4706
两条 MOS2 观测完全相同的参数候选。
← CDFS MOS audit 首页 ← 0101 单观测 benchmark ← 0201 negative control
gti_fraction ≥ 0.95；页面据此再在 0800730101 的 Al line-contrast–GTI 平面取 joint frontier。只有同时具有 0101 的 Al/continuum 对比增益、足够 GTI，并且不把 0201 剪到 95% 以下，才值得进入后续 matched 4.background 和 XSherpa 验证。3398 / 4706
保留至少 95% 的 2.filter GTI。
106 reps
所有共同候选的 point-estimate frontier。
88 reps
先过 0201 ≥95%，再在 0101 右上方向筛选。
L_Al/C = Al-K sideband-excess / [rate(1.00–1.40 keV) + rate(1.85–2.00 keV)]；纵轴是 100×[(L_Al/C)_strategy/(L_Al/C)_2.filter−1]，误差棒是 nested-PHA Gehrels delta-method 误差。它是 detector-space 筛选指标，不是拟合线 flux、SP flux 或 calibrated photon flux。广带 Q = rate(1.4–1.8)/[rate(1.0–1.4)+rate(1.8–3.2)] 只作为表中的 concordance control。实心点表示 0201 保留 ≥95%，空心点表示越过这个 guardrail。
R_FOV−alpha R_corner 的统计量做筛选。r22 的 Blocks 只作用于该正确缩放的 residual，并在每段原始 2.filter live run 内以 Gaussian-measurements Bayesian Blocks 分段；它不是原始 FOV rate 的 Blocks，也不是 Poisson 两成分 SP 拟合。sym 4.5、75-s、p0=0.05、q=0.25、z=2、15-s padding 得到 0101 99.08%、Al +0.72±1.15%，0201 100.00%、Al 0.00±1.39%。r23 仅把同一 Blocks 规则的 low-FOV alpha calibration 扩到 0.35/0.50：320 个通过 control 的新增点中，35 个数值不同坐标均被已有前沿支配，其余为已有 GTI 的等价重现。因此 alpha 灵敏度没有继续推高边界。r22 的小步只是在 point estimate 上连接/细化已有前沿，幅度小于一倍统计误差，仍不是 SP 检出。目前最有用的 joint candidates
下表不是最终推荐清单，而是用来决定下一轮 matched 抽谱的代表点。Al-rate retention 是主坐标的保护性互检：它应与 0 相容；0201 的 Al 和广带 Q 也应与 0 相容。hybrid 的共识版本保留任一 selector 接受的时间；保守版本只保留两个都接受的时间。
| strategy | 0101 GTI | 0101 Al / Al rate | 0101 broad Q diagnostic | 0201 GTI | 0201 Al | 0201 broad Q | screen |
|---|---|---|---|---|---|---|---|
scaled-residual BB consensus: sym 4.5s x alpha=0.2, cell=75 s, p0=0.05, q=0.25, z=2, pad=15 sscaled-residual BB: sym 4.5, alpha low 0.2, 75s, p0=0.05, q=0.25, z>2, pad 15s, retain union | 99.08% | +0.72% ± 1.15% Al rate +0.32% ± 0.66% | +0.24% ± 0.83% | 100.00% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
scaled-residual BB consensus: sym 4.25s x alpha=0.2, cell=75 s, p0=0.1, q=0.25, z=1.5, pad=15 sscaled-residual BB: sym 4.25, alpha low 0.2, 75s, p0=0.1, q=0.25, z>1.5, pad 15s, retain union | 98.51% | +1.03% ± 1.21% Al rate +0.41% ± 0.71% | +0.47% ± 0.86% | 100.00% | +0.00% ± 1.39% | +0.00% ± 1.06% | control-safe |
alpha-tail paired CUSUM consensus: sym 4.5s x alpha=0.2, cell=75, q=0.15, k=0.1, H=2.5alpha-tail paired CUSUM: sym 4.5, alpha low 0.2, 75s, retain union | 98.97% | +0.79% ± 1.16% Al rate +0.39% ± 0.67% | +0.39% ± 0.83% | 99.99% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
upper symmetric-line paired CUSUM consensus: sym 4.5s x 90, q=0.35, k=0.25, H=4 (both)symmetric upper-line grid 4.5 sigma × paired CUSUM confirmed_both retain union | 99.37% | +0.54% ± 1.10% Al rate +0.32% ± 0.61% | +0.39% ± 0.81% | 99.99% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
FOV+corner CUSUM consensus: sym3p5s × 90_q0.35_k0.5_h4 (both)paired FOV/corner CUSUM confirmed_both retain union: symmetric_3p5sigma × paired_cusum90_q0p35_k0p5_h4 | 98.03% | +1.93% ± 1.28% Al rate +0.59% ± 0.74% | +1.26% ± 0.90% | 99.99% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
corner-CUSUM consensus: sym4s × 90 q=0.35 k=0.25 H=4corner-scaled high-band CUSUM retain union: symmetric_4sigma × corner_cusum90_q0p35_k0p25_h4 | 98.56% | +1.22% ± 1.21% Al rate +0.44% ± 0.70% | +0.67% ± 0.86% | 99.99% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
FOV+corner CUSUM consensus: sym3p5s × 90_q0.35_k0.5_h4 (both)paired FOV/corner CUSUM confirmed_both retain union: symmetric_3p5sigma × paired_cusum90_q0p35_k0p5_h4 | 98.03% | +1.93% ± 1.28% Al rate +0.59% ± 0.74% | +1.26% ± 0.90% | 99.99% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
FOV+corner CUSUM consensus: sym3s × 90_q0.35_k0.5_h4 (both)paired FOV/corner CUSUM confirmed_both retain union: symmetric_3sigma × paired_cusum90_q0p35_k0p5_h4 | 96.47% | +2.73% ± 1.43% Al rate +0.60% ± 0.84% | +1.87% ± 0.98% | 99.95% | +0.06% ± 1.40% | +0.00% ± 1.06% | control-safe |
corner-CUSUM consensus: sym2p5s × 45 q=0.25 k=0.25 H=3corner-scaled high-band CUSUM retain union: symmetric_2p5sigma × corner_cusum45_q0p25_k0p25_h3 | 88.81% | +5.45% ± 2.01% Al rate +0.81% ± 1.24% | +3.89% ± 1.32% | 99.17% | +0.14% ± 1.47% | +0.05% ± 1.09% | control-safe |
CUSUM consensus: sym2p5s × 90 q=0.25 k=0.25 H=3high-band Poisson CUSUM retain union: symmetric_2p5sigma × cusum_fov90_q0p25_k0p25_h3 | 89.06% | +5.42% ± 1.99% Al rate +0.97% ± 1.22% | +3.83% ± 1.31% | 99.17% | +0.14% ± 1.47% | +0.05% ± 1.09% | control-safe |
CUSUM consensus: sym2p5s × 60 q=0.25 k=0.25 H=3high-band Poisson CUSUM retain union: symmetric_2p5sigma × cusum_fov60_q0p25_k0p25_h3 | 87.96% | +6.53% ± 2.11% Al rate +1.06% ± 1.28% | +4.01% ± 1.36% | 99.13% | +0.18% ± 1.47% | +0.05% ± 1.10% | control-safe |
aligned consensus: sym3s × BB90/.01/z1.5/close330 · dt +15 stiming-aligned retain union: symmetric_3sigma × bb_fov90_p00p01_z1p5_pad0s, bridge <= 330 s, BB-BTI shift +15 s | 95.01% | +3.57% ± 1.55% Al rate +0.90% ± 0.93% | +2.59% ± 1.05% | 100.00% | +0.00% ± 1.39% | +0.00% ± 1.06% | control-safe |
aligned conservative: sym2p5s × BB60/.01/z1.5/close120 · dt -30 stiming-aligned retain intersection: symmetric_2p5sigma × bb_fov60_p00p01_z1p5_pad0s, bridge <= 120 s, BB-BTI shift -30 s | 85.49% | +7.53% ± 2.27% Al rate +1.19% ± 1.38% | +4.67% ± 1.46% | 99.13% | +0.18% ± 1.47% | +0.05% ± 1.10% | joint frontier |
conservative: sym2p5s × bb60/.01/z1p5p15shybrid retain intersection: symmetric_2p5sigma × bb_fov60_p00p01_z1p5_pad15s | 85.26% | +7.67% ± 2.30% Al rate +1.17% ± 1.39% | +4.82% ± 1.47% | 99.13% | +0.18% ± 1.47% | +0.05% ± 1.10% | control-safe |
sym 2p5σsymmetric |z| <= 2.5 | 87.96% | +6.53% ± 2.11% Al rate +1.06% ± 1.28% | +4.01% ± 1.36% | 99.13% | +0.18% ± 1.47% | +0.05% ± 1.10% | control-safe |
BB30/.01/z1p5/p15sBB high-band FOV 30-s: p0=0.01, excess>1.5 sigma, +/-15 s | 64.13% | +11.93% ± 3.71% Al rate +1.25% ± 2.23% | +7.13% ± 2.30% | 100.00% | +0.00% ± 1.39% | +0.00% ± 1.06% | joint frontier |
standard_2filteroriginal 2.filter | 100.00% | +0.00% ± 0.00% Al rate +0.00% ± 0.00% | +0.00% ± 0.00% | 100.00% | +0.00% ± 0.00% | +0.00% ± 0.00% | control-safe |