0800730201 MOS2 negative control

← CDFS MOS audit 首页　 ← 0101 MOS2 strategy benchmark

这是方法控制，不是 SP 检出。 所有 66 个候选均从 0800730201 原始 2.filter 支持出发，使用同一组策略定义和 3B background/RMF 的 detector-space source-PHA screen。没有为每个候选重新运行 ESAS 4.background，因此结果只回答“指标是否在稳定观测上保持接近零”。

2.filter support

15.480 ks

trace 内保留比例 99.10%。

line-free old metric

12 frontier

范围 -7.26% 至 +10.42%。

Al line contrast

8 frontier

范围 -6.73% 至 +1.07%。

1.4–3.2 keV control

6 frontier

范围 -2.90% 至 +0.52%。

negative-control 结论：旧的 line-free continuum reduction 在这条近稳态观测上仍能产生约 +10.42% 的 point estimate；但它的最大正偏差只有 1.90σ。Al 指标的最大值为 +1.07%（0.66σ），宽带指标最大值为 +0.52%（0.46σ），两者都没有超过 2σ。这个结果支持你的判断：旧指标会把随机谱涨落误读成“SP 变少”，而 line-contrast 指标在无 flare 观测上基本守住 0。

三种指标并排比较

三张图使用相同横轴：GTI 相对于 2.filter 的保留率。第一张是旧指标 R=(C_{2.filter}-C_{strategy})/C_{2.filter}；第二张是 Al line/continuum；第三张是 Q=R(1.4–1.8)/[R(1.0–1.4)+R(1.8–3.2)]。在无真实 flare 的控制样本中，后两者应围绕 0，且误差棒覆盖 0。

0800730201 MOS2 old line-free continuum reduction — 旧指标仍出现正的点估计 frontier；这不是 SP 证据，而是选择子集后源谱统计涨落和背景差分造成的 screening artifact。

0800730201 MOS2 Al line contrast — Al line-contrast 的正负变化都在统计误差内；最大正偏差不到 1σ，说明稳定观测没有被系统性推向“SP 减少”。

0800730201 MOS2 broad line contrast — 1.4–3.2 keV 宽带对照同样接近 0；1.8–3.2 keV 仍需响应和仪器结构审计，因此它是 concordance control，不是纯连续谱通量。

为什么这个控制很重要

0201 的 trace 在 2.filter 支持内保留约 99.1%，标准 2.filter 选中的 51-s rate 的均值约 2.188 c/s、标准差约 0.201 c/s；相比 0101 的约 68.7% 支持和约 0.303 c/s 波动，它确实是更接近稳态的测试。若某个指标在这里稳定地产生正 gain，它不能再被解释为 SP 去除；若它只在 0101 这类高波动观测中上升，而在 0201 守住 0，才符合“flare-dependent” 的必要条件。

Si 线在这里反而暴露了一个保护边界：激进的 quantile 子集可出现约 −16.9±7.0% 的 Si ratio（约 −2.42σ），因此 Si 不能作为主优化轴；它只能作为“出现异常时拒绝候选”的交叉诊断。Al 主指标和宽带 Q 在同一候选上没有对应的显著正 gain，这正是我们需要的 negative-control 行为。

0800730201 MOS2 standard QDP selection — QDP/2.filter 选择视图。

0800730201 MOS2 complete raw lightcurve with GTIs — 完整 raw lightcurve 上的 GTI 位置。

0800730201 MOS2 native QDP FOV lightcurve — 复现的 native QDP FOV lightcurve。

66 个候选的数值

负的 line-contrast gain 不是“反向 SP flare”的确定证据；它首先说明该候选在稳态控制样本上没有得到正的改善。只有当结果在多个观测中重复、超过误差、与 Al/宽带两个投影一致，并通过 matched 4.background 与 XSherpa residual/released-sky gates，才可以讨论 SP。

策略	GTI/2.filter	旧 continuum reduction	Al ratio gain	broad Q gain	Al status
`2.filter` original 2.filter	100.00%	+0.00% ± 0.00%	+0.00% ± 0.00%	+0.00% ± 0.00%	frontier
`sym 0.5` symmetric \|z\| <= 0.5	34.83%	-7.26% ± 8.90%	-6.73% ± 6.22%	-2.90% ± 4.19%	dominated
`sym 0.6` symmetric \|z\| <= 0.6	43.85%	-1.11% ± 7.30%	-2.31% ± 5.29%	-0.32% ± 3.51%	dominated
`sym 0.7` symmetric \|z\| <= 0.7	49.61%	-4.75% ± 6.50%	-2.09% ± 4.63%	-0.71% ± 3.07%	dominated
`sym 0.8` symmetric \|z\| <= 0.8	55.70%	-3.11% ± 5.74%	-2.16% ± 4.13%	-1.01% ± 2.73%	dominated
`sym 0.9` symmetric \|z\| <= 0.9	64.48%	-0.46% ± 4.79%	-0.03% ± 3.57%	-0.58% ± 2.36%	dominated
`sym 1` symmetric \|z\| <= 1	69.13%	-1.11% ± 4.32%	-0.01% ± 3.25%	-0.79% ± 2.16%	dominated
`sym 1.2` symmetric \|z\| <= 1.2	77.31%	-2.71% ± 3.49%	-1.06% ± 2.73%	-1.78% ± 1.83%	dominated
`sym 1.4` symmetric \|z\| <= 1.4	83.88%	-4.25% ± 2.80%	-2.56% ± 2.32%	-2.28% ± 1.60%	dominated
`sym 1.6` symmetric \|z\| <= 1.6	88.97%	-2.59% ± 2.28%	-1.47% ± 2.07%	-1.42% ± 1.45%	dominated
`sym 1.8` symmetric \|z\| <= 1.8	92.92%	+0.03% ± 1.87%	+0.17% ± 1.89%	+0.01% ± 1.34%	dominated
`sym 2` symmetric \|z\| <= 2	95.94%	+2.05% ± 1.55%	+1.07% ± 1.74%	+0.39% ± 1.24%	frontier
`sym 2.25` symmetric \|z\| <= 2.25	97.86%	+1.40% ± 1.19%	+1.05% ± 1.60%	+0.37% ± 1.16%	frontier
`sym 2.5` symmetric \|z\| <= 2.5	99.13%	+0.07% ± 0.76%	+0.18% ± 1.47%	+0.05% ± 1.10%	dominated
`sym 2.75` symmetric \|z\| <= 2.75	99.54%	-0.26% ± 0.55%	+0.26% ± 1.43%	+0.01% ± 1.07%	frontier
`sym 3` symmetric \|z\| <= 3	99.75%	-0.17% ± 0.45%	+0.20% ± 1.41%	+0.01% ± 1.06%	frontier
`sym 3.25` symmetric \|z\| <= 3.25	99.88%	-0.31% ± 0.29%	+0.08% ± 1.40%	+0.03% ± 1.06%	dominated
`sym 3.5` symmetric \|z\| <= 3.5	99.93%	-0.18% ± 0.28%	+0.04% ± 1.39%	+0.03% ± 1.06%	dominated
`sym 4` symmetric \|z\| <= 4	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	frontier
`U2/p0` upper BTI z > 2, +/-0 s	98.53%	+1.29% ± 1.04%	+0.01% ± 1.55%	+0.07% ± 1.14%	dominated
`U2/p25.5` upper BTI z > 2, +/-25.5 s	92.18%	+2.28% ± 2.05%	-1.43% ± 1.99%	-1.07% ± 1.39%	dominated
`U2/p51` upper BTI z > 2, +/-51 s	86.56%	-0.08% ± 2.61%	-1.67% ± 2.27%	-0.97% ± 1.58%	dominated
`U2.5/p0` upper BTI z > 2.5, +/-0 s	99.60%	+0.81% ± 0.69%	+0.15% ± 1.45%	+0.21% ± 1.08%	dominated
`U2.5/p25.5` upper BTI z > 2.5, +/-25.5 s	96.14%	+2.24% ± 1.53%	-0.33% ± 1.75%	-0.24% ± 1.25%	dominated
`U2.5/p51` upper BTI z > 2.5, +/-51 s	93.00%	+1.39% ± 1.92%	-0.40% ± 1.94%	-0.50% ± 1.35%	dominated
`U3/p0` upper BTI z > 3, +/-0 s	99.92%	-0.05% ± 0.35%	+0.10% ± 1.40%	+0.00% ± 1.06%	frontier
`U3/p25.5` upper BTI z > 3, +/-25.5 s	98.93%	+0.92% ± 0.91%	+0.07% ± 1.51%	-0.03% ± 1.12%	dominated
`U3/p51` upper BTI z > 3, +/-51 s	97.94%	+1.15% ± 1.15%	+0.01% ± 1.59%	+0.09% ± 1.17%	dominated
`U3.5/p0` upper BTI z > 3.5, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`U3.5/p25.5` upper BTI z > 3.5, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`U3.5/p51` upper BTI z > 3.5, +/-51 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`U4/p0` upper BTI z > 4, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`U4/p25.5` upper BTI z > 4, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`U4/p51` upper BTI z > 4, +/-51 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E2.5/0.5/p0` sustained z > 2.5 core + z > 0.5 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E2.5/0.5/p25.5` sustained z > 2.5 core + z > 0.5 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E2.5/1/p0` sustained z > 2.5 core + z > 1 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E2.5/1/p25.5` sustained z > 2.5 core + z > 1 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3/0.5/p0` sustained z > 3 core + z > 0.5 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3/0.5/p25.5` sustained z > 3 core + z > 0.5 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3/1/p0` sustained z > 3 core + z > 1 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3/1/p25.5` sustained z > 3 core + z > 1 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3.5/0.5/p0` sustained z > 3.5 core + z > 0.5 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3.5/0.5/p25.5` sustained z > 3.5 core + z > 0.5 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3.5/1/p0` sustained z > 3.5 core + z > 1 shoulder, +/-0 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`E3.5/1/p25.5` sustained z > 3.5 core + z > 1 shoulder, +/-25.5 s	99.95%	-0.12% ± 0.28%	+0.04% ± 1.39%	+0.00% ± 1.06%	equivalent
`Q2/p0` remove highest 2% decision rates, +/-0 s	97.75%	+0.49% ± 1.14%	-0.58% ± 1.59%	-0.25% ± 1.18%	dominated
`Q2/p25.5` remove highest 2% decision rates, +/-25.5 s	89.45%	+0.85% ± 2.32%	-2.79% ± 2.12%	-1.46% ± 1.49%	dominated
`Q5/p0` remove highest 5% decision rates, +/-0 s	94.95%	+1.32% ± 1.66%	-1.04% ± 1.80%	-0.73% ± 1.29%	dominated
`Q5/p25.5` remove highest 5% decision rates, +/-25.5 s	78.85%	+0.41% ± 3.39%	-3.35% ± 2.69%	-1.96% ± 1.85%	dominated
`Q7.5/p0` remove highest 7.5% decision rates, +/-0 s	91.49%	+3.01% ± 2.16%	-0.72% ± 2.05%	-0.56% ± 1.45%	dominated
`Q7.5/p25.5` remove highest 7.5% decision rates, +/-25.5 s	65.90%	+2.74% ± 4.67%	-1.85% ± 3.52%	-1.43% ± 2.39%	dominated
`Q10/p0` remove highest 10% decision rates, +/-0 s	89.68%	+3.37% ± 2.37%	-0.67% ± 2.16%	-0.30% ± 1.51%	dominated
`Q10/p25.5` remove highest 10% decision rates, +/-25.5 s	60.81%	+2.86% ± 5.20%	-1.88% ± 3.86%	-0.59% ± 2.65%	dominated
`Q15/p0` remove highest 15% decision rates, +/-0 s	84.17%	+5.67% ± 2.99%	-1.12% ± 2.49%	-0.53% ± 1.71%	dominated
`Q15/p25.5` remove highest 15% decision rates, +/-25.5 s	48.09%	+8.51% ± 6.71%	-2.74% ± 5.00%	-1.80% ± 3.40%	dominated
`Q20/p0` remove highest 20% decision rates, +/-0 s	79.22%	+4.47% ± 3.43%	-4.02% ± 2.71%	-2.08% ± 1.86%	dominated
`Q20/p25.5` remove highest 20% decision rates, +/-25.5 s	39.70%	+10.42% ± 7.93%	-2.19% ± 6.02%	-0.49% ± 4.18%	dominated
`L2/p0` 601-s rolling-median excess > 2 sigma, +/-0 s	97.57%	+0.19% ± 1.16%	-0.63% ± 1.60%	+0.00% ± 1.19%	dominated
`L2/p25.5` 601-s rolling-median excess > 2 sigma, +/-25.5 s	88.52%	+1.95% ± 2.45%	-2.19% ± 2.19%	-1.16% ± 1.52%	dominated
`L2.5/p0` 601-s rolling-median excess > 2.5 sigma, +/-0 s	98.99%	+1.53% ± 0.96%	+0.61% ± 1.51%	+0.52% ± 1.12%	frontier
`L2.5/p25.5` 601-s rolling-median excess > 2.5 sigma, +/-25.5 s	94.73%	+1.05% ± 1.67%	-0.55% ± 1.82%	-0.56% ± 1.30%	dominated
`L3/p0` 601-s rolling-median excess > 3 sigma, +/-0 s	99.67%	+0.53% ± 0.63%	+0.19% ± 1.44%	+0.30% ± 1.08%	dominated
`L3/p25.5` 601-s rolling-median excess > 3 sigma, +/-25.5 s	97.66%	+1.81% ± 1.25%	+0.09% ± 1.63%	+0.30% ± 1.19%	dominated
`L3.5/p0` 601-s rolling-median excess > 3.5 sigma, +/-0 s	99.92%	-0.06% ± 0.35%	+0.10% ± 1.40%	+0.03% ± 1.06%	dominated
`L3.5/p25.5` 601-s rolling-median excess > 3.5 sigma, +/-25.5 s	99.26%	+0.39% ± 0.76%	-0.27% ± 1.47%	-0.15% ± 1.10%	dominated

下一步：将 0201 作为固定 negative control 保留在所有后续策略 benchmark 中。任何新策略如果同时在 0101 提升 line contrast、在 0201 保持 line contrast≈0，并且保留足够 GTI，才值得进入逐策略 matched 4.background 抽谱。