PPC residual wrong-FIX analysis
Reference truth is used only after the run to label wrong FIX; the source and gate columns use data already present in the real-time POS/selector outputs.
Wrong-FIX threshold: 0.500 m 3D error.
Key findings
min_sat9 wrong source: unknown=1179; rule matched: unknown=1179.- Best low-collateral deployable gate for
min_sat9 in this POS-only replay: nsat < 8 or (nsat <= 11 and ratio <= 15) catches 25.276% of wrong FIX while touching 1.056% of good FIX.
integrity_gate wrong source: unknown=927; rule matched: unknown=927.- Best low-collateral deployable gate for
integrity_gate in this POS-only replay: nis_per_obs > 20 catches 2.913% of wrong FIX while touching 1.591% of good FIX.
wrong_basin_escape wrong source: unknown=894; rule matched: unknown=894.- Best low-collateral deployable gate for
wrong_basin_escape in this POS-only replay: nis_per_obs > 20 catches 2.908% of wrong FIX while touching 1.591% of good FIX.
consensus_escape wrong source: unknown=869; rule matched: unknown=869.- Best low-collateral deployable gate for
consensus_escape in this POS-only replay: nis_per_obs > 20 catches 2.992% of wrong FIX while touching 1.591% of good FIX.
online_consensus_kinematic_tokyo3_fgo wrong source: unknown=668; rule matched: unknown=668.- Best low-collateral deployable gate for
online_consensus_kinematic_tokyo3_fgo in this POS-only replay: nis_per_obs > 20 catches 3.293% of wrong FIX while touching 1.591% of good FIX.
staged_integrity wrong source: unknown=574; rule matched: unknown=574.- Best low-collateral deployable gate for
staged_integrity in this POS-only replay: nis_per_obs > 20 catches 2.962% of wrong FIX while touching 1.602% of good FIX.
Profile summary
| Profile |
FIX epochs |
Wrong FIX |
Wrong/FIX |
Wrong source |
Rule matched |
| min_sat9 |
41800 |
1179 |
2.821% |
unknown=1179 |
unknown=1179 |
| integrity_gate |
41220 |
927 |
2.249% |
unknown=927 |
unknown=927 |
| wrong_basin_escape |
41185 |
894 |
2.171% |
unknown=894 |
unknown=894 |
| consensus_escape |
41160 |
869 |
2.111% |
unknown=869 |
unknown=869 |
| online_consensus_kinematic_tokyo3_fgo |
40963 |
668 |
1.631% |
unknown=668 |
unknown=668 |
| staged_integrity |
40896 |
574 |
1.404% |
unknown=574 |
unknown=574 |
Leave-one-run-out integrity-gate validation
six-fold leave-one-run-out; <=2% training correct-FIX harm; minimize >5m, wrong FIX, then FIX loss.
| Held-out run |
Selected min sat / ceiling / ratio |
min-sat9 wrong |
LOO wrong |
LOO >5 m |
| tokyo_run1 |
8 / 13 / 10.000 |
440 |
304 |
25 |
| tokyo_run2 |
8 / 12 / 15.000 |
95 |
46 |
13 |
| tokyo_run3 |
9 / 11 / 30.000 |
335 |
310 |
16 |
| nagoya_run1 |
8 / 12 / 15.000 |
46 |
31 |
5 |
| nagoya_run2 |
8 / 12 / 15.000 |
90 |
82 |
9 |
| nagoya_run3 |
9 / 12 / 20.000 |
90 |
76 |
12 |
min_sat9
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
9029 |
440 |
4.873% |
93 |
29 |
17.332 m |
188219.400-188225.600 (6.400 s) |
| tokyo_run2 |
7484 |
95 |
1.269% |
19 |
10 |
12.175 m |
177925.600-177927.200 (1.800 s) |
| tokyo_run3 |
12003 |
335 |
2.791% |
17 |
8 |
4.926 m |
181708.800-181747.600 (39.000 s) |
| nagoya_run1 |
6069 |
46 |
0.758% |
12 |
7 |
34.397 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4748 |
123 |
2.591% |
43 |
37 |
28.393 m |
556737.000-556746.200 (9.400 s) |
| nagoya_run3 |
2467 |
140 |
5.675% |
65 |
61 |
105.781 m |
554493.600-554495.200 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
31.700/140.600/273.000 |
7.300/172.700/554.000 |
| baseline_m |
2952.167/9441.529/9622.160 |
3040.234/9312.071/9371.640 |
| nsat |
21.000/30.000/32.000 |
13.000/26.000/27.000 |
| prefit_rms_m |
0.939/10.668/23.600 |
3.537/16.268/42.161 |
| prefit_max_m |
6.001/35.933/99.661 |
12.436/58.499/132.875 |
| post_rms_m |
0.489/0.746/0.825 |
0.681/0.940/1.093 |
| post_max_m |
2.684/2.956/2.979 |
2.739/2.973/2.986 |
| nis_per_obs |
0.346/2.074/4.806 |
0.783/4.857/7.945 |
| outliers |
4.000/18.000/26.000 |
13.000/43.000/50.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
298 |
25.276% |
429 |
1.056% |
| ratio < 6 |
410 |
34.775% |
4749 |
11.691% |
| ratio < 8 |
622 |
52.757% |
6282 |
15.465% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
30 |
2.545% |
646 |
1.590% |
| nis_per_obs > 50 |
12 |
1.018% |
221 |
0.544% |
| prefit_rms > 5 m |
544 |
46.141% |
6547 |
16.117% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
integrity_gate
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
8802 |
291 |
3.306% |
20 |
10 |
8.915 m |
189208.800-189214.200 (5.600 s) |
| tokyo_run2 |
7349 |
46 |
0.626% |
13 |
7 |
12.175 m |
178223.400-178224.600 (1.400 s) |
| tokyo_run3 |
11938 |
315 |
2.639% |
17 |
8 |
6.047 m |
181708.800-181747.600 (39.000 s) |
| nagoya_run1 |
6036 |
33 |
0.547% |
5 |
3 |
12.017 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4723 |
119 |
2.520% |
42 |
36 |
28.390 m |
556737.000-556746.200 (9.400 s) |
| nagoya_run3 |
2372 |
123 |
5.185% |
56 |
52 |
105.794 m |
554493.600-554495.200 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
32.200/143.900/281.900 |
11.800/405.400/1778.100 |
| baseline_m |
2952.169/9441.529/9631.654 |
3043.482/9371.609/9371.640 |
| nsat |
21.000/30.000/32.000 |
14.000/26.000/27.000 |
| prefit_rms_m |
0.947/10.994/23.650 |
6.695/19.429/46.124 |
| prefit_max_m |
6.090/36.157/99.775 |
19.935/69.870/133.704 |
| post_rms_m |
0.489/0.745/0.825 |
0.670/0.917/1.052 |
| post_max_m |
2.686/2.956/2.979 |
2.795/2.976/2.988 |
| nis_per_obs |
0.346/2.053/4.801 |
0.922/4.635/8.969 |
| outliers |
4.000/18.000/26.000 |
19.000/47.000/52.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
0 |
0.000% |
9 |
0.022% |
| ratio < 6 |
223 |
24.056% |
4511 |
11.195% |
| ratio < 8 |
379 |
40.885% |
5969 |
14.814% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
27 |
2.913% |
641 |
1.591% |
| nis_per_obs > 50 |
12 |
1.294% |
220 |
0.546% |
| prefit_rms > 5 m |
509 |
54.908% |
6526 |
16.196% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
wrong_basin_escape
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
8802 |
291 |
3.306% |
20 |
10 |
8.915 m |
189208.800-189214.200 (5.600 s) |
| tokyo_run2 |
7349 |
46 |
0.626% |
13 |
7 |
12.175 m |
178223.400-178224.600 (1.400 s) |
| tokyo_run3 |
11938 |
315 |
2.639% |
17 |
8 |
6.047 m |
181708.800-181747.600 (39.000 s) |
| nagoya_run1 |
6036 |
33 |
0.547% |
5 |
3 |
12.017 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4688 |
86 |
1.834% |
10 |
4 |
8.031 m |
556737.000-556746.200 (9.400 s) |
| nagoya_run3 |
2372 |
123 |
5.185% |
56 |
52 |
105.794 m |
554493.600-554495.200 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
32.200/143.900/282.000 |
11.300/476.000/1799.400 |
| baseline_m |
2952.169/9441.529/9631.654 |
2988.007/9371.611/9371.640 |
| nsat |
21.000/30.000/32.000 |
14.000/27.000/27.000 |
| prefit_rms_m |
0.947/10.953/23.649 |
6.322/19.546/46.126 |
| prefit_max_m |
6.088/36.119/99.775 |
19.274/71.166/133.707 |
| post_rms_m |
0.489/0.745/0.825 |
0.681/0.921/1.056 |
| post_max_m |
2.686/2.956/2.979 |
2.796/2.976/2.987 |
| nis_per_obs |
0.346/2.053/4.801 |
0.864/4.815/8.614 |
| outliers |
4.000/18.000/26.000 |
19.000/42.000/52.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
0 |
0.000% |
9 |
0.022% |
| ratio < 6 |
212 |
23.714% |
4511 |
11.196% |
| ratio < 8 |
367 |
41.051% |
5969 |
14.815% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
26 |
2.908% |
641 |
1.591% |
| nis_per_obs > 50 |
11 |
1.230% |
220 |
0.546% |
| prefit_rms > 5 m |
476 |
53.244% |
6524 |
16.192% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
consensus_escape
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
8802 |
291 |
3.306% |
20 |
10 |
8.915 m |
189208.800-189214.200 (5.600 s) |
| tokyo_run2 |
7349 |
46 |
0.626% |
13 |
7 |
12.175 m |
178223.400-178224.600 (1.400 s) |
| tokyo_run3 |
11938 |
315 |
2.639% |
17 |
8 |
6.047 m |
181708.800-181747.600 (39.000 s) |
| nagoya_run1 |
6036 |
33 |
0.547% |
5 |
3 |
12.017 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4688 |
86 |
1.834% |
10 |
4 |
8.031 m |
556737.000-556746.200 (9.400 s) |
| nagoya_run3 |
2347 |
98 |
4.176% |
31 |
27 |
105.849 m |
554493.600-554495.200 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
32.200/143.900/282.000 |
12.600/490.500/1929.900 |
| baseline_m |
2952.169/9441.529/9631.654 |
2974.659/9371.611/9371.641 |
| nsat |
21.000/30.000/32.000 |
14.000/27.000/27.000 |
| prefit_rms_m |
0.947/10.953/23.649 |
5.577/19.596/47.439 |
| prefit_max_m |
6.088/36.119/99.775 |
17.148/71.577/133.713 |
| post_rms_m |
0.489/0.745/0.825 |
0.675/0.923/1.065 |
| post_max_m |
2.686/2.956/2.979 |
2.792/2.976/2.988 |
| nis_per_obs |
0.346/2.053/4.801 |
0.978/4.867/9.379 |
| outliers |
4.000/18.000/26.000 |
18.000/42.000/53.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
0 |
0.000% |
9 |
0.022% |
| ratio < 6 |
196 |
22.555% |
4511 |
11.196% |
| ratio < 8 |
343 |
39.471% |
5969 |
14.815% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
26 |
2.992% |
641 |
1.591% |
| nis_per_obs > 50 |
11 |
1.266% |
220 |
0.546% |
| prefit_rms > 5 m |
451 |
51.899% |
6524 |
16.192% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
online_consensus_kinematic_tokyo3_fgo
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
8792 |
283 |
3.219% |
12 |
2 |
4.429 m |
189208.800-189214.200 (5.600 s) |
| tokyo_run2 |
7339 |
39 |
0.531% |
6 |
0 |
6.745 m |
177620.600-177621.400 (1.000 s) |
| tokyo_run3 |
11784 |
161 |
1.366% |
9 |
0 |
6.047 m |
181708.800-181720.000 (11.400 s) |
| nagoya_run1 |
6036 |
33 |
0.547% |
5 |
3 |
12.017 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4684 |
82 |
1.751% |
6 |
0 |
5.191 m |
556737.000-556746.200 (9.400 s) |
| nagoya_run3 |
2328 |
70 |
3.007% |
4 |
0 |
7.194 m |
554711.200-554712.800 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
32.200/143.900/281.900 |
17.400/844.900/5218.400 |
| baseline_m |
2952.170/9441.529/9631.654 |
3057.840/9371.630/9381.523 |
| nsat |
21.000/30.000/32.000 |
13.000/21.000/27.000 |
| prefit_rms_m |
0.943/10.747/23.637 |
3.157/41.996/49.175 |
| prefit_max_m |
6.053/35.991/99.775 |
12.046/117.622/133.779 |
| post_rms_m |
0.489/0.745/0.825 |
0.627/0.951/1.104 |
| post_max_m |
2.685/2.956/2.979 |
2.734/2.967/2.985 |
| nis_per_obs |
0.345/2.052/4.797 |
0.659/5.321/9.573 |
| outliers |
4.000/18.000/26.000 |
10.000/38.000/43.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
0 |
0.000% |
43 |
0.107% |
| ratio < 6 |
154 |
23.054% |
4521 |
11.220% |
| ratio < 8 |
207 |
30.988% |
5981 |
14.843% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
22 |
3.293% |
641 |
1.591% |
| nis_per_obs > 50 |
7 |
1.048% |
220 |
0.546% |
| prefit_rms > 5 m |
276 |
41.317% |
6505 |
16.143% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
staged_integrity
| Run |
FIX epochs |
Wrong FIX |
Wrong/FIX |
>5 m |
>10 m |
Wrong p95 |
Longest wrong span |
| tokyo_run1 |
8792 |
251 |
2.855% |
12 |
2 |
4.438 m |
189208.800-189214.200 (5.600 s) |
| tokyo_run2 |
7339 |
34 |
0.463% |
6 |
0 |
6.745 m |
177722.600-177723.400 (1.000 s) |
| tokyo_run3 |
11784 |
161 |
1.366% |
9 |
0 |
6.047 m |
181708.800-181720.000 (11.400 s) |
| nagoya_run1 |
6036 |
33 |
0.547% |
5 |
3 |
12.017 m |
551591.400-551593.400 (2.200 s) |
| nagoya_run2 |
4617 |
25 |
0.541% |
6 |
0 |
7.977 m |
556741.400-556742.800 (1.600 s) |
| nagoya_run3 |
2328 |
70 |
3.007% |
4 |
0 |
7.194 m |
554711.200-554712.800 (1.800 s) |
Real-time discriminator percentiles
| Metric |
Good FIX p50/p90/p95 |
Wrong FIX p50/p90/p95 |
| ratio |
32.200/144.400/284.700 |
18.200/747.200/6564.200 |
| baseline_m |
2952.169/9441.529/9630.451 |
3047.457/8064.896/9536.737 |
| nsat |
21.000/30.000/32.000 |
13.000/25.000/27.000 |
| prefit_rms_m |
0.942/10.680/23.600 |
2.746/15.338/19.448 |
| prefit_max_m |
6.040/35.927/99.668 |
11.308/55.129/70.253 |
| post_rms_m |
0.489/0.745/0.825 |
0.622/0.959/1.105 |
| post_max_m |
2.685/2.956/2.979 |
2.737/2.970/2.986 |
| nis_per_obs |
0.345/2.057/4.806 |
0.626/5.397/8.969 |
| outliers |
4.000/18.000/26.000 |
9.000/38.000/49.000 |
Deployable gate simulation
| Gate |
Wrong caught |
Wrong caught |
Good FIX harmed |
Good FIX harmed |
| nsat < 8 or (nsat <= 11 and ratio <= 15) |
0 |
0.000% |
43 |
0.107% |
| ratio < 6 |
129 |
22.474% |
4520 |
11.210% |
| ratio < 8 |
172 |
29.965% |
5979 |
14.828% |
| post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |
| post_rms > 6 m |
0 |
0.000% |
0 |
0.000% |
| nis_per_obs > 20 |
17 |
2.962% |
646 |
1.602% |
| nis_per_obs > 50 |
7 |
1.220% |
220 |
0.546% |
| prefit_rms > 5 m |
219 |
38.153% |
6495 |
16.108% |
| baseline > 7000 m and post_rms > 4 m |
0 |
0.000% |
0 |
0.000% |