 筛查，sol）、MSC-R21-9（fast_regions 到 READY，sol，PI 批准）、MSC-R59b（pn 考试产品，脱离式 Virgo 监视）、RT-M127（R70，codexnim）/RT-M128（R85，hermes）、MSC-S59/S60（luna）、手册刷新 3 阶段 4（Hermes，本任务）。In flight (08 §3, rev 46): MSC-R85b (per-band QPB/SP/SWCX for the near-repeat pair, sol), MSC-R86 (solar-wind SWCX screen, sol), MSC-R21-9 (fast_regions to READY, sol, PI-approved), MSC-R59b (pn exam products, detached Virgo monitor), RT-M127 (R70, codexnim) / RT-M128 (R85, hermes), MSC-S59/S60 (luna), and handbook refresh 3 stage 4 (Hermes, this job).
- 部署：本刷新不部署（Claude 复审后部署）。Deployed: none this day (the handbook is not deployed; Claude deploys after review).
来源：Source: cognition-os/08_RUNNING.md revision 46 §3，revision 46 §3, cognition-os/06_AGENT_LOG.md rev 40–46 各行the rev 40–46 rows
问题 / Problems
Problems
- 派发器自己的错误（已在记录中）：rev 40 派发刷新 3、MSC-R75b/R76、RT-M118/M119、S45/S46 时没有先做 lang 负载检查；当时 lang 因其他项目（S2Fitting、SR）占用而处于 81 % CPU / 71 % 内存。行动：新任务尽量改派给代理执行；在负载降到 60 % 以下之前不再派发。The dispatcher's own error (recorded): rev 40 launched refresh 3, MSC-R75b/R76, RT-M118/M119 and S45/S46 without first checking the lang load; lang was at 81 per cent CPU and 71 per cent memory from other projects (S2Fitting, SR). Action: new jobs were routed to agents instead, and no further launches until the load falls below 60 per cent.
- 技能演练 7 对手册技能的判定（codexnim，2026-10-20 22:55 EDT 评分）：兄弟项目 ParetoGTICal 的线上站 14 个页面里 0 个有密码门（用 curl 检查；DXC 是 12 个）；
deploy_live.sh期望hermes_site而 DXC 用hermes_build/site；"至少 10 % 内容变化"这道闸与 LOG 计数器都不是机械可检的；LOG 合并已逾期 3.3 倍。行动：修复项排队。Skill drill 7's verdict on the handbook skill (codexnim, graded 2026-10-20 22:55 EDT): the sibling project ParetoGTICal's live site has 0 of 14 pages gated (checked by curl; DXC has 12);deploy_live.shexpectshermes_sitewhile DXC useshermes_build/site; the "at least 10 per cent content change" gate and the LOG counter are not mechanically checkable; LOG consolidation is 3.3 times overdue. Action: fixes queued.
来源：Source: cognition-os/08_RUNNING.md revision 40 头部，revision 40 header, cognition-os/06_AGENT_LOG.md rev 40 行the rev 40 row
待 PI 决定 / For the PI
For the PI
- 无（rev 40 之后没有新的待 PI 决定项；D031 已采纳、评分推迟到外部 SP 门通过）。None (no new item awaits the PI after rev 40; D031 is adopted and scoring is deferred until the external-SP gate passes).
来源：Source: cognition-os/04_DECISION_LOG.md D031
DiffuseXrayCalibration：把 XMM 的弥漫辐射测量校准到可引用
DiffuseXrayCalibration: making XMM diffuse-emission measurements quotable
XMM-Newton 的 EPIC 相机看弥漫辐射时，隔着两种仪器自身的背景：准静态粒子背景（QPB）与软质子（SP）。本项目校准 ESAS 减数据 + Sherpa 拟合这条流水线，使同一天空区域在任何观测、任何离轴角、任何 SP/QPB 状态下给出同一个答案；做不到，就给出这个差异的测量值——系统误差预算。最显眼的症状是每个指向在拼接图边缘变亮（edge brightening）。为什么要做，见为什么存在一页。
XMM-Newton's EPIC cameras see diffuse emission through two backgrounds of the instrument's own making: the quiescent particle background (QPB) and soft protons (SP). This project calibrates the pipeline that removes them — ESAS reduction plus Sherpa fitting — so that one piece of sky gives the same answer in every observation, at every off-axis angle, and under every SP/QPB state; where it cannot, it delivers the measured difference itself: a systematic budget. The visible symptom is that every pointing brightens toward the edge of a mosaic (edge brightening). Why this matters is on the why the project exists page.
这一页回答四个问题：这是什么（本段）；现在到哪里了（目标表）；什么还没解决（未解决问题表）；下一步做什么（1–4 周计划与如何用手继续）。
This page answers four questions: what this is (this paragraph); where it stands (the objectives table); what is unsolved (the open-problems table); what to do next (the 1–4 week plan and how to continue by hand).
来源：Source: cognition-os/00_AI_BRIEFING.md, cognition-os/02_CURRENT_QUESTION.md
最新骨架：一条流水线，三个考题
Latest skeleton: one pipeline, three exam steps
来源：Source: cognition-os/02_CURRENT_QUESTION.md, MosaicSelfCal/BENCHMARK.md, cognition-os/tool_iteration.md
目标表：PI 的三步目标与基准测试目标
Objectives: the PI's three-step goal and the benchmark goals
| 目标Objective | 状态Status | 当前最佳数字Best current number | 趋势Trend |
|---|---|---|---|
| 第 1 步：抽谱（region + spectra）——高信噪比、低 SP、低未分辨点源/展源残差Step 1: regions and spectra — high S/N, low SP, low unresolved point/extended residual | 进行中in progress | CDFS P0：\(\tau/\bar{I}=0.246\)，\(\sigma/\bar{I}=0.245\)CDFS P0: \(\tau/\bar{I}=0.246\), \(\sigma/\bar{I}=0.245\) | → 掩膜规则已定（D012/D021），分数尚未收敛→ mask rule decided (D012/D021); score not yet converged |
| 第 2 步：拟合（fit sky 和 Src）——同一点指向内跨观测稳定；跨离轴角稳定Step 2: fit sky and Src — stable across observations of the same pointing, and across off-axis angle | 进行中in progress | M31 配对斜率跨 SP 分解横跨 \(-0.73\) 到 \(+0.91\) %/arcmin（仅 QPB \(+1.51\pm0.17\)；自由谱形 \(-0.73\pm0.29\)；时间切片颜色 \(+0.91\)；颜色先验 \(+0.01\pm0.26\)）；联合多环带（耀发模板）\(-0.161\pm0.262\)，与零一致M31 pair slope spans \(-0.73\) to \(+0.91\) per cent per arcmin across SP decompositions (QPB only \(+1.51\pm0.17\); free shape \(-0.73\pm0.29\); time-slice colour \(+0.91\); colour prior \(+0.01\pm0.26\)); joint multi-annulus with the flare template \(-0.161\pm0.262\), consistent with zero | ↔ 斜率由 SP 分解决定；重复指向在 SP 扣除后彼此不一致（R37，约 \(4\sigma\)）；SP 门仍不过：谱精度 \(\le\) 13 % 只有 6/31 台相机，角区先验的 6/9 是循环的（自由 QPB 下 1/9）↔ the slope is decided by the SP decomposition; repeat pointings disagree with each other after SP subtraction (R37, about \(4\sigma\)); the SP gate still fails: a spectral precision of \(\le\) 13 per cent in only 6 of 31 cameras, and the corner prior's 6 of 9 is circular (1 of 9 with QPB free) |
| 第 3 步：成像（image，去掉 QPB 与 SP）——同指向叠加一致；不同指向拼接看不到 vignettingStep 3: images with QPB and SP removed — same-pointing stacks agree; mosaics show no visible vignetting | 未解决open | M31 边缘步长（MOS2，0.4–1.25 keV，8–10′ 对 12–14′）。一个固定估计量上（像素集合、曝光分母、权重全部冻结）：测得 SP 分支 \(+3.29\) %，统计工作尺度 \(\pm4.34\) 个百分点；点源曲线另贡献 0.39 个百分点；指向间散布 \(\tau\approx10.2\) 个百分点，工作 95 % 预测区间 \(-23\) 到 \(+29\) %。不扣 SP 时（R34 校正表）\(+9.37\pm3.20\) % 同相机、\(+9.06\pm3.22\) % 曝光匹配；点源离轴校正把它降到 \(+3.28\pm4.31\) %（配对 \(-2.54\pm1.01\) 个百分点）The M31 edge step (MOS2, 0.4–1.25 keV, 8–10′ vs 12–14′). On one fixed estimand (pixel membership, exposure denominator and weights all frozen): the measured-SP branch is \(+3.29\) per cent with a statistical working scale of \(\pm4.34\) percentage points; the point-source curve adds 0.39 percentage points; between-pointing scatter \(\tau\approx10.2\) percentage points, working 95 per cent prediction interval \(-23\) to \(+29\) per cent. Unsubtracted (the R34 corrected table) \(+9.37\pm3.20\) per cent same-camera and \(+9.06\pm3.22\) per cent exposure-matched; the point-source off-axis correction brings it to \(+3.28\pm4.31\) per cent (paired \(-2.54\pm1.01\) percentage points) | ↔ 不扣 SP 的步长稳健；扣 SP 后的残差与零一致但指向间散布 \(\tau\approx10\) 个百分点占主导——报告"与零一致、散布主导"，不报单一数字（MSC-R72/R74）。点源项是施加点源响应曲线的效应，\(\textit{是否真实修正未定}\)（rev 39 的"曝光伪影"读法已撤回，rev 40 的"真信号"措辞也被撤回；RT-M118）。曝光图定标未定：预登记规则在当前样本量下不可能通过（RT-M119），且 \(\beta=0.10\) 的响应残差就把同相机步长移 \(-6.86\) 个百分点（MSC-R77）。归零所需 \(\beta\) 为 MOS1 \(0.21\pm0.24\)、MOS2 \(0.043\pm0.065\)（MSC-R80）；目录法实测 MOS \(+0.04\) 到 \(+0.07\)、同天空重叠 MOS2 \(+0.018\pm0.022\) 到 \(+0.040\pm0.011\)（MSC-R81b/R79，均 Provisional）→ 独立计数法复核中（MSC-R82）、盲预测已派（MSC-R83）且它以排除离轴曲线的方式失败：失败的那对 MOS2 是同一块天、同一个探测器位置（相距 \(0.33'\)），所以残差是观测之间的差异，SP/QPB 状态不同而原因未识别（MSC-R84/R85）↔ the unsubtracted step is robust; the SP-subtracted residual is consistent with zero but between-pointing scatter of \(\tau\approx10\) percentage points dominates — report "consistent with zero, scatter-dominated", not a single number (MSC-R72/R74). The point-source term is the effect of applying the point-source response curve, and \(\textit{whether it is a real correction is open}\) (the rev-39 "exposure artefact" reading is withdrawn and so is the rev-40 "real signal" wording; RT-M118). The exposure-map calibration is inconclusive: the pre-registered rule could not be passed at this sample size (RT-M119), and a response residual of \(\beta=0.10\) alone moves the same-camera step by \(-6.86\) percentage points (MSC-R77). The \(\beta\) needed to zero the step is MOS1 \(0.21\pm0.24\), MOS2 \(0.043\pm0.065\) (MSC-R80); the catalogue method measures MOS \(+0.04\) to \(+0.07\) and same-sky overlaps MOS2 \(+0.018\pm0.022\) to \(+0.040\pm0.011\) (MSC-R81b/R79, both Provisional) → an independent counting method is re-checking it (MSC-R82), the blind prediction is dispatched (MSC-R83) and it failed in a way that rules out the off-axis curve: the failing MOS2 pair is the same sky at the same detector position (\(0.33'\) apart), so the residual is a between-observation difference with different SP/QPB states and an unidentified cause (MSC-R84/R85) |
| 基准测试基线：同一固定天域跨 ObsID 的 \(\tau/\bar{I}\) 与离轴斜率（CDFS 与 M31）Benchmark baseline: between-ObsID \(\tau/\bar{I}\) and off-axis slope for one fixed sky cell (CDFS and M31) | 已建立established | CDFS P0：\(\tau/\bar{I}=0.246\)，斜率 \(+0.152/10'\)；M31 P0：\(\tau/\bar{I}=0.14\)，斜率 \(+0.25/10'\)CDFS P0: \(\tau/\bar{I}=0.246\), slope \(+0.152/10'\); M31 P0: \(\tau/\bar{I}=0.14\), slope \(+0.25/10'\) | → 这是待改进的基线，不是成果→ this is the baseline to improve, not a result |
| 考试集（exam v4 r240）：7 个 problem、18 个 ObsID、45 个 case 被减出来并审计Exam set (v4 r240): 7 problems, 18 ObsIDs, 45 cases reduced and audited | 完成done | 40 PASS，5 EXCLUDED，0 FAIL40 PASS, 5 EXCLUDED, 0 FAIL | → 冻结集（D020），后续分数都在它上面算→ frozen set (D020); every later score is computed on it |
| 快速抽谱（R21，PI 2026-10-19 11:23 EDT 的新目标）：对任意区域快速抽出谱与响应（RMF、ARF、QPB）Fast extraction (R21, the PI's new goal of 2026-10-19 11:23 EDT): spectra and responses (RMF, ARF, QPB) for arbitrary regions, fast | 进行中in progress | 热缓存 0.16 s/区域（Virgo 上 100 个 MOS2 区域 16.0 s）；冷/热输出一致；原生 prepare 的 ARF+RMF 每相机约 930–990 swarm cache 0.16 s per region on Virgo (100 MOS2 regions in 16.0 s); outputs identical cold/warm; native prepare ARF+RMF ~930–990 s per camera | → R21-2 对齐已过（BACKSCAL 6/6 在 0.1 % 内，ARF 6/6 在 1 % 内）；R21-6 提速完成，R21-7 持久缓存在跑→ R21-2 parity passed (BACKSCAL 6/6 within 0.1 %, ARF 6/6 within 1 %); R21-6 speed-up done, R21-7 persistent cache running |
来源：Source: MosaicSelfCal/BENCHMARK.md, cognition-os/02_CURRENT_QUESTION.md, MosaicSelfCal/results/cdfs_pipeline_comparison_v1.csv, MosaicSelfCal/results/exam_v4_r240/SUMMARY.md, cognition-os/tool_iteration.md 趋势箭头：→ 未收敛；↔ 两个估计矛盾；"持平"= 配对区间含 0。第 3 行的三个数字是三个不同的估计量，不可混用：固定估计量步长（MSC-R74）、不扣 SP 的 R34 校正表、以及施加点源离轴校正后的 R34 分支（MSC-R60）。Trend arrows: → not converged; ↔ two estimates disagree; "No significant difference" = the paired interval contains 0. The three numbers in the step-3 row are three different estimands and must not be mixed: the fixed-estimand step (MSC-R74), the unsubtracted R34 corrected table, and the R34 branch after the point-source off-axis correction (MSC-R60).
未解决的问题
Open problems
| 问题Problem | 为何重要Why it matters | 当前最佳假设Best hypothesis | 下一个检验Next test | 负责人Owner | 状态Status |
|---|---|---|---|---|---|
| 边缘增亮（edge brightening）到底是 SP、QPB，还是仪器响应？Is edge brightening soft protons, the QPB, or the instrument response? | 这是 mosaics 里最显眼的失败，也是 PI 三步目标的第 3 步It is the most visible failure in mosaics and step 3 of the PI goal | 步长稳健、归因未定，且数据不足以定归因：不扣 SP 时 \(+9.37\pm3.20\) %（同相机）跨相机一致（R34）。在一个固定估计量上（像素集合、曝光分母、权重全部冻结），测得 SP 分支是 \(+3.29\) %、统计工作尺度 \(\pm4.34\) 个百分点，但九个指向本身散布 \(\tau\approx10.2\) 个百分点，工作 95 % 预测区间 \(-23\) 到 \(+29\) %——报告"与零一致、散布主导"，不报单一数字（MSC-R72/R74）。点源离轴曲线把步长削掉约 40 %（配对 \(-2.54\pm1.01\) 个百分点）而不消除相机分歧；点源项（\(-2.52\) 个百分点）是施加点源响应曲线的效应，\(\textit{是否真实修正未定}\)（rev 39 的"曝光伪影"读法与 rev 40 的"真信号"措辞都已撤回；RT-M118）。能段签名不偏向 SP/QPB/相机项中任何一个（R45）。SP 归一化无法从外部定住（谱 \(\le\) 13 % 只有 6/31 台相机；角区先验的 6/9 是循环的，自由 QPB 下 1/9），重复指向彼此不一致（R37），QPB 模板修正不充分（R42）——归因是数据受限的，不是分析疏漏。rev 45–46 又排除了一个候选：离轴响应残差被盲预测正面检验排除（失败的那对 MOS2 是同一位置、相距 \(0.33'\)，离轴曲线对它两个观测几乎相同；MSC-R84），逐项分解把差异定位到 QPB 与 SP 的观测间差而原因未识别（MSC-R85）。目录法测到的 MOS 残余降为条件性（RT-M124），孔径测光既不能确认也不能排除它（MSC-R82）The step is robust, its attribution is not, and the data cannot pin it: unsubtracted, \(+9.37\pm3.20\) per cent (same-camera), consistent across cameras (R34). On one fixed estimand (pixel membership, exposure denominator and weights all frozen) the measured-SP branch is \(+3.29\) per cent with a statistical working scale of \(\pm4.34\) percentage points, but the nine pointings themselves scatter by \(\tau\approx10.2\) percentage points, with a working 95 per cent prediction interval of \(-23\) to \(+29\) per cent — report "consistent with zero, scatter-dominated", not a single number (MSC-R72/R74). The point-source off-axis curve removes about 40 per cent of the step (paired \(-2.54\pm1.01\) percentage points) and does not remove the camera disagreement; the point-source term (\(-2.52\) percentage points) is the effect of applying the point-source response curve, and \(\textit{whether it is a real correction is open}\) (both the rev-39 "exposure artefact" reading and the rev-40 "real" wording are withdrawn; RT-M118). The energy signature favours none of SP, QPB or the camera terms (R45). The SP normalisation cannot be pinned externally (a spectral precision of \(\le\) 13 per cent in only 6 of 31 cameras; the corner prior's 6 of 9 is circular, 1 of 9 with QPB free), repeat pointings disagree (R37), and the QPB-template fix is not sufficient (R42) — the attribution is data-limited, not an analysis slip. Rev 45–46 removes one more candidate: the off-axis response residual is excluded by a positive test — the failing MOS2 pair is the same position (\(0.33'\) apart), so the off-axis curve is nearly identical for its two observations (MSC-R84), and the per-term decomposition locates the difference in between-observation QPB and SP differences with the cause not identified (MSC-R85). The catalogue method's MOS residual is downgraded to conditional (RT-M124), and aperture photometry can neither confirm nor exclude it (MSC-R82) | D031（已采纳的归因规则）规定：外部 SP 约束不到位，任何归因只能判为"未识别"。下一个真正的检验是把 SP 门做出来：带一致的 SP 谱形（MSC-R71）、角区 QPB 与 SP 的联合拟合（MSC-R73，已因循环先验被红队修正为 1/9）。同时 MSC-R75b 用档案里的重复源（每单元 \(\ge\) 50 个）重做曝光图离轴定标，MSC-R76 检查曝光图是否被错误地应用到 SP/QPB 模板上。rev 45–46 的新事实：盲预测以排除离轴曲线的方式失败（失败的那对 MOS2 是同一位置、相距 \(0.33'\)，见上），所以离轴响应这条解释已被一个正面检验排除；逐项分解把差异定位到 QPB 与 SP 的观测间差（原因未识别），下一步按能段分离（MSC-R85b）并用太阳风通量筛查 SWCX（MSC-R86）D031 (the adopted attribution rule) prescribes that without an external SP constraint any attribution can only be judged "unidentified". The next real test is therefore to get the SP gate to pass: a band-consistent SP spectral shape (MSC-R71), and the joint corner-QPB and SP fit (MSC-R73, corrected by the red team to 1 of 9 because of the circular prior). In parallel MSC-R75b redoes the exposure-map off-axis calibration with archive repeat sources (at least 50 per cell) and MSC-R76 checks whether the X-ray exposure map is being applied to the SP/QPB templates. The new fact of rev 45–46: the blind prediction failed in a way that rules out the off-axis curve (the failing MOS2 pair is the same position, \(0.33'\) apart — see above), so the off-axis response is now excluded by a positive test; the per-term decomposition locates the difference in the between-observation QPB and SP differences (cause not identified), and the next step separates them by band (MSC-R85b) and screens SWCX with the solar-wind flux (MSC-R86) | MSC-R71 / R73 / R75b / R76 / R85b / R86 | 运行中running |
| SP 在探测器上的空间分布是什么形状？What is the spatial distribution of SP across the detector? | 拟合用的模型 \( \text{rsp\_flat}(A_{\text{sky}}\cdot\text{SP}) \) 默认 SP 在探测器上是均匀的；如果不是，所有 SP 扣除都带系统误差The fit model \( \text{rsp\_flat}(A_{\text{sky}}\cdot\text{SP}) \) assumes SP is uniform across the detector; if it is not, every SP subtraction carries a systematic | 已测（R24）：从耀发时段直接测得的 M31 软带 SP 空间模板与 ESAS 耀发图一致；R20y 联合多环带斜率 \(+0.011\pm0.261\) %/arcmin（名义 \(-0.161\)），与零一致。闭合证据：R24 的 31/31 安静 PHA 重放精确一致，MOS2−MOS1 超出名义预测 \(+0.023\pm0.014\)（1.7\(\sigma\)）Measured (R24): the M31 soft-band SP spatial template taken directly from flare intervals matches the ESAS flare map; the R20y joint multi-annulus slope is \(+0.011\pm0.261\) per cent per arcmin (nominal \(-0.161\)), consistent with zero. Closing evidence: R24's 31/31 quiet-PHA replay is exact, and MOS2−MOS1 exceeds the nominal prediction by \(+0.023\pm0.014\) (1.7\(\sigma\)) | 已闭合：软带模板已由 R24 从耀发时段直接测得。剩下的问题不再是形状，而是每观测的归一化（见上一行）Closed: the soft-band template was measured directly from flare intervals by R24. What remains is not the shape but the per-observation normalisation (previous row) | R24 | 已闭合closed |
| MOS1/MOS2 的 SP 归一比在 2012 年发生台阶（\(0.68\to1.47\)），原因是什么？The MOS1/MOS2 SP-normalisation ratio steps in 2012 (\(0.68\to1.47\)); what causes it? | 任何跨 2012-12 合并 SP 先验或比较都是错的；这也决定了 SP 扣除能否跨年代使用Any SP prior or comparison pooled across 2012-12 is wrong, and it decides whether an SP subtraction can be used across epochs | 没有单一原因站得住：面积效应被排除；掩膜足迹、角区模板陈旧、硬带 QPB 标度都被逐一排除；自由 SP 谱形下看不到 2012 台阶（R20s：1.61\(\sigma\)）No single cause stands: the area effect is refuted; the mask footprint, stale corner templates and the hard-band QPB scale are excluded one by one; with free SP shapes no 2012 step is seen (R20s: 1.61\(\sigma\)) | R20w 已答（4 个 M31 候选里只有 0690600401 是可用的定标源；时间可变 SP 颜色 0.4–1.25/2–7 keV MOS1 0.564、MOS2 0.401）；软带径向模板仍分不开。剩下的检验在 SP 门里：带一致的 SP 谱形（MSC-R71）——固定形状已被数据否决（\(\Delta\) 统计量约 \(-530\)）R20w is answered (of 4 M31 candidates only 0690600401 is a usable calibrator; the time-variable SP colour 0.4–1.25/2–7 keV is 0.564 for MOS1 and 0.401 for MOS2); the soft radial templates remain indistinguishable. The remaining test lives in the SP gate: a band-consistent SP spectral shape (MSC-R71) — the fixed shape is rejected by the data (a \(\Delta\) statistic of about \(-530\)) | MSC-R71 | 检验中testing |
| 重复指向能否测出 SP 的绝对水平？Can repeat pointings measure the absolute SP level? | SP 水平是第 3 步成像的关键输入；测不出来就无法给 SP 扣除定标The SP level is the key input to step 3; without it the SP subtraction cannot be normalised | 不能，而且重复对彼此不一致（R37）：0109270301/0402561501 的 MOS2 在 SP 扣除后从 \(+7.9\pm3.8\) % 变到 \(+23.8\pm5.8\) %（约 \(4\sigma\)）；近重复对给出的绝对 SP 精度 \(\ge\) 200 %——比需要的 4–13 % 差一个量级以上No, and repeat pairs disagree with each other (R37): the MOS2 pair 0109270301/0402561501 moves from \(+7.9\pm3.8\) to \(+23.8\pm5.8\) per cent after SP subtraction (about \(4\sigma\)); near-repeats give an absolute SP precision of \(\ge\) 200 per cent — an order of magnitude worse than the 4–13 per cent needed | R36 的结论已把这条路标为数据受限：1 个百分点的步长需要约 140–170 个科学观测（i.i.d. 外推），且系统底不随 \(\sqrt{N}\) 下降；改走外部定标（D028 的外部 SP 约束门）R36 has marked this route data-limited: a 1-percentage-point step needs about 140–170 science observations (i.i.d. extrapolation), and the systematic floor does not fall as \(\sqrt{N}\); the route moves to external calibration (D028's external-SP gate) | D028 （草稿）(draft) | 未解决open |
| 叠加（stacking）能否把边缘误差平均掉？Can stacking average out the edge error? | CGM 叠加是本项目下游的主要用途；若共模误差消不掉，所有叠加晕的径向轮廓都带 \(\pm\)4 % 的系统Stacking CGM observations is the main downstream use; if a common-mode error survives, every stacked halo profile carries a \(\pm\)4 per cent systematic | 不能（R38）：随机滚动平均不掉共模径向误差——对齐叠加在 12–14′ 处保留 \(\pm\)4 % 天空，任意 N；8′ 偏移 + N=200 只把边缘-only 偏差降到 +0.7 %，但线性项给 \(-1.4\pm1.3\) %No (R38): random roll does not average out a common-mode radial error — an aligned stack keeps \(\pm\)4 per cent sky at 12–14′ for any N; an 8′ offset with N=200 reduces the edge-only bias to +0.7 per cent but a linear term gives \(-1.4\pm1.3\) per cent | 已到结论：没有方案能把全体天空降到 1 % 以内；叠加前必须先修边缘，或把 \(\pm\)4 % 写进误差预算Concluded: no scheme brings the whole sky within 1 per cent; fix the edge before stacking, or carry the \(\pm\)4 per cent in the budget | R38 | 已回答answered |
| QPB 空间模板在 10–14′ 是否被低估？Is the QPB spatial template underestimated at 10–14′? | 这是 R39 提出的非 SP 机制候选：那里 QPB 约占计数的 35–36 %，模板低估 5.8–9.1 % 就会产生 +4–6 个百分点的边缘The non-SP candidate raised by R39: QPB is about 35–36 per cent of counts there, and a 5.8–9.1 per cent template underestimate would produce a +4–6 percentage point edge | 已到结论（R42/R44/R48 + RT-M101/RT-M106）：预先声明的充分性规则不满足——全 FWC/径向替换给出非正净项，SP 扣除后的 MOS1/同相机/曝光匹配未定义。R44 清点出 SAS 20 的 ESAS QPB 模板用的是 209 条 SOC FWC 记录里的 128 条、完全没有 2012-12-11 之后的 E3 数据；但把 E3 数据补进去只把 M31 步长移 \(-0.94\) 个百分点（约 \(1.6\sigma\)），而"晚期年代实质性过时"的读法被红队撤回（10 次触发都解析不了：7/10 在 WCDFS，LOO 摆动 35–66 个百分点）。年代效应 \(\le\) 1 个百分点，不是主因Concluded (R42/R44/R48 + RT-M101/RT-M106): the pre-stated sufficiency rule is not met — the full FWC/radial replacement gives non-positive net terms, leaving SP-subtracted MOS1/samecam/exposure-matched undefined. R44's inventory shows the SAS 20 ESAS QPB template uses 128 of the 209 SOC FWC entries and no E3 (post-2012-12-11) data at all; but adding the E3 data moves the M31 step by only \(-0.94\) percentage points (about \(1.6\sigma\)), and the "materially stale at late epochs" reading was withdrawn by the red team (the 10 trips are not resolvable: 7 of 10 in WCDFS, leave-one-out swings of 35–66 percentage points). The epoch effect is \(\le\) 1 percentage point, not the main cause | 闭合证据：R42（充分性规则不满足）+ R48（E3 只移 \(-0.94\) 个百分点）+ RT-M106（"过时"读法撤回）+ MSC-R53（原始旧库重建与生产不闭合，0.009 个百分点）。QPB 线关闭，精力转向 SP 门与曝光图定标Closing evidence: R42 (sufficiency rule not met) + R48 (E3 moves only \(-0.94\) percentage points) + RT-M106 (the "stale" reading withdrawn) + MSC-R53 (the raw-old rebuild does not close on production, 0.009 percentage points). The QPB line is closed; effort moves to the SP gate and the exposure-map calibration | R42 / R48 | 已闭合closed |
| 点源掩膜该掩多深？（PI 2026-10-20 18:29 EDT 的要求：不要什么都掩，用定标选阈值）How deep should the point-source mask go? (the PI's request of 2026-10-20 18:29 EDT: do not mask everything, choose the threshold by calibration) | 掩膜深度本身就能造出一条边缘：掩得越深，外环比内环少掩的部分就变成一条假边缘The mask depth alone can create an edge: the deeper the mask, the part that is masked less in the outer ring becomes a spurious edge | 考试掩膜本来就是半径均匀的通量切割 \(F>5\times10^{-15}\)（半径 30/45/90″），不是 DET_ML。把更暗的源也掩上（F > 1e-15）会把 M31 步长从 \(+5.82\) 抬到 \(+10.05\) %（\(+4.2\pm2.6\) 个百分点），因为暗源在 12–14′ 的密度比 8–10′ 低 44 %——这是离轴目录不完备，不是天空。规范 5e-15 掩膜在 8–10′ 环去掉 8.9 %、在 12–14′ 环去掉 18.7 %The exam masks are already a radius-uniform flux cut \(F>5\times10^{-15}\) (radii 30/45/90″), not DET_ML. Masking fainter sources too (F > 1e-15) raises the M31 step from \(+5.82\) to \(+10.05\) per cent (\(+4.2\pm2.6\) percentage points) because faint sources are 44 per cent less dense at 12–14′ than at 8–10′ — off-axis catalogue incompleteness, not sky. The canonical 5e-15 mask removes 8.9 per cent of the 8–10′ ring and 18.7 per cent of the 12–14′ ring | MSC-R70：5e-15 处的掩膜完备度对离轴角的依赖、完整通量/DET_ML 阶梯加随机掩膜对照、用 logN-logS 推未分辨源的面亮度；给出边缘偏差最小的掩膜规则MSC-R70: mask completeness versus off-axis angle at 5e-15, the full flux/DET_ML ladder with random-mask controls, unresolved-source surface brightness from logN-logS; recommend the mask rule with the least edge bias | MSC-R70 | 运行中running |
| pn 相机的 QPB 为什么被高估？Why is the pn camera's QPB over-normalised? | pn 目前只能做交叉检查（T01）；修不好就不能升级为主要相机pn is a cross-check only (T01); until it is fixed it cannot be promoted | 不是角区光子泄漏（OOT 只占 \(2.5\ \%\)，边缘 \(0.5\ \%\)）；更像 FWC 模板迁移，最佳修正是 FoV 12-14 keV 归一化Not corner photon leakage (OOT \(2.5\ \%\), edge \(0.5\ \%\)); likelier an FWC template transfer, best corrected by the FoV 12-14 keV normalisation | R5b：pn 减数据 + pn/MOS 联合拟合，held-out 检验 pn QPB 标度R5b: pn reduction plus joint pn/MOS fits, with a held-out test of the pn QPB scale | R5b | 检验中testing |
| 暗天区里 SP 与天空的简并如何打破？How is the SP-sky degeneracy broken in faint fields? | WCDFS 是最暗的检验场；在那里谱拟合给出的 SP 归一不可靠，成像又过度扣除WCDFS is the faintest testbed; there the spectral SP norm is unreliable and the image over-subtracts | 分组方式本身就是一个系统误差（G1 把 2-12 keV 打进一个 bin，天空差 \(18\ \%\)）；用 G2 式边界The grouping choice is itself a systematic (G1 packs 2-12 keV into one bin, moving the sky by \(18\ \%\)); use G2-style boundaries | R3c：联合似然比检验（天空共享 vs 每观测自由，SP 每观测自由）R3c: joint likelihood-ratio test (sky shared vs free per observation, SP free per observation) | R3c | 部分完成partly done |
| SAS 22.1 后端能否用于生产？Can the SAS 22.1 backend be used for production? | SAS 20 与 22.1 的 ARF 差 \(+7\) 到 \(+22\ \%\)；用错后端会让所有天空水平偏移The SAS 20 and 22.1 ARFs differ by \(+7\) to \(+22\ \%\); the wrong backend shifts every sky level | 差异可归因：ARF 来自 arfgen 开关，QPB 来自 FWC 来源 + MOS1 自建角区；但 pn 端到端与可复现性声明还没过The differences are attributable (ARF from arfgen switches, QPB from the FWC source plus MOS1's own corner), but pn end-to-end and the reproducibility statement have not passed | R15 / esas v1.8.0：pn 端到端、ANOMFL、可复现性声明，然后 PI 签字R15 / esas v1.8.0: pn end to end, ANOMFL, a reproducibility statement, then PI sign-off | R15 | 检验中testing |
来源：Source: cognition-os/REDTEAM_QUEUE.md, MosaicSelfCal/RESEARCH_PLAN.md, cognition-os/08_RUNNING.md
未来 1–4 周
The next 1–4 weeks
计划顺序照 MosaicSelfCal/RESEARCH_PLAN.md 的 SP 路线图（PI 2026-10-17）与 cognition-os/08_RUNNING.md §7 的建议方向。每条后面是它要回答的问题。
The order follows the SP roadmap in MosaicSelfCal/RESEARCH_PLAN.md (PI 2026-10-17) and the proposed directions in cognition-os/08_RUNNING.md §7. Each item names the question it answers.
| 时间When | 做什么What | 回答什么问题Question answered | 状态State |
|---|---|---|---|
| 已完成（本时段）done (this window) | R44 FWC 库存；R45 能段签名；R46 图像分解；R47 软带锚；R48 E3 QPB；R50 软带失配诊断；R21-7 持久缓存；MSC-R72 误差预算；MSC-R74 固定估计量；MSC-R75 曝光图离轴定标（未过自预登记规则）；MSC-R79 同天空弥漫重叠；MSC-R80 归零所需 \(\beta\)；MSC-R81/R81b 目录曝光图；MSC-R82 孔径测光复核（未定）；MSC-R83 盲预测（失败）；MSC-R84 离轴曲线不闭合；MSC-R85 近重复对分解（原因未识别）；MSC-R70 掩膜完备度；红队 RT-M108–RT-M126 全部三分派；文献 S29–S58R44 FWC inventory; R45 energy signature; R46 image decomposition; R47 softest-band anchor; R48 E3 QPB; R50 softest-band misfit diagnosis; R21-7 persistent cache; MSC-R72 error budget; MSC-R74 fixed estimand; MSC-R75 exposure-map off-axis calibration (did not pass its pre-registered rule); MSC-R79 same-sky diffuse overlaps; MSC-R80 the \(\beta\) needed to zero the step; MSC-R81/R81b catalogue exposure map; MSC-R82 aperture-photometry re-check (inconclusive); MSC-R83 blind prediction (failed); MSC-R84 the off-axis curve does not close R83; MSC-R85 near-repeat pair decomposition (cause not identified); MSC-R70 mask completeness; red teams RT-M108–RT-M126 all triaged; literature S29–S58 | QPB 线可以关闭了（年代效应 \(\le\) 1 个百分点）；能段签名不偏向任何一个候选；SP 门仍不过；曝光图定标修好 MOS1 却过度修正 MOS2；rev 45–46：盲预测以排除离轴曲线的方式失败，逐项分解把差异定位到 QPB 与 SP 的观测间差而原因未识别The QPB line can be closed (epoch effect \(\le\) 1 percentage point); the energy signature favours no candidate; the SP gate still fails; the exposure-map calibration fixes MOS1 but over-corrects MOS2; rev 45–46: the blind prediction failed in a way that rules out the off-axis curve, and the per-term decomposition locates the difference in between-observation QPB and SP differences with the cause not identified | 已完成done |
| 现在进行in flight | MSC-R85b（近重复对的逐能段 QPB/SP/SWCX，sol）；MSC-R86（太阳风 SWCX 筛查，sol）；MSC-R21-9（fast_regions 到 READY，sol，PI 批准）；MSC-R59b（pn 考试产品，脱离式 Virgo 监视）；RT-M127 复核 R70（codexnim）；RT-M128 复核 R85（hermes）；MSC-S59/S60（luna）MSC-R85b (per-band QPB/SP/SWCX for the near-repeat pair, sol); MSC-R86 (solar-wind SWCX screen, sol); MSC-R21-9 (fast_regions to READY, sol, PI-approved); MSC-R59b (pn exam products, detached Virgo monitor); RT-M127 re-checking R70 (codexnim); RT-M128 re-checking R85 (hermes); MSC-S59/S60 (luna) | 是什么让同一块天在两条观测之间差 24 %（QPB、SP、SWCX，还是 MOS2 探测器状态）；点源掩膜该掩多深；pn 能否成为第二台主相机What makes the same sky differ by 24 per cent between two observations (QPB, SP, SWCX, or a MOS2 detector state); how deep the point-source mask should go; whether pn can become a second primary camera | 运行中running |
| 接下来next | D031 已采纳（= D028 v2，协议已封存、sha256 以 73902f30 开头）。评分推迟，直到外部 SP 门通过——现在它是失败的（谱 \(\le\) 13 % 只有 6/31 台相机；角区先验的 6/9 是循环的，自由 QPB 下 1/9）。因此下一步是把 SP 门做出来：MSC-R71 的带一致 SP 谱形、MSC-R73 的角区 QPB 联合拟合（按红队修正后的 1/9 计）。在门通过之前不看分数D031 is adopted (= D028 v2, protocol sealed, sha256 starting 73902f30). Scoring is deferred until the external SP gate passes — it currently fails (a spectral precision of \(\le\) 13 per cent in only 6 of 31 cameras; the corner prior's 6 of 9 is circular, 1 of 9 with QPB free). So the next step is to make the SP gate pass: MSC-R71's band-consistent SP spectral shape and MSC-R73's joint corner-QPB fit (counted at the red team's corrected 1 of 9). No score is looked at before the gate passes | 在没有外部 SP 约束的情况下，归因只能判为"未识别"；规则先行的价值就在此Without an external SP constraint the attribution can only be judged "unidentified"; that is the value of the rule-first approach | 待定pending |
| 并行in parallel | R21 快速抽谱家族（R21-2/R21-4/R21-6/R21-7 与 MSC-R21-8 已完成，全部 NOT READY：held-out r015 QPB +0.106 %、r084 ARF 1.07 % 仍未过，精确 FWC 41 s/区域）；R5b pn 减数据与联合拟合；R15 / esas v1.8.0 READY 门the R21 fast-extraction family (R21-2/R21-4/R21-6/R21-7 and MSC-R21-8 are done, all NOT READY: held-out r015 QPB +0.106 per cent and r084 ARF 1.07 per cent still fail, exact-FWC 41 s per region); R5b pn reduction and joint fits; the R15 / esas v1.8.0 READY gate | 任意区域快速抽谱（PI 当前目标）；pn 能否成为第二台主相机；SAS 22.1 能否用于生产fast spectra for arbitrary regions (the PI's current goal); can pn become a second primary camera; can SAS 22.1 be used for production | 运行中running |
| 2–4 周2–4 weeks | 工具发布：edgestep 包、xsherpa SP 响应、esas SP 成像模板；然后 DXC 用固定版本重测Tool releases: the edgestep package, the xsherpa SP response, the ESAS SP image template; then a DXC re-test pinned to the version | 修正能否落到工具里并被独立复现Whether the fixes land in the tools and are independently reproduced | 待定pending |
| 验收标准acceptance | 高 SP / 低 SP 时间桶给同一面亮度；M31/LMC/WCDFS 配对斜率归零；mosaic 看不到 vignettingHigh- and low-SP time bins of one pointing give the same surface brightness; M31/LMC/WCDFS pair slopes zero; mosaics show no visible vignetting | PI 三步目标是否达成Whether the PI's three-step goal is met | 未达成not met |
来源：Source: MosaicSelfCal/RESEARCH_PLAN.md, cognition-os/08_RUNNING.md, cognition-os/tool_iteration.md
本次更新 / 数据截至
What's new / Data as of
本次更新What's new
2026-10-21 12:19 UTC（2026-10-21 08:19 EDT）：刷新 3 阶段 2b（H-DXC-3）。把内容推进到 cognition-os/08_RUNNING.md revision 46（2026-10-21 08:06 EDT），并补上阶段 2 遗漏的物理页图。要点：
2026-10-21 12:19 UTC (2026-10-21 08:19 EDT): refresh 3 stage 2b (H-DXC-3).The content is brought up to cognition-os/08_RUNNING.md revision 46 (2026-10-21 08:06 EDT), and the physics-page figure that stage 2 missed is added. The essentials:
- 盲预测失败，且失败的方式排除了离轴响应（MSC-R83/R84/R85）：冻结流程的盲预测里 M31 近重复对 0109270301/0402561501 的 MOS2 差 \(z=4.03\)（联合 \(\chi^2=26.74/6\)，\(p=1.6\times10^{-4}\)），MOS1 同对一致（\(z=0.10\)）。把 MOS 离轴残余曲线乘回去不闭合（\(\chi^2\) 26.74/6 → 25.74–26.21/6；近重复 \(z\) 4.03 → 3.95）——因为该对相距仅 \(0.33'\)、同一探测器位置，离轴曲线对它两个观测几乎相同。逐项分解：原始计数差在两台 MOS 相机里几乎相等，MOS2 的失败来自减掉更大的 QPB（+0.240e-6）与 SP（+0.160e-6）差；同一 CCD 子集 \(z=3.92\)，共同天空上 MOS2−MOS1 只 1.22\(\sigma\)。原因未识别 → MSC-R85b 按能段分离，MSC-R86 用太阳风通量筛查 SWCX
- 目录法 MOS 残余降为条件性；孔径测光未定（RT-M124，MSC-R82，RT-M123b）：RT-M124 显示 MSC-R81b 的 MOS 残差依赖访问散射模型（30 % 散射变体下 \(z=0.42\)–\(1.23\)），且没有单一 \(\beta\) 能同时归零两台相机的步长（所需比值 5.74，实测 0.84–2.78）→ R81b 残余 Questioned。MSC-R82 未定：事件文件孔径与目录计数给同一斜率（MOS1 \(\Delta\beta+0.0146\pm0.139\)、MOS2 \(+0.0066\pm0.107\)），每相机仅 5–7 对，不能确认或排除 \(\beta\approx0.04\)–\(0.07\)。RT-M123b：\(\beta\approx0\) 在原生 RAWX/RAWY 成立，非坐标/roll/基线假象，量级随权重变（约 40 %），年代下降项 Questioned
- D032 主机次序（PI）：数据处理只在 Virgo；谱拟合按 lang2 → Virgo → lang 的次序动态负载，Virgo 上拟合让位于抽谱（拟合在 nice 19 下运行，抽谱等待时按 PID 暂停）。这取代 D023 的"Virgo 是默认拟合环境"与 T02 的"从不在 Virgo 上拟合"
- 点源掩膜完备度（MSC-R70）：F>5e-15 在 12–14′ 不完备（代理功效 65.9 % M31、29.1 % WCDFS、18.3 % LMC）；M31 步长 6.19 \(\pm\) 3.97 %（5e-15）对 9.98 \(\pm\) 4.90 %（1e-15），配对 +3.79 \(\pm\) 2.33 个百分点；目录−随机对照 3.17 \(\pm\) 7.19 个百分点；仅完备度偏差未识别 → RT-M127
- fast_regions 至 READY（MSC-R21-9，PI 批准）：r015 QPB / r084 ARF 平价原因、精确面积 FWC、全部 100 区域的原生响应平价，然后是手册原则页（PI）
- The blind prediction failed, and the way it failed rules out the off-axis response (MSC-R83/R84/R85): in the frozen-pipeline blind prediction the M31 near-repeat pair 0109270301/0402561501 differs in MOS2 at \(z=4.03\) (joint \(\chi^2=26.74/6\), \(p=1.6\times10^{-4}\)), while MOS1 of the same pair agrees (\(z=0.10\)). Multiplying the MOS off-axis residual curve back in does not close it (\(\chi^2\) 26.74/6 → 25.74–26.21/6; the near-repeat \(z\) 4.03 → 3.95) — because the pair is only \(0.33'\) apart at the same detector position, so the off-axis curve is nearly identical for its two observations. The per-term decomposition: the raw count differences are nearly equal in both MOS cameras, and the MOS2 failure comes from subtracting larger QPB (+0.240e-6) and SP (+0.160e-6) differences; the same-CCD subset gives \(z=3.92\), and on common sky MOS2−MOS1 is only 1.22\(\sigma\). The cause is not identified → MSC-R85b separates by band and MSC-R86 screens SWCX with the solar-wind flux
- The catalogue MOS residual is downgraded to conditional; aperture photometry is inconclusive (RT-M124, MSC-R82, RT-M123b): RT-M124 shows MSC-R81b's MOS residual depends on the visit-scatter model (under its 30 per cent scatter variant \(z=0.42\)–\(1.23\)), and no single \(\beta\) zeroes both cameras' steps (the needed ratio is 5.74 against 0.84–2.78 measured) → R81b's residual is Questioned. MSC-R82 is inconclusive: event-file apertures and catalogue counting give the same slope (MOS1 \(\Delta\beta+0.0146\pm0.139\), MOS2 \(+0.0066\pm0.107\)), with only 5–7 pairs per camera, so it can neither confirm nor exclude \(\beta\approx0.04\)–\(0.07\). RT-M123b: \(\beta\approx0\) holds in native RAWX/RAWY, not a coordinate/roll/baseline artefact; its magnitude is weight-dependent (about 40 per cent) and its epoch-decline terms are Questioned
- The D032 host order (PI): data processing only on Virgo; spectral fits go to the first host with headroom, lang2 → Virgo → lang, with dynamic load; fits on Virgo yield to extraction (they run under nice 19 and are paused by PID when extraction waits). This supersedes D023's "Virgo is the default fit environment" and T02's "never fit on Virgo"
- Point-source mask completeness (MSC-R70): F>5e-15 is not complete at 12–14′ (surrogate power 65.9 % M31, 29.1 % WCDFS, 18.3 % LMC); the M31 step is 6.19 \(\pm\) 3.97 % (5e-15) versus 9.98 \(\pm\) 4.90 % (1e-15), paired +3.79 \(\pm\) 2.33 percentage points; the catalogue-minus-random contrast is 3.17 \(\pm\) 7.19 percentage points; the completeness-only bias is not identified → RT-M127
- fast_regions to READY (MSC-R21-9, PI-approved): the r015 QPB / r084 ARF parity causes, exact-area FWC, native-response parity on all 100 regions, then the handbook principle page (PI)
2026-10-21 03:2x UTC（2026-10-20 23:2x EDT）：刷新 3（H-DXC-3）。把内容推进到 cognition-os/08_RUNNING.md revision 40（2026-10-20 22:55 EDT），新增 2026-10-21 的组会段，并把 revision 29–40 补录进 2026-10-20 的组会段。要点：
2026-10-21 03:2x UTC (2026-10-20 23:2x EDT): refresh 3 (H-DXC-3).The content is brought up to cognition-os/08_RUNNING.md revision 40 (2026-10-20 22:55 EDT); a 2026-10-21 group-meeting block is added and revisions 29–40 are appended to the 2026-10-20 block. The essentials:
- 点源离轴曲线削掉约 40 % 的步长；这一项的含义仍开放：8–10′ 对 12–14′、0.4–1.25 keV 的点源比 MOS1/pn \(+1.067\pm0.767\) %、MOS2/pn \(+3.913\pm0.774\) %；按每相机用到原环带后，测得 SP 分支同相机 \(+5.818\to+3.281\pm4.308\) %（配对 \(-2.537\pm1.009\) 个百分点）。把曝光分母完全冻结后该项仍是 \(-2.52\) 个百分点；它是施加点源响应曲线的效应，是否真实修正取决于该曲线是否真实的响应误差——rev 39 的"曝光伪影"读法与 rev 40 的"真信号"措辞都已撤回（MSC-R58/R60/R74，RT-M118）
- 一个固定估计量上的步长：与零一致，指向间散布主导：像素集合、曝光分母、权重全部冻结后 \(+3.29\) %、统计 \(\pm4.34\) 个百分点；九个指向散布 \(\tau=10.163\) 个百分点（\(I^2=61.5\) %），工作 95 % 预测区间 \(-23\) 到 \(+29\) %；单删任一指向均值都保持正号（\(+1.05\) 到 \(+4.73\) %），没有单个指向决定符号（MSC-R74，临时；RT-M118 复核中）
- 曝光图离轴定标降级为"未定"：协议先封存再跑；联合接受规则只在有 13–15 个重复源的晚期单元失败（heldout z 1.08/1.71），但红队检查（RT-M119）指出该规则在当前样本量下本来就不可能通过（零假设下预期 2.8 个单元失败），弥漫臂对 \(\beta=0.20\) 的移动（\(-7.3\) 个百分点）小于其 \(\sigma\approx7.6\)——所以"vignetting 残差不是主因"不是判决。功效重放显示步长对响应残差非常敏感：\(\beta=0.10\) 就把同相机步长移 \(-6.86\) 个百分点（MSC-R77）。描述性曲线把 MOS1 \(+13.8\to+4.2\) % 但把 MOS2 \(+3.0\to-6.3\) %——修好一个、过度修正另一个（MSC-R75，PI 批准）
- SP 水平仍不可认证；角区先验的通过数是循环的：谱精度 \(\le\) 13 % 只有 6/31 台相机、\(\le\) 4 % 是 0/31，固定 SP 谱形被数据否决（\(\Delta\) 统计量约 \(-530\)）；放开 SP 斜率让全/硬带归一一致（19/31 在 1\(\sigma\) 内）但精度仍 0/31。角区 QPB 先验下 6/9 通过，红队证明这是 q=1 循环先验的产物：QPB 自由时 1/9，加上漏掉的转移误差 3/9。D031 SP 门按 1/9 计（MSC-R65/R68/R71/R73，RT-M112/M113/M116）
- 点源掩膜定标（PI 要求）：考试掩膜本来就是半径均匀通量切割 \(F>5\times10^{-15}\)，不是 DET_ML；掩更暗的源（F > 1e-15）把步长抬 \(+4.2\pm2.6\) 个百分点，因为暗源在 12–14′ 密度低 44 %（离轴目录不完备）。规范掩膜在内环去 8.9 %、外环去 18.7 %（MSC-R69/R70，RT-M117）
- QPB 线关闭：ESAS QPB 模板完全没有 2012-12-11 之后的 FWC 数据（209 条里用 128 条），但补上 E3 只移 \(-0.94\) 个百分点（约 \(1.6\sigma\)），"晚期过时"读法被红队撤回；年代效应 \(\le\) 1 个百分点（R44/R48/MSC-R53，RT-M106）
- 文献：官方确认 MOS/pn 在约 12′ 处有最高 20 % 的通量差：CAL-TN-0018（2025-02，逐字核对）说 vignetting 定标调查仍在进行，无 0.3–1.5 keV 单独界限、无 MOS1/MOS2 拆分；无已发布修正；IACHEC 2026 列 MOS2 污染物累积（0.5 keV 约 \(-20\) %）；无已发布的每观测 SP 归一或 8′ 外 MOS1/MOS2 弥漫比（S29–S44）
- 本站自查：线上组会存档页 62 处公式从未渲染：上一轮留下的双反斜杠定界符不被 KaTeX 认识，公式在浏览器里是字面文本；已修正并用无头 Chromium 复核（14 个渲染公式、0 个错误）
- The point-source off-axis curve removes about 40 per cent of the step; what that means is still open: at 8–10′ vs 12–14′, 0.4–1.25 keV, the point-source ratio is \(+1.067\pm0.767\) per cent for MOS1/pn and \(+3.913\pm0.774\) per cent for MOS2/pn; applied per camera to the original annuli, the measured-SP same-camera step moves \(+5.818\to+3.281\pm4.308\) per cent (paired \(-2.537\pm1.009\) percentage points). With the exposure denominator fully frozen the term is still \(-2.52\) percentage points; it is the effect of applying the point-source response curve, and whether it is a real correction depends on whether that curve is a real response error — both the rev-39 "exposure artefact" reading and the rev-40 "real signal" wording are withdrawn (MSC-R58/R60/R74, RT-M118)
- The step on one fixed estimand: consistent with zero, between-pointing scatter dominates: with pixel membership, exposure denominator and weights all frozen, \(+3.29\) per cent, statistical \(\pm4.34\) percentage points; the nine pointings scatter by \(\tau=10.163\) percentage points (\(I^2=61.5\) per cent), working 95 per cent prediction interval \(-23\) to \(+29\) per cent; deleting any single pointing keeps the mean positive (\(+1.05\) to \(+4.73\) per cent), so no single pointing sets the sign (MSC-R74, provisional; RT-M118 re-checking)
- The exposure-map off-axis calibration is downgraded to inconclusive: the protocol was sealed before the run; the joint acceptance rule fails only in the late cells with 13–15 repeat sources (heldout z 1.08/1.71), but a red-team check (RT-M119) shows the rule could not have been passed at this sample size (2.8 cell fails expected under the null) and the diffuse arm's move at \(\beta=0.20\) (\(-7.3\) percentage points) is smaller than its \(\sigma\approx7.6\) — so "vignetting residual not the main cause" is not a verdict. The power replay shows the step is very sensitive to a response residual: \(\beta=0.10\) alone moves the same-camera step by \(-6.86\) percentage points (MSC-R77). The descriptive curve moves MOS1 \(+13.8\to+4.2\) per cent but MOS2 \(+3.0\to-6.3\) per cent — it fixes one and over-corrects the other (MSC-R75, PI-approved)
- The SP level is still not certifiable; the corner prior's pass count is circular: a spectral precision of \(\le\) 13 per cent in only 6 of 31 cameras and \(\le\) 4 per cent in 0 of 31, with the fixed SP shape rejected by the data (a \(\Delta\) statistic of about \(-530\)); freeing the SP slopes makes the full- and hard-band norms agree (19 of 31 within 1\(\sigma\)) but the precision still reaches 0 of 31. With a corner QPB prior 6 of 9 pass, and the red team shows this is a product of the circular q=1 prior: 1 of 9 with QPB free, 3 of 9 with the omitted transfer error added. The D031 SP gate counts 1 of 9 (MSC-R65/R68/R71/R73, RT-M112/M113/M116)
- Point-source mask calibration (the PI's request): the exam masks are already a radius-uniform flux cut \(F>5\times10^{-15}\), not DET_ML; masking fainter sources (F > 1e-15) raises the step by \(+4.2\pm2.6\) percentage points because faint sources are 44 per cent less dense at 12–14′ (off-axis catalogue incompleteness). The canonical mask removes 8.9 per cent of the inner ring and 18.7 per cent of the outer (MSC-R69/R70, RT-M117)
- The QPB line is closed: the ESAS QPB template contains no FWC data after 2012-12-11 (128 of 209 entries used), but adding E3 moves the step by only \(-0.94\) percentage points (about \(1.6\sigma\)) and the "stale at late epochs" reading was withdrawn by the red team; the epoch effect is \(\le\) 1 percentage point (R44/R48/MSC-R53, RT-M106)
- Literature: the SOC confirms MOS/pn flux differences up to 20 per cent at about 12′: CAL-TN-0018 (2025-02, verified verbatim) says the vignetting-calibration investigation is ongoing, with no 0.3–1.5 keV-only bound and no MOS1/MOS2 split; no released correction; IACHEC 2026 lists MOS2 contaminant build-up (about \(-20\) per cent at 0.5 keV); no published per-observation SP normalisation and no published MOS1/MOS2 diffuse ratio beyond 8′ (S29–S44)
- Site self-check: 62 formulas on the live meeting-archive page never rendered: doubled-backslash delimiters left by the previous refresh are not recognised by KaTeX, so the formulas showed as literal text in the browser; fixed and re-verified in headless Chromium (14 rendered formulas, 0 errors)
2026-10-20 09:25 EDT：刷新 2（H-DXC-2）。把内容推进到 cognition-os/08_RUNNING.md revision 28（2026-10-20 08:43 EDT），并新增 2026-10-20 的组会段。要点：
2026-10-20 09:25 EDT: refresh 2 (H-DXC-2).The content is brought up to cognition-os/08_RUNNING.md revision 28 (2026-10-20 08:43 EDT), and a 2026-10-20 group-meeting block is added. The essentials:
- M31 边缘步长有了校正表（R34，已验证）：0.4–1.25 keV、内部 8–10′ 对外部 12–14′、MOS2——不扣软质子 \(+9.37\pm3.20\) %（同相机）、\(+9.06\pm3.22\) %（曝光匹配）；名义 SP 扣除后 \(+6.28\pm4.51\) / \(+4.74\pm3.99\)；测得模板后 \(+5.82\pm4.34\) / \(+4.38\pm3.94\)。旧的 \(+8.9\pm2.4\) / \(+9.3\pm2.4\)（RT-M88）与 \(-1.66\)/\(+1.62\)（R29）废弃
- SP 归一化无法从外部定住（R32/R33/R36）：没有任何路线能给到每观测 4–13 % 的可认证精度；按 i.i.d. 外推，1 个百分点的步长需要约 140–170 个科学观测，且系统底不随 \(\sqrt{N}\) 下降
- 重复指向彼此不一致（R37）：0109270301/0402561501 的 MOS2 在 SP 扣除后从 \(+7.9\pm3.8\) % 变到 \(+23.8\pm5.8\) %（约 \(4\sigma\)）；近重复对给出的绝对 SP 精度 \(\ge\) 200 %
- 叠加保留共模边缘（R38）：随机滚动平均不掉共模径向误差——对齐叠加在 12–14′ 处保留 \(\pm\)4 % 天空，任意 N；没有方案能把全体天空降到 1 % 以内
- QPB 外模板假设被检验，不充分（R39/R42）：预先声明的规则（区间含 0 且修正误差 \(<\)2 个百分点）不满足；只剩年代-FWC 漂移转移（同相机 \(+5.08\pm4.35\) %）
- 官方 ESAS 耀发图 CCF 里有负像素（R43）：每行 MOS FLAREMAP 有 111k–289k 个负像素（共 608,400）；截断到 0 只改变步长 \(+0.17\) / \(+0.23\) 个百分点
- 目录默认 DR14（D029）：4XMM-DR14 按观测逐个使用，DR14s 仅在点名理由时；edetect 每观测省 39–86 分钟（中位 60）；34 项测试通过
- 快速抽谱提速（R21-6）：热缓存 0.16 s/区域（Virgo 上 100 个 MOS2 区域 16.0 s），冷/热输出一致；R21-7 持久缓存在跑；专家顾问判定三步目标未达成
- The M31 edge step has a corrected table (R34, verified): 0.4–1.25 keV, inner 8–10′ vs outer 12–14′, MOS2 — unsubtracted \(+9.37\pm3.20\) per cent (same-camera), \(+9.06\pm3.22\) per cent (exposure-matched); after nominal SP \(+6.28\pm4.51\) / \(+4.74\pm3.99\); after the measured template \(+5.82\pm4.34\) / \(+4.38\pm3.94\). The old \(+8.9\pm2.4\) / \(+9.3\pm2.4\) (RT-M88) and \(-1.66\)/\(+1.62\) (R29) are deprecated
- The SP normalisation cannot be pinned externally (R32/R33/R36): no route gives a certifiable 4–13 per cent per-observation precision; by i.i.d. extrapolation a 1-percentage-point step needs about 140–170 science observations, and the systematic floor does not fall as \(\sqrt{N}\)
- Repeat pointings disagree with each other (R37): the MOS2 pair 0109270301/0402561501 moves from \(+7.9\pm3.8\) to \(+23.8\pm5.8\) per cent after SP subtraction (about \(4\sigma\)); near-repeats give an absolute SP precision of \(\ge\) 200 per cent
- Stacks keep the common-mode edge (R38): random roll does not average out a common-mode radial error — an aligned stack keeps \(\pm\)4 per cent sky at 12–14′ for any N; no scheme brings the whole sky within 1 per cent
- The QPB outer-template lead was tested and is not sufficient (R39/R42): the pre-stated rule (interval contains 0 and correction error \(<\)2 percentage points) is not met; only the epoch-FWC drift transfer remains (same-camera \(+5.08\pm4.35\) per cent)
- The official ESAS flare-map CCF has negative pixels (R43): every MOS FLAREMAP row has 111k–289k negative pixels of 608,400; clipping at 0 changes the step by only \(+0.17\) / \(+0.23\) percentage points
- Catalogue default DR14 (D029): 4XMM-DR14 per observation by default, DR14s only with a named reason; edetect saves 39–86 minutes (median 60) per ObsID; 34 tests pass
- Fast extraction speed-up (R21-6): warm cache 0.16 s per region (100 MOS2 regions in 16.0 s on Virgo), outputs identical cold/warm; R21-7 persistent cache running; the expert advisor judged the three-step goal not reached
本页所有数字取自下方列出的记录文件；没有一项是本手册自己算出来的。
Every number on this page is read from the record files listed below; none is computed by this handbook.
数据截至Data as of
项目状态：2026-10-21 08:06 EDT（cognition-os/08_RUNNING.md revision 46）。
Project state: 2026-10-21 08:06 EDT (cognition-os/08_RUNNING.md revision 46).
本手册读到的最新科学记录：2026-10-21 08:06 EDT（rev 46：盲预测失败——M31 近重复对的 MOS2 差 \(z=4.03\)，而该对相距仅 \(0.33'\)、同一探测器位置，所以离轴曲线不能解释它（MSC-R84）；逐项分解把差异定位到 QPB（+0.240e-6）与 SP（+0.160e-6）的观测间差，原始计数差在两台 MOS 里几乎相等，原因未识别（MSC-R85）→ MSC-R85b 按能段分离；目录法 MOS 残余降为条件性（RT-M124），孔径测光未定（MSC-R82）；点源掩膜 F>5e-15 在 12–14′ 不完备但仅完备度偏差未识别（MSC-R70 → RT-M127）；D032 主机次序：拟合 lang2 → Virgo → lang，抽谱优先）。
The newest science record this handbook read: 2026-10-21 08:06 EDT (rev 46: the blind prediction failed — the M31 near-repeat pair differs in MOS2 at \(z=4.03\), and the pair is only \(0.33'\) apart at the same detector position, so the off-axis curve cannot explain it (MSC-R84); the per-term decomposition locates the difference in the between-observation QPB (+0.240e-6) and SP (+0.160e-6) differences, with the raw count differences nearly equal in both MOS cameras and the cause not identified (MSC-R85) → MSC-R85b separates by band; the catalogue MOS residual is downgraded to conditional (RT-M124) and aperture photometry is inconclusive (MSC-R82); the point-source mask F>5e-15 is incomplete at 12–14′ but the completeness-only bias is not identified (MSC-R70 → RT-M127); the D032 host order: fits go lang2 → Virgo → lang, extraction first).
更晚的记录（评审、决定、新结果）以它们为准；本页在下次刷新时更新。
Later records (reviews, decisions, new results) take precedence; this page is updated at the next refresh.