feat(plan-selection): group selection MILP over execution contexts as Stage 3's default search (#1095) by zzylol · Pull Request #1144 · ProjectASAP/ASAPPlanner

feat(plan-selection): group selection MILP over execution contexts behind ASAP_STAGE3_SEARCH=milp (#1095)

zzylol

committed

d99c1f9

fix(plan-selection): a shareable candidate reads a scan at its own window; window-independent scans keep their usage per window (#1095)

zzylol

committed

1b66859

fix(plan-selection): adopt main's Stage 1 exports and cost_inputs name after rebasing onto the Stages 0–2 milestone (#1095)

zzylol

committed

1566f6a

feat(plan-selection): selection MILP solved per independent part, LP gap diagnosis, tighter big-M bounds (#1095 §5 3a, §6.1) (#1155)

zzylol

committed

f3d83bb

perf(plan-selection): keep accuracy builds across pair checks and queries (#1095 §8) (#1156)

zzylol

committed

9f32a1f

feat(plan-selection): one ingestion context per refresh window in the selection MILP (#1095) (#1157)

zzylol

committed

73072e6

fix(plan-selection): use the tumbling-window names (#1163) in the group MILP after rebasing onto main

zzylol

committed

2766e42

feat(plan-selection): the group MILP is Stage 3's default search (#1095 C6)

zzylol

committed

079507f

fix(plan-selection): use the remaining tumbling-window names (#1163 part 2) in the group MILP

zzylol

committed

b7b4ce3

Navigation Menu

Repository navigation

feat(plan-selection): group selection MILP over execution contexts as Stage 3's default search (#1095) - #1144

feat(plan-selection): group selection MILP over execution contexts as Stage 3's default search (#1095)#1144
zzylol merged 9 commits into
mainfrom
feat/1095-group-milp

Commits on Nov 2, 2026

feat(plan-selection): group selection MILP over execution contexts behind ASAP_STAGE3_SEARCH=milp (#1095)

fix(plan-selection): a shareable candidate reads a scan at its own window; window-independent scans keep their usage per window (#1095)

fix(plan-selection): adopt main's Stage 1 exports and cost_inputs name after rebasing onto the Stages 0–2 milestone (#1095)

feat(plan-selection): selection MILP solved per independent part, LP gap diagnosis, tighter big-M bounds (#1095 §5 3a, §6.1) (#1155)

perf(plan-selection): keep accuracy builds across pair checks and queries (#1095 §8) (#1156)

feat(plan-selection): one ingestion context per refresh window in the selection MILP (#1095) (#1157)

fix(plan-selection): use the tumbling-window names (#1163) in the group MILP after rebasing onto main

feat(plan-selection): the group MILP is Stage 3's default search (#1095 C6)

fix(plan-selection): use the remaining tumbling-window names (#1163 part 2) in the group MILP