guluchen · Nov 8, 2026 · Nov 6, 2026 · Nov 7, 2026 · Nov 7, 2026 · Nov 7, 2026
diff --git a/NOTICE b/NOTICE
@@ -372,6 +372,19 @@ TN_496_2_27to30), TN_540_4_24to34 (was TN_540_4_23to34), TN_648_14_22to28 (was T
 TN_684_2_24to29 (was TN_684_2_22to29); COLOR_1027_1_31to37 (was COLOR_1027_1_28to37) and
 COLOR_2791_1_28to61 (was COLOR_2791_1_26to61), section 9.
 
+Renamed 2026-11-07 (same matrices, upper bounds unchanged): lower bounds raised by DistQLDPC's optional CC engine in
+12-36 h follow-up runs of GH-225 (#225; records in optimization/experiments/GH-225/lb-push/). Each was reached by two runs
+with different search trees: main 755e0c21's CC default (gated erasure) and main 1239e9a2's CC default plus ungated
+erasure, or, for TN_684, relabelled qubits and rows (-seed=1, -seed=2), or, for COLOR_2791, a different row order. This is
+the same engine, not an independent second method. BB_756_16_34, TN_288_2_24, TN_360_4_24 and TN_496_2_30 are now exact:
+the lower bound meets the GH-146 upper-bound witness. Stems: BB_1080_4_34to48 (was BB_1080_4_32to48), BB_756_16_34 (was BB_756_16_32to34), BB_864_4_34to40 (was
+BB_864_4_32to40), COLOR_1027_1_33to37 (was COLOR_1027_1_31to37), COLOR_2791_1_30to61 (was COLOR_2791_1_28to61),
+TN_288_2_24 (was TN_288_2_22to24), TN_324_4_24to25 (was TN_324_4_23to25), TN_360_4_24 (was TN_360_4_23to24),
+TN_396_4_24to26 (was TN_396_4_23to26), TN_432_8_25to29 (was TN_432_8_23to29), TN_468_4_25to29 (was
+TN_468_4_23to29), TN_496_2_30 (was TN_496_2_28to30), TN_504_2_25to34 (was TN_504_2_24to34), TN_540_4_25to34 (was
+TN_540_4_24to34), TN_576_2_26to37 (was TN_576_2_24to37), TN_612_2_26to40 (was TN_612_2_24to40), TN_648_10_23to32
+(was TN_648_10_22to32), TN_648_14_23to28 (was TN_648_14_22to28), TN_684_2_25to29 (was TN_684_2_24to29).
+
 9. Triangular 6.6.6 color-code matrices (GH-156; tiered 2026-11-04)
 ------------------------------------------------------------------
 
@@ -390,6 +403,8 @@ Placed and renamed 2026-11-04: COLOR_7_1_3 (tier0), COLOR_19_1_5 (tier0), COLOR_
 Renamed 2026-11-06 (same matrices, upper bounds unchanged; CC lower bounds reproduced by runs with different
 search trees, same engine, see section 5): COLOR_1027_1_31to37 (was COLOR_1027_1_28to37), COLOR_2791_1_28to61
 (was COLOR_2791_1_26to61).
+Renamed 2026-11-07 (same matrices, upper bounds unchanged; see section 5): COLOR_1027_1_33to37 (was
+COLOR_1027_1_31to37), COLOR_2791_1_30to61 (was COLOR_2791_1_28to61).
 
 10. Twisted-torus generalized toric / BB matrices (GH-159; tiered 2026-11-04)
 --------------------------------------------------------------------------

diff --git a/README.md b/README.md
@@ -40,7 +40,7 @@ For a code named `<code>`, put four text files in one directory (default: `data/
 
 **File format:** each line is one binary row; entries are `0` or `1` separated by spaces. Lines starting with `#` are comments. All four matrices must have the same number of columns `n` (qubits).
 
-**Stem names** use `{family}_{n}_{k}_{d}` (e.g. `BB_72_12_6`, `LP_34_20_2`), family in `BB`, `GB`, `LP`, `TN`. When the distance is not known, the last field holds the current proven bounds: `{lb}to{ub}` or `ge{lb}` (e.g. `TN_648_10_22to32`, `BB_756_16_32to34`; upper bounds from verified witnesses, see [data/matrices/README.md](data/matrices/README.md)). Upstream `AJ_*` / `xu_*` / `PK_*` ids and renames are recorded in [NOTICE](NOTICE) and [data/matrices/README.md](data/matrices/README.md).
+**Stem names** use `{family}_{n}_{k}_{d}` (e.g. `BB_72_12_6`, `LP_34_20_2`), family in `BB`, `GB`, `LP`, `TN`. When the distance is not known, the last field holds the current proven bounds: `{lb}to{ub}` or `ge{lb}` (e.g. `TN_648_10_23to32`, `BB_864_4_34to40`; upper bounds from verified witnesses, see [data/matrices/README.md](data/matrices/README.md)). Upstream `AJ_*` / `xu_*` / `PK_*` ids and renames are recorded in [NOTICE](NOTICE) and [data/matrices/README.md](data/matrices/README.md).
 
 Example layout:
 

diff --git a/data/matrices/README.md b/data/matrices/README.md
@@ -33,11 +33,11 @@ certified).
 | tier2 | medium filter; main 10–100 s on yfclab2 | TN_108_2_12 [[108,2,12]], BB_144_14_14 [[144,14,14]], TN_144_2_13 [[144,2,13]], TN_200_10_10 [[200,10,10]], BB_248_10_18 [[248,10,18]], BB_340_16_18 [[340,16,18]], AMC_522_6_20 [[522,6,20]], LP_544_80_12 [[544,80,12]] |
 | tier3 | decisive set; main 100–2000 s on yfclab2 | TN_250_10_15 [[250,10,15]], BB_294_10_20 [[294,10,20]], HGP_481_1_16 [[481,1,16]], LP_714_100_16 [[714,100,16]], LP_1020_136_16 [[1020,136,16]], HGP_1261_121_6 [[1261,121,6]], HGP_2866_256_8 [[2866,256,8]] |
 | tier4 | hard; main 2000–5000 s on yfclab2; timeout 2000 s Mac / 5000 s yfclab2 | (none) |
-| tier5 | open set; main > 5000 s or unsolved on yfclab2; timeout 21600 s (6 h) | TN_180_2_17 [[180,2,17]], TN_216_4_18 [[216,4,18]], TN_252_2_21 [[252,2,21]], TN_288_2_22to24 [[288,2,22–24]], TN_324_4_23to25 [[324,4,23–25]], COLOR_331_1_21 [[331,1,21]], BB_360_12_24 [[360,12,24]], TN_360_4_23to24 [[360,4,23–24]], TN_396_4_23to26 [[396,4,23–26]], TN_432_8_23to29 [[432,8,23–29]], TN_468_4_23to29 [[468,4,23–29]], TN_496_2_28to30 [[496,2,28–30]], TN_504_2_24to34 [[504,2,24–34]], TN_540_4_24to34 [[540,4,24–34]], TN_576_2_24to37 [[576,2,24–37]], TN_612_2_24to40 [[612,2,24–40]], TN_648_10_22to32 [[648,10,22–32]], TN_648_14_22to28 [[648,14,22–28]], TN_684_2_24to29 [[684,2,24–29]], BB_756_16_32to34 [[756,16,32–34]], BB_864_4_32to40 [[864,4,32–40]], COLOR_1027_1_31to37 [[1027,1,31–37]], LP_1054_140_20 [[1054,140,20]], BB_1080_4_32to48 [[1080,4,32–48]], HGP_1105_1_24 [[1105,1,24]], LP_1122_148_20 [[1122,148,20]], LP_1428_184_24 [[1428,184,24]], LP_1768_224_22 [[1768,224,22]], LP_2610_744_16 [[2610,744,16]], COLOR_2791_1_28to61 [[2791,1,28–61]] |
+| tier5 | open set; main > 5000 s or unsolved on yfclab2; timeout 21600 s (6 h) | TN_180_2_17 [[180,2,17]], TN_216_4_18 [[216,4,18]], TN_252_2_21 [[252,2,21]], TN_288_2_24 [[288,2,24]], TN_324_4_24to25 [[324,4,24–25]], COLOR_331_1_21 [[331,1,21]], BB_360_12_24 [[360,12,24]], TN_360_4_24 [[360,4,24]], TN_396_4_24to26 [[396,4,24–26]], TN_432_8_25to29 [[432,8,25–29]], TN_468_4_25to29 [[468,4,25–29]], TN_496_2_30 [[496,2,30]], TN_504_2_25to34 [[504,2,25–34]], TN_540_4_25to34 [[540,4,25–34]], TN_576_2_26to37 [[576,2,26–37]], TN_612_2_26to40 [[612,2,26–40]], TN_648_10_23to32 [[648,10,23–32]], TN_648_14_23to28 [[648,14,23–28]], TN_684_2_25to29 [[684,2,25–29]], BB_756_16_34 [[756,16,34]], BB_864_4_34to40 [[864,4,34–40]], COLOR_1027_1_33to37 [[1027,1,33–37]], LP_1054_140_20 [[1054,140,20]], BB_1080_4_34to48 [[1080,4,34–48]], HGP_1105_1_24 [[1105,1,24]], LP_1122_148_20 [[1122,148,20]], LP_1428_184_24 [[1428,184,24]], LP_1768_224_22 [[1768,224,22]], LP_2610_744_16 [[2610,744,16]], COLOR_2791_1_30to61 [[2791,1,30–61]] |
 
 Counts: tier0 26, tier1 9, tier2 8, tier3 7, tier4 0, tier5 30 (total 80).
 
-**Tier assignment rule (PI 2026-10-12):** by main's time on yfclab2 (slower of no-card / card-mto, `optimization/baselines/SERVER.md`): < 1 s tier0, 1–10 s tier1, 10–100 s tier2, 100–2000 s tier3, 2000–5000 s tier4, > 5000 s or unsolved tier5. Re-check placements when main changes. 2026-11-04 (main 02879a93 baseline of all 75 codes, 5000 s panel limit for every code, `optimization/baselines/SERVER.md`; PI plan 2026-11-04): BB_288_12_18 tier2 -> tier1 (9.58 s), GB_144_12_12 tier2 -> tier1 (1.10 s), TN_144_2_13 tier3 -> tier2 (88.17 s), TN_250_10_15 tier4 -> tier3 (1283 s), LP_1020_136_16 tier5 -> tier3 (393 s); the 29 library codes placed by the same rule (tier0: COLOR_7_1_3, HGP_13_1_3, COLOR_19_1_5, BB_30_4_6, HGP_41_1_5, HGP_45_9_3, HGP_106_16_4, AMC_108_6_9, HGP_113_1_8, BB_120_8_12, AMC_216_6_12, HGP_306_36_5, AMC_330_6_14; tier1: COLOR_127_1_13, BB_186_10_14, HGP_265_1_12, AMC_438_6_17, HGP_610_64_6; tier2: BB_248_10_18, BB_340_16_18, AMC_522_6_20; tier3: BB_294_10_20, HGP_481_1_16, HGP_1261_121_6, HGP_2866_256_8; tier5: COLOR_331_1_21, COLOR_1027_1_31to37, HGP_1105_1_24, COLOR_2791_1_28to61). Codes that main solves within 6 h but over 5000 s stay in tier5 (TN_180_2_17, TN_216_4_18, BB_360_12_24, LP_1122_148_20); HGP_1105_1_24 is in tier5 because MaxSAT does not solve it in 6 h (the CC engine does, in 72-189 s). COLOR_331_1_21 goes to tier 5: MaxSAT solves it in 4300-4600 s on some seeds but its slower-mode 5-seed median is >= 5000 s (no-card seeds 0 and 2 over 5000 s; seed 0 solved after 6472 s), so tier 4 is empty after this re-tier. 2026-10-23 (main c914ef5 baseline, PR #142; PI): LP_442_68_10 tier2 -> tier1, BB_288_12_18 tier3 -> tier2, LP_714_100_16 tier4 -> tier3 (LP_1020_136_16 pending a 5-seed measurement). 2026-10-21 (main cbb0642, -frontier=auto; PI): BB_108_8_10, BB_144_12_12, LP_238_44_6 tier1 -> tier0; GB_144_12_12, LP_544_80_12, TN_200_10_10 tier3 -> tier2; LP_714_100_16 tier5 -> tier4 (optimization/baselines/SERVER.md). 2026-10-14 (main 398054e, GH-107; PI): BB_144_14_14 tier3 -> tier2 (72.7 s), BB_288_12_18 tier4 -> tier3 (919 s).
+**Tier assignment rule (PI 2026-10-12):** by main's time on yfclab2 (slower of no-card / card-mto, `optimization/baselines/SERVER.md`): < 1 s tier0, 1–10 s tier1, 10–100 s tier2, 100–2000 s tier3, 2000–5000 s tier4, > 5000 s or unsolved tier5. Re-check placements when main changes. 2026-11-04 (main 02879a93 baseline of all 75 codes, 5000 s panel limit for every code, `optimization/baselines/SERVER.md`; PI plan 2026-11-04): BB_288_12_18 tier2 -> tier1 (9.58 s), GB_144_12_12 tier2 -> tier1 (1.10 s), TN_144_2_13 tier3 -> tier2 (88.17 s), TN_250_10_15 tier4 -> tier3 (1283 s), LP_1020_136_16 tier5 -> tier3 (393 s); the 29 library codes placed by the same rule (tier0: COLOR_7_1_3, HGP_13_1_3, COLOR_19_1_5, BB_30_4_6, HGP_41_1_5, HGP_45_9_3, HGP_106_16_4, AMC_108_6_9, HGP_113_1_8, BB_120_8_12, AMC_216_6_12, HGP_306_36_5, AMC_330_6_14; tier1: COLOR_127_1_13, BB_186_10_14, HGP_265_1_12, AMC_438_6_17, HGP_610_64_6; tier2: BB_248_10_18, BB_340_16_18, AMC_522_6_20; tier3: BB_294_10_20, HGP_481_1_16, HGP_1261_121_6, HGP_2866_256_8; tier5: COLOR_331_1_21, COLOR_1027_1_33to37, HGP_1105_1_24, COLOR_2791_1_30to61). Codes that main solves within 6 h but over 5000 s stay in tier5 (TN_180_2_17, TN_216_4_18, BB_360_12_24, LP_1122_148_20); HGP_1105_1_24 is in tier5 because MaxSAT does not solve it in 6 h (the CC engine does, in 72-189 s). COLOR_331_1_21 goes to tier 5: MaxSAT solves it in 4300-4600 s on some seeds but its slower-mode 5-seed median is >= 5000 s (no-card seeds 0 and 2 over 5000 s; seed 0 solved after 6472 s), so tier 4 is empty after this re-tier. 2026-10-23 (main c914ef5 baseline, PR #142; PI): LP_442_68_10 tier2 -> tier1, BB_288_12_18 tier3 -> tier2, LP_714_100_16 tier4 -> tier3 (LP_1020_136_16 pending a 5-seed measurement). 2026-10-21 (main cbb0642, -frontier=auto; PI): BB_108_8_10, BB_144_12_12, LP_238_44_6 tier1 -> tier0; GB_144_12_12, LP_544_80_12, TN_200_10_10 tier3 -> tier2; LP_714_100_16 tier5 -> tier4 (optimization/baselines/SERVER.md). 2026-10-14 (main 398054e, GH-107; PI): BB_144_14_14 tier3 -> tier2 (72.7 s), BB_288_12_18 tier4 -> tier3 (919 s).
 
 **Timeouts.** tier4: 2000 s per instance on the Mac, 5000 s on yfclab2; tier5: 21600 s (6 h) per instance.
 tier0–tier3 are the filtering tiers of the
@@ -66,6 +66,25 @@ and [credit request and per-pass records](https://github.com/guluchen/DistQLDPC/
 
 | Date | Old | New | Note |
 |------|-----|-----|------|
+| 2026-11-07 | BB_1080_4_32to48 | BB_1080_4_34to48 | lb 34: CC lower bound from the GH-225 follow-up (dedicated server, up to 36 h; records `optimization/experiments/GH-225/lb-push/`). C = main 755e0c21's CC default (gated erasure); E = main 1239e9a2's CC default plus ungated `erasure`; same engine, not an independent second method; the last level is EMPTY in each run listed; ub unchanged (GH-146 witness; copy in `optimization/experiments/GH-146/witnesses/renamed-2026-11-07/`). C 40 586 s; E 44 390 s; level 32: 36.4G vs 7.1G nodes; X half under the verified dual map; odd levels skipped by the parity rule (every column of Hx and Hz has odd weight, re-checked in pure Python) |
+| 2026-11-07 | BB_756_16_32to34 | BB_756_16_34 | d = 34 (solved): lb 34 as for BB_1080 above: C 59 834 s; E 65 917 s; level 32: 55.2G vs 10.5G nodes; X half under the verified dual map; odd levels skipped by parity; both runs report `o 34`; the lower bound meets the GH-146 weight-34 witness, VALID in `VERIFY.txt` |
+| 2026-11-07 | BB_864_4_32to40 | BB_864_4_34to40 | lb 34 (as above): C 58 134 s; E 64 158 s; level 32: 51.3G vs 10.0G nodes; ub unchanged |
+| 2026-11-07 | COLOR_1027_1_31to37 | COLOR_1027_1_33to37 | lb 33 (as above): C 78 893 s; E with `-cc-roworder=index` 77 146 s (lb 32: 38 288 s and 37 190 s); level 32 EMPTY, X half only (Hx = Hz), 38.0G vs 36.3G nodes: the two runs differ in binary and erasure gating, not in row order; ub unchanged (family formula d = 37) |
+| 2026-11-07 | COLOR_2791_1_28to61 | COLOR_2791_1_30to61 | lb 30 (as above): C 42 701 s; C with `-cc-roworder=bfs` 79 340 s (lb 29: 20 415 s and 36 930 s); level 29 EMPTY, X half only, 33.5G vs 54.9G nodes. C went on to lb 31 at 88 551 s; that is one tree and stays PLAUSIBLE; ub unchanged (family formula d = 61) |
+| 2026-11-07 | TN_288_2_22to24 | TN_288_2_24 | d = 24 (solved): C 98 733 s; E 100 460 s (lb 23: 29 449 s and 30 037 s); level 23 EMPTY on both halves, X 11.6G / Z 8.4G vs X 7.1G / Z 4.6G nodes; both runs report `o 24`; the lower bound meets the GH-146 weight-24 witness, VALID in `VERIFY.txt` |
+| 2026-11-07 | TN_324_4_23to25 | TN_324_4_24to25 | lb 24 (as above): C 72 074 s; E 72 807 s; ub unchanged |
+| 2026-11-07 | TN_360_4_23to24 | TN_360_4_24 | d = 24 (solved): C 47 415 s; E 48 186 s; level 23 EMPTY on both halves, X 5.3G / Z 6.2G vs X 2.4G / Z 2.6G nodes; both runs report `o 24`; the lower bound meets the GH-146 weight-24 witness, VALID in `VERIFY.txt` |
+| 2026-11-07 | TN_396_4_23to26 | TN_396_4_24to26 | lb 24 (as above): C 47 489 s; E 47 923 s; ub unchanged |
+| 2026-11-07 | TN_432_8_23to29 | TN_432_8_25to29 | lb 25 (as above): C 87 063 s; E 89 519 s (lb 24: 27 189 s and 28 040 s; without erasure 38 938 s, 12 h); ub unchanged |
+| 2026-11-07 | TN_468_4_23to29 | TN_468_4_25to29 | lb 25 (as above): C 107 865 s; E 110 270 s; ub unchanged |
+| 2026-11-07 | TN_496_2_28to30 | TN_496_2_30 | d = 30 (solved): C 117 232 s; E 119 072 s (lb 29: 46 997 s and 47 826 s); level 29 EMPTY, X half under the verified XZ-dual map, 17.2G vs 8.1G nodes; both runs report `o 30`; the lower bound meets the GH-146 weight-30 witness, VALID in `VERIFY.txt` |
+| 2026-11-07 | TN_504_2_24to34 | TN_504_2_25to34 | lb 25 (as above): C 54 147 s; E 56 955 s; ub unchanged |
+| 2026-11-07 | TN_540_4_24to34 | TN_540_4_25to34 | lb 25 (as above): C 57 564 s; E 63 505 s; ub unchanged |
+| 2026-11-07 | TN_576_2_24to37 | TN_576_2_26to37 | lb 26 (as above): C 79 691 s; E 88 962 s (lb 25: 26 157 s and 29 372 s; without erasure 34 316 s, 12 h); ub unchanged |
+| 2026-11-07 | TN_612_2_24to40 | TN_612_2_26to40 | lb 26 (as above): C 97 484 s; E 107 037 s (lb 25: 32 844 s and 36 062 s; without erasure 42 105 s, 12 h); ub unchanged |
+| 2026-11-07 | TN_648_10_22to32 | TN_648_10_23to32 | lb 23 (as above): C 53 620 s; E 58 436 s; ub unchanged |
+| 2026-11-07 | TN_648_14_22to28 | TN_648_14_23to28 | lb 23 (as above): C 70 302 s; E 63 576 s; ub unchanged |
+| 2026-11-07 | TN_684_2_24to29 | TN_684_2_25to29 | lb 25 (as above): E with `-seed=1` 106 893 s and with `-seed=2` 105 414 s (relabelled qubits and rows); level 24 EMPTY on both halves, X 4.52G / Z 6.46G vs X 4.53G / Z 5.96G nodes; ub unchanged |
 | 2026-11-06 | TN_324_4_22to25 | TN_324_4_23to25 | lb 23: CC lower bound first reached by the GH-225 erasure arm (issue #225; binary 5a201fd = the GH-217 CC default plus `erasure`, yfclab2, 6 h, seed 0): lb 23 at 20 985 s. Reproduction (12 h, main 1239e9a2 with `-cc-rules=excl,pack,orbf,parity,leaf,overlap,charge,tiebreak,sparse,sparseov,sparsemrv,erasure`, ungated): relabelled qubits and rows `-seed=1` lb 23 at 30 309 s, `-seed=2` at 31 369 s; level 22 EMPTY on both halves; node counts differ from the original at every high level; same engine, not an independent second method; records `optimization/experiments/GH-225/lb-repro/`. Without erasure: lb 22 (6 h seed 0; 12 h `-seed=1`). ub unchanged (GH-146 witness; copy under the new stem in `optimization/experiments/GH-146/witnesses/renamed-2026-11-06/`) |
 | 2026-11-06 | TN_496_2_27to30 | TN_496_2_28to30 | lb 28 (as above): GH-225 erasure arm 17 996 s; GH-227 candidate (#227, main 755e0c21, gated erasure in the default) 19 683 s; reproduction with erasure `-seed=1` 20 699 s, `-seed=2` 19 095 s (`-cc-roworder=bfs` 19 856 s, the same tree as the GH-225 run); without erasure lb 27 at seed 0 (6 h) but lb 28 with `-seed=1` (24 919 s, 12 h); X half under the verified XZ-dual map; ub unchanged |
 | 2026-11-06 | TN_540_4_23to34 | TN_540_4_24to34 | lb 24 (as above): GH-225 erasure arm 19 168 s; GH-227 candidate 18 399 s; reproduction with erasure `-seed=1` 35 026 s, `-seed=2` 32 893 s; without erasure lb 23 at seed 0 (6 h), lb 24 with `-seed=1` (42 322 s, 12 h); both halves EMPTY; ub unchanged |
@@ -224,8 +243,9 @@ and [credit request and per-pass records](https://github.com/guluchen/DistQLDPC/
 ## Open items
 
 - **Former numeric tier5 names** (BB_756_16_34, BB_864_4_40, BB_1080_4_54, TN_180_2_17 ... TN_684_2_29): the old third field may be a literature value or an upstream index; it is kept in the rename history but not in the stem until DistQLDPC confirms the distance.
-- **TN_648_10_22to32:** d <= 32 (GH-146 verified witness; the bundled bases contain weight-52
-  nontrivial logicals); lb 22 from GH-157 (2026-10-31); the exact distance is unknown.
+- **TN_648_10_23to32:** d <= 32 (GH-146 verified witness; the bundled bases contain weight-52
+  nontrivial logicals); lb 22 from GH-157 (2026-10-31), lb 23 from the GH-225 follow-up (2026-11-07); the exact
+  distance is unknown.
 - **Upper-bound witnesses (`{lb}to{ub}` stems):** the upper bounds of the tier5 `{lb}to{ub}` stems renamed on
   2026-10-23 come from the GH-146 finder; one nontrivial logical per stem is stored as
   `optimization/experiments/GH-146/witnesses/renamed/<stem>.<X|Z>.w<ub>.txt` (0-based qubit indices) and can be
@@ -244,6 +264,12 @@ and [credit request and per-pass records](https://github.com/guluchen/DistQLDPC/
   Five TN stems were renamed on 2026-11-06 (TN_324_4_23to25, TN_496_2_28to30, TN_540_4_24to34, TN_648_14_22to28,
   TN_684_2_24to29); copies of their witnesses under the current stems are in
   `optimization/experiments/GH-146/witnesses/renamed-2026-11-06/` (same supports, `VERIFY.txt`: 5/5 VALID).
+  The 2026-11-06 copies of TN_324, TN_496, TN_540, TN_648_14 and TN_684 no longer resolve after the 2026-11-07 renames.
+  Seventeen more stems were renamed on 2026-11-07 (BB_1080_4_34to48, BB_756_16_34, BB_864_4_34to40, TN_288_2_24,
+  TN_324_4_24to25, TN_360_4_24, TN_396_4_24to26, TN_432_8_25to29, TN_468_4_25to29, TN_496_2_30, TN_504_2_25to34,
+  TN_540_4_25to34, TN_576_2_26to37, TN_612_2_26to40, TN_648_10_23to32, TN_648_14_23to28, TN_684_2_25to29); copies under
+  the current stems are in `optimization/experiments/GH-146/witnesses/renamed-2026-11-07/` (same supports, `VERIFY.txt`:
+  17/17 VALID).
 - **GH-152 lower bounds (2026-10-29):** the lower bounds of the six stems above come from the optional CC engine with
   `-cc-aut=on` (6 h, yfclab2) and were reproduced with a different search order (`mrv`, 12 h) on the same labels;
   this is the same algorithm, not an independent method. Record: issue #152 and