ParetoGTICal 工作台ParetoGTICal workbench

四个问题Four questions · 目标表Objectives · 开放问题表Open problems · 计划Plan · 骨架图Skeleton · 动画Animation · 本次更新What's new · 数据截至Data as of

这是什么项目? XMM-Newton 望远镜每次观测都会被软质子 (soft proton) 耀斑污染，必须用"好时间段 (good time interval, GTI)"把耀斑时间删掉。删多删少是一对矛盾：多删 → 剩下的数据更干净但更少；少删 → 数据多但更脏。ParetoGTICal 把这个取舍当作一个帕累托问题 (Pareto problem):在单次观测内部，用保留活时间 f_T 与连续谱约减因子 R = C_all/C_K 两个坐标给每个候选 GTI 打分，并用 Al-K、Si-K 荧光线的不变性做门 (gate),防止"挑噪声"伪装成"删耀斑"。它属于父项目 DiffuseXrayCalibration （校准 XMM EPIC 弥散发射分析），但有自己的问题、自己的认知操作系统 (cognition OS) 和自己的预注册 (pre-registration) 纪律。

What is this project? Every XMM-Newton observation is polluted by soft-proton flares, and "bad" time must be removed by keeping only good time intervals (GTIs). How much to cut is a genuine conflict: cut more and the retained data are cleaner but fewer; cut less and there are more data but dirtier ones. ParetoGTICal treats this trade-off as a Pareto problem: within one observation, every candidate GTI is scored by two coordinates — retained live time f_T and the continuum reduction factor R = C_all/C_K — with an Al-K / Si-K fluorescence-line stability gate that stops "selecting noise" from masquerading as "cutting flares". It is a sub-project of DiffuseXrayCalibration (calibrating the XMM EPIC diffuse-emission analysis), with its own question, its own cognition OS and its own pre-registration discipline.

四个问题，一屏作答Four questions, answered on one screen

现在在哪？
Where does it stand?: P001 已关闭（D006），主要问题上的答案是否定的：在等保留率下，率排名在任何主集上都不比标准 2.filter 更好 —— M31 MOS1 （13/25,Z = 0.02)、M31 MOS2 参考（21/29,Z = 1.64)、CDFS MOS1 （24/29,Z = 1.55)、CDFS MOS2 （27/29,Z = 1.73)。CDFS 上有方向一致但极小的增益（约 0.3%,描述性，按 D7 规则记"持平"）。角落 (T017) 按数据集分别回答：在 CDFS 两台相机上够准（σ = 0.0044 / 0.0038),在 M31 MOS1 上太吵（σ = 0.013)。; P001 is closed (D006) with a negative answer on its primary question: at equal retention, rate-rank does not beat the standard 2.filter on any primary set — M31 MOS1 (13/25, Z = 0.02), M31 MOS2 reference (21/29, Z = 1.64), CDFS MOS1 (24/29, Z = 1.55), CDFS MOS2 (27/29, Z = 1.73). On CDFS there is a consistent but tiny gain (about 0.3% in R; descriptive, "no significant difference" under rule D7). The corner (T017) is answered per set: precise enough on both CDFS cameras (σ = 0.0044 / 0.0038), too noisy on M31 MOS1 (σ = 0.013).
还有什么没解决？
What is unsolved?: ① 下游成功判据（D002 第 3 步：曝光损失 ⇄ 连续谱增益的兑换率，经父项目的响应折叠拟合）—— E3 失败后，这是让任何 GTI 政策改变变得值得的唯一途径；② 耦合/Al 问题（T005、T016),被 CDFS 的探索性发现变得更尖锐：被删时间的角落数子水平高 2–4%,而门内 Al 线比并不跟着走。; ① The downstream success criterion (D002 step 3: the exchange rate between exposure loss and continuum gain, through the parent's response-folded fit) — after E3 failed, this is the only route by which any change of GTI policy could be justified; ② the coupling/Al question (T005, T016), sharpened by the CDFS exploratory finding: rejected time has a 2–4% higher corner particle level, yet the as-gated Al-line ratio does not follow it.
接下来做什么？
What do I do next?: 两条路由 PI 选择，本手册不代选:(a) 下游成功判据（D002 第 3 步）;（b) 就"Al 线是否跟随角落粒子水平"立新预注册（未经评审）。步骤格式见"如何接手"页 §5。; The PI chooses between two routes — this handbook does not choose: (a) the downstream success criterion (D002 step 3); (b) a new pre-registration on whether the Al line tracks the corner particle level (not reviewed). Step formats on the "By hand" page §5.

目标表（PI 工作单，D002,2026-10-13）Objectives (the PI's work order, D002, 2026-10-13)

#	目标Objective	状态Status	最好的当前数字（来源）Best current number (source)	趋势Trend
1	事实纠错认知 OSFact-correct the cognition OS	DONE	D002 + I002：训练/保留集（19 训练、11 保留）2026-09-14 已冻结且已消费；两个度量族 R 与 G_Al；数据是 M31 MOS2 不是 CDFS。D002 + I002: the train/holdout split (19 training, 11 holdout) was frozen and scored on 2026-09-14; two metric families R and G_Al exist; the data are M31 MOS2, not CDFS. 来源：Source: `ParetoGTICal/cognition-os/04_DECISION_LOG.md` (D002); `insights/I002_historical_evidence_reconciliation.md`	→
2	扩展到 MOS1 + 空白场对照Extend to MOS1 + blank-field control	DONE	M31 MOS1 盲测（E002）与 CDFS 空白场盲测（E003,34/34 ObsID × 2 相机）均完成。E3 在所有主集上失败；CDFS 的增益方向一致但只有约 0.3%（描述性）。角落：CDFS 可用，M31 MOS1 太吵。Both halves done: the blind M31 MOS1 run (E002) and the CDFS blank-field blind run (E003, 34/34 ObsIDs × 2 cameras). E3 fails on every primary set; the CDFS gain is consistent in sign but only about 0.3% (descriptive). Corner: usable on CDFS, too noisy on M31 MOS1. `results/p001_m31_mos1/p001_endpoints_m31.json`; `results/p001_cdfs/p001_endpoints_all.json`; `cognition-os/experiments/E002_p001_m31_mos1.md`; `E003_p001_cdfs_blank_field.md`	→
3	定义下游成功判据Define the downstream success criterion	待定PENDING	尚无数字。它要回答：1% 的活时间值多少连续谱约减？由父项目的响应折叠弥散天空拟合给出。E3 失败后，这个问题变成"选哪个保留率",而非"选哪个时间"。No number yet. It must answer: how much continuum reduction is 1% of live time worth? Given by the parent project's response-folded diffuse-sky fit. After E3 failed, this becomes "which retention to choose", not "which time to keep". `ParetoGTICal/cognition-os/00_AI_BRIEFING.md` §Next steps (D002 step 3); `04_DECISION_LOG.md` D005	→
4	开放问题：未曝光角落有没有用？Open question: is the unexposed corner useful?	已按数据集回答（D006）ANSWERED per set (D006)	精度（叠加 σ）:CDFS MOS1 = 0.0044、MOS2 = 0.0038 → USABLE;M31 MOS2 参考 = 0.0098 → 勉强可用；M31 MOS1 = 0.0133 → TOO_NOISY。它还在 CDFS 上看到了真实的 K/B 粒子级差：被删时间的角落 2.5–8 keV 率高 1.9%（MOS1,−4.3σ）与 3.9%（MOS2,−10σ）。新的开放子问题（未经预注册）：角落的粒子水平变化不带入门内 Al 线比。Precision (stacked σ): CDFS MOS1 = 0.0044, MOS2 = 0.0038 → USABLE; M31 MOS2 reference = 0.0098 → usable by a hair; M31 MOS1 = 0.0133 → TOO_NOISY. It also sees a real K/B particle-level difference on CDFS: rejected time has a 1.9% (MOS1, −4.3σ) and 3.9% (MOS2, −10σ) higher corner 2.5–8 keV rate. New open sub-question (not pre-registered): the corner's particle-level change does not carry into the as-gated Al-line ratio. `results/p001_cdfs/p001_endpoints_all.json` (E5); `03_TRUST_TABLE.md` T017; `E003_p001_cdfs_blank_field.md` §Reading 3§Reading 3	→

开放问题表Open problems

问题Problem	为什么重要Why it matters	当前最好的假设Best current hypothesis	下一个检验Next test	所有者 / 状态Owner / status
R 筛选在等保留率下是否优于归档 2.filter?Does the R screen beat the archived 2.filter at equal retention?	这是 P001 的主要终点；若不优于，方法的全部价值就只剩"选保留率"这一层。This was P001's primary endpoint; if not, the method's whole value reduces to "choosing a retention".	不优于（已关闭）：四个主集的 E3 全部失败 —— M31 MOS1 13/25（Z = 0.02）、M31 MOS2 21/29（Z = 1.64）、CDFS MOS1 24/29（Z = 1.55）、CDFS MOS2 27/29（Z = 1.73）。CDFS 上符号一致但增益约 0.3%，远低于 Z > 3;按 D7 记"持平"。Not better (closed): E3 fails on all four primary sets — M31 MOS1 13/25 (Z = 0.02), M31 MOS2 21/29 (Z = 1.64), CDFS MOS1 24/29 (Z = 1.55), CDFS MOS2 27/29 (Z = 1.73). On CDFS the sign is consistent but the gain is about 0.3%, far below Z > 3; "no significant difference" under D7.	已由 CDFS 盲测完成（E003）：证据收据 `p001_endpoints_all.json`。问题关闭。Done by the CDFS blind run (E003); receipt `p001_endpoints_all.json`. The question is closed.	T019 / 在 M31 与 CDFS 上被驳倒REFUTED on M31 and CDFST019 / REFUTED on M31 and CDFS
角落的粒子水平变化为什么不带入门内 Al 线比？（探索性，未预注册）Why does the corner's particle-level change not carry into the as-gated Al-line ratio? (exploratory, not pre-registered)	若 Al 线不跟随 2.5–8 keV 角落连续谱，则线门（L_K = L_B 这条设计级约束）可能有未建模的 K/B 系统误差；这直接决定线门是否可信（T005/T016）。If the Al line does not track the 2.5–8 keV corner continuum, the line gate (the design-level constraint L_K = L_B) may carry an unmodelled K/B systematic; this decides whether the line gate is trustworthy (T005/T016).	两个竞争解释：Al 线本来就不跟随角落连续谱（粒子谱形随率变化），或线测量里有未建模的 K/B 系统误差。CDFS 上的差距：门内 +0.006 / +0.013 对角落隐含的 −0.02 / −0.04,相隔 3.2σ（MOS1）与 7.7σ（MOS2，忽略两个叠加之间的相关）。Two competing readings: the Al line genuinely does not track the corner continuum (the particle spectral shape changes with rate), or the line measurement has an unmodelled K/B systematic. The CDFS gap: as-gated +0.006 / +0.013 vs the corner-implied −0.02 / −0.04, i.e. 3.2σ (MOS1) and 7.7σ (MOS2) apart, ignoring correlations between the two stacks.	建议（未经评审）：立一份新预注册，在 CDFS 上直接检验"Al 线是否跟随角落粒子水平"。在此之前，任何数字都不算证据。Suggestion (not reviewed): a new pre-registration testing directly whether the Al line tracks the corner particle level on CDFS. Until then, no number from this is evidence.	T005, T016, T017 / 探索性（未预注册）EXPLORATORY (not pre-registered)T005, T016, T017 / EXPLORATORY (not pre-registered)
反事实 Al 亏损（−0.028 ± 0.013, MOS2）是真信号吗？Is the counterfactual Al deficit (−0.028 ± 0.013, MOS2) real?	若真，则 L_K = L_B 这个设计级约束本身有偏差；若假，则是未建模系统误差。它决定线门是否可信。If real, the design-level constraint L_K = L_B is itself biased; if not, it is an unmodelled systematic. It decides whether the line gate is trustworthy.	CDFS 上门内线比为 +0.0057 ± 0.0062（MOS1）与 +0.0129 ± 0.0055（MOS2）—— 方向与 M31 MOS2 的门内亏损相反；反事实耦合修正后为 −0.0223 ± 0.0065（MOS1）与 −0.0153 ± 0.0058（MOS2）。MOS1 与角落隐含值相符，MOS2 仍差 3.4σ。问题按 T005 仍未决，且与上面的角落张力相连。On CDFS the as-gated line ratio is +0.0057 ± 0.0062 (MOS1) and +0.0129 ± 0.0055 (MOS2) — the opposite sign to M31 MOS2's as-gated deficit; after the counterfactual coupling correction it is −0.0223 ± 0.0065 (MOS1) and −0.0153 ± 0.0058 (MOS2). MOS1 agrees with the corner-implied value; MOS2 remains 3.4σ short. Open per T005, and tied to the corner tension above.	新的预注册（检验 Al 线是否跟随角落粒子水平）；在此之前不关闭。A new pre-registration (testing whether the Al line tracks the corner particle level); not closed before that.	T005, T016 / QUESTIONEDT005, T016 / QUESTIONED
角落能否预测 K/B 粒子级变化？Can the corner predict the K/B particle-level change?	角落在 FOV 外、不经望远镜聚焦，是最干净的粒子监测器；若够准，就能直接检验"Al 亏损 = 粒子变化"。The corner is outside the FOV and receives no focused soft protons — the cleanest particle monitor; if precise enough it can test "Al deficit = particle change" directly.	已回答（按数据集，D006）：CDFS 上够准（σ = 0.0044 / 0.0038）且看到了真实的 K/B 粒子级差（被删时间高 1.9% / 3.9%）;M31 MOS1 上太吵（σ = 0.0133）。所以 2.filter 也删掉了粒子背景升高的时段，不只是软质子时段。Answered (per set, D006): precise enough on CDFS (σ = 0.0044 / 0.0038) and it sees a real K/B particle-level difference (rejected time 1.9% / 3.9% higher); too noisy on M31 MOS1 (σ = 0.0133). So 2.filter also cuts periods of elevated particle background, not only soft-proton time.	已由 CDFS E5 完成。遗留的子问题（未预注册）：Al 线不跟随角落 —— 见上面新行。Done by the CDFS E5. Remaining sub-question (not pre-registered): the Al line does not follow the corner — see the new row above.	T017 / 已按数据集回答ANSWERED per setT017 / ANSWERED per set
被移除的成分是软质子还是可变天体源？Is the removed component soft protons or variable celestial sources?	这是"删掉的东西到底是什么"的身份问题；M31 盘里有很多可变 X 射线双星，是天体源解释的真实竞争者。This is the identity question — what was actually removed; the M31 disk holds many variable X-ray binaries, a genuine celestial competitor.	A1:移除的超额 ~79% 无线 (Al −5.96σ);B1:空间延展、非紧凑，更像软质子花样；但仍不能排除多个微弱可变源。A1: the removed excess is ~79% line-free (Al −5.96σ); B1: extended, not compact — closer to the soft-proton pattern; many faint variable sources are not excluded.	CDFS E6（描述性）：CDFS 的中位增益（+0.0035 / +0.0029）比 M31（+0.0005 / +0.0011）大 3–10 倍，但同样远低于阈值 —— 与"移除成分主要是软质子、M31 的可变源压低了增益"相容，不构成证明。The CDFS E6 (descriptive): the median CDFS gain (+0.0035 / +0.0029) is 3–10× M31's (+0.0005 / +0.0011) but still far below threshold — compatible with "the removed component is mostly soft protons, with M31's variable sources diluting the gain", not proof of it.	T009, T018 / NOT ESTABLISHEDT009, T018 / NOT ESTABLISHED

来源：ParetoGTICal/cognition-os/03_TRUST_TABLE.md (T005, T008, T009, T016–T019, T021);04_DECISION_LOG.md (D002, D005, D006);E002_p001_m31_mos1.md;E003_p001_cdfs_blank_field.md。Source: ParetoGTICal/cognition-os/03_TRUST_TABLE.md (T005, T008, T009, T016–T019, T021); 04_DECISION_LOG.md (D002, D005, D006); E002_p001_m31_mos1.md; E003_p001_cdfs_blank_field.md.

ParetoGTICal skeleton: event list → light curve → candidate GTIs → K/B → R and line gate → null → P001 endpoints → missing downstream criterion — 最新骨架。虚线框是全项目唯一缺失的下游环节：用父项目的响应折叠 (response-folded) 弥散天空拟合定义"曝光损失 ⇄ 连续谱增益"的兑换率（D002 第 3 步）。 The latest skeleton. The dashed box is the one missing downstream piece: defining the exchange rate between exposure loss and continuum gain through the parent project's response-folded diffuse-sky fit (D002 step 3).

未来 1–4 周计划（PI 顺序， briefing "Next steps"）Plan for the next 1–4 weeks (PI order, briefing "Next steps")

CDFS 盲测已完成（2026-10-13,E003）：34/34 ObsID × 2 相机，E1–E6 已评估并记入 D006。P001 关闭。The CDFS blind run is done (2026-10-13, E003): 34/34 ObsIDs × 2 cameras, E1–E6 evaluated and recorded in D006. P001 is closed.
定义下游成功判据（D002 第 3 步）：在父项目的响应折叠拟合里，量化 1% 活时间损失对应多少连续谱增益，把 f_T–R 平面变成一个可决策的标量。E3 失败后，这是让任何 GTI 政策改变值得的唯一途径。完整步骤格式见"如何接手"页第 5 节。Define the downstream success criterion (D002 step 3): in the parent project's response-folded fit, quantify how much continuum gain 1% of live time buys, turning the f_T–R plane into a decidable scalar. After E3 failed this is the only route to justifying any change of GTI policy. Full step format on the "By hand" page, §5.
可能的（建议，未经评审）新预注册：Al 线是否跟随角落粒子水平 —— 由 PI 决定是否立项；在此之前它是探索性观察，不算证据。完整步骤格式见"如何接手"页第 5 节。A possible (suggestion, not reviewed) new pre-registration: whether the Al line tracks the corner particle level — the PI decides whether to open it; until then it is an exploratory observation, not evidence. Full step format on the "By hand" page, §5.

来源：ParetoGTICal/cognition-os/00_AI_BRIEFING.md §Next steps;02_CURRENT_QUESTION.md "Status (D006)";04_DECISION_LOG.md D006。Source: ParetoGTICal/cognition-os/00_AI_BRIEFING.md §Next steps; 02_CURRENT_QUESTION.md "Status (D006)"; 04_DECISION_LOG.md D006.

一分钟看懂核心矛盾（动画）The core conflict in one minute (animation)

光变曲线 + 阈值滑块 → GTI、f_T、保留计数实时更新Light curve + threshold slider → GTI, f_T and retained counts update live

拖动阈值：蓝色是保留的时间 K,红色是被删的 B。f_T 是保留活时间比例；R 用玩具模型模拟"删高率时间为什么会让 R>1"。

Drag the threshold: blue is kept time K, red is rejected B. f_T is the retained live-time fraction; R uses a toy model to show why cutting high-rate time gives R > 1.

阈值Threshold 2.6 count/s

本次更新What's new

2026-10-13（第二次更新，D006 刷新）:

2026-10-13 (second edition, the D006 refresh):

P001 已关闭（D006），主要问题上的答案是否定的：等保留率下率排名在任何主集上都不优于 2.filter。P001 is closed (D006) with a negative answer on its primary question: rate-rank at equal retention does not beat 2.filter on any primary set.
CDFS 盲测完成：34/34 ObsID × 2 相机（E003）;E1 在主集上 NOT_EVALUABLE（无规则零控制），敏感性集上 PASS（0/9）。The CDFS blind run is complete: 34/34 ObsIDs × 2 cameras (E003); E1 NOT_EVALUABLE on the primary set (no rule zero control), PASS on the sensitivity set (0/9).
E3 在 CDFS 两台相机上失败：MOS1 24/29（符号 p = 2.7e-4,Stouffer Z = 1.55）;MOS2 27/29（符号 p = 8.1e-7,Stouffer Z = 1.73）。预注册要求 Z > 3。E3 fails on both CDFS cameras: MOS1 24/29 (sign p = 2.7e-4, Stouffer Z = 1.55); MOS2 27/29 (sign p = 8.1e-7, Stouffer Z = 1.73). The pre-registered bar is Z > 3.
CDFS 增益方向一致但只有约 0.3%（描述性；按 D7 规则，选滤波器时记"持平"）。The CDFS gain is consistent in sign but only about 0.3% (descriptive; "no significant difference" for filter choice under rule D7).
角落 (T017) 已按数据集回答：CDFS 可用（σ = 0.0044 / 0.0038）且看到被删时间的粒子水平高 1.9% / 3.9%;M31 MOS1 太吵（σ = 0.013）。The corner (T017) is answered per set: usable on CDFS (σ = 0.0044 / 0.0038) and seeing a 1.9% / 3.9% higher particle level in rejected time; too noisy on M31 MOS1 (σ = 0.013).
新的探索性张力（未预注册）：角落的粒子水平变化不带入门内 Al 线比（相隔 3.2σ 与 7.7σ）。New exploratory tension (not pre-registered): the corner's particle-level change does not carry into the as-gated Al-line ratio (3.2σ and 7.7σ apart).
还原中的两个执行教训进了 PITFALLS:U 开头的曝光不一定是定标曝光（0555780601 只有 U002 科学曝光，被误标 CalClosed 后纠正）;`XMM_Data_nvme` 名字里有 nvme 但实际在 HDD 上。Two execution lessons from the reduction entered PITFALLS: a U-prefixed exposure id is not necessarily calibration (0555780601 has only U002 science exposures; wrongly skipped as "CalClosed" and corrected); and `XMM_Data_nvme` has "nvme" in its name but lives on an HDD.
下一步由 PI 在两条路里选择:(a) 下游成功判据（D002 第 3 步);(b) Al 线是否跟随角落粒子水平的新预注册。The next step is the PI's choice between two routes: (a) the downstream success criterion (D002 step 3); (b) a new pre-registration on whether the Al line tracks the corner particle level.

2026-10-13(第一版）:本手册随 ParetoGTICal 迁移入 DXC 后建立，覆盖 D001–D005:P001 预注册批准 + 代码冻结 (46d8df3) + 回归门通过（26,111 个数字，最大相对差 0）,M31 MOS1 盲测完成（E002）。当时所有 CDFS P001 结果均为"待定"。

2026-10-13 (first edition): this handbook was built after ParetoGTICal migrated into DXC, covering D001–D005: P001 pre-registration approved + code frozen (46d8df3) + regression gate passed (26,111 numbers, max relative difference 0), and the blind M31 MOS1 run (E002). At that time all CDFS P001 results were "pending".

数据截至Data as of

2026-10-13（所读最新来源：cognition-os/04_DECISION_LOG.md D006 与 02_CURRENT_QUESTION.md "Status (D006)",以及 E003_p001_cdfs_blank_field.md 与收据 results/p001_cdfs/p001_endpoints_all.json,均为 2026-10-13）。

2026-10-13 (newest sources read: cognition-os/04_DECISION_LOG.md D006 and 02_CURRENT_QUESTION.md "Status (D006)", plus E003_p001_cdfs_blank_field.md and the receipt results/p001_cdfs/p001_endpoints_all.json, all dated 2026-10-13).