Fix selective rerip owner dependency cycles

[0hmX]

What

Stacked on #167.

• Record directed failed-route → blocking-owner dependencies.
• Correct the owner inclusion check by walking those dependencies before the next selective rerip.
• Fall back to the solver's global rerip as soon as adding the current owner would close a dependency cycle.
• Clear learned dependency history after escalation so a fresh routing round does not inherit the stale cycle.
• Report the fallback as failed_owner_cycle in selective-rerip statistics.

Root cause

The bugreport87 graph is valid. One cramped port changes route 32 onto a Z0 path through that port. Later routes rotate blockers and eventually form the directed dependency cycle 51 → 48 → 49 → 50 → 51.

Previously, selective rerip only reacted when the same failed-route/owner pair repeated. Each edge in this larger cycle looked new, so the solver kept displacing another route without recognizing that it had returned to the same dependency chain.

The normal global-rerip solver completes the same graph. This fix preserves selective reripping for acyclic conflicts and escalates immediately when the complete owner dependency graph becomes cyclic.

Impact

The exact repro completes instead of running until the CI job timeout. No fixture, snapshot, or test changes are included in this stacked diff.

Validation

• bun run typecheck
• bunx biome format lib/selective-rerip-tiny-hyper-graph-solver.ts tests/solver/bugreport87-unused-port-repro.test.ts
• bun test tests/selective-rerip-tiny-hyper-graph-solver.test.ts tests/solver/port-chokepoint-repro.test.ts tests/solver/same-net-shared-bottleneck-repro.test.ts tests/solver/bugreport87-unused-port-repro.test.ts
• Result: 5 passed, 0 failed; exact repro completed locally in about 7.2 seconds.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
tiny-hypergraph	Ready	Preview	Aug 13, 2026 11:50pm

[tscircuitbot]

Benchmark This PR

Run benchmarks by commenting on this PR:

/benchmark [benchmark.sh args...]
/benchmark --dataset <hg07|srj18> --limit <n> --solver <core|poly> --families <default|default+deep|all> --concurrency <n|auto>

Everything after /benchmark is forwarded directly to ./benchmark.sh.
Every command runs current main and the PR head sequentially on the same Blacksmith machine, then reports their results side by side.

Examples:

• /benchmark -> all committed completed SRJ18 Pipeline7 cases, core solver
• /benchmark --limit 4 -> first four committed Pipeline7 cases
• /benchmark --dataset srj18 --concurrency auto -> srj18 samples with runner CPU-count concurrency
• /benchmark --sample sample003 -> one committed SRJ18 Pipeline7 case
• /benchmark --dataset hg07 --limit 20 --solver poly -> first 20 hg07 samples with the poly solver
• /benchmark --dataset hg07 --limit 40 --families default+deep -> first 40 hg07 samples with deeper candidate families

Any PR whose title contains [BENCHMARK TEST] will automatically run the benchmark workflow on PR updates.

[0hmX]

/benchmark

[tscircuitbot]

Tiny Hypergraph Same Machine Benchmark Results

Main c60c552 and PR e7ba41d ran sequentially on blacksmith-01kzyp7prtp3v1hgf1d2fbekdy-32vcpu with identical arguments.

Run	Dataset	Solver	Families	Success	Routes	Final Cost	Avg Hops	P50	P95
Main	srj18	core	default	100.0%	100.0%	1.739	5.786	2.938s	13.398s
PR	srj18	core	default	100.0%	100.0%	1.739	5.786	2.912s	14.050s

Main Branch Details

Metric	Value
Dataset	srj18
Source	dataset-srj18 committed Pipeline7 tiny-hypergraph inputs
Revision	source dataset-srj18@c0aad90256a95256fcac814f9f7da81a82a2fdea; tscircuit-autorouter@ac87ba2740c450c226ef84a5bbcef7fb31369297
Sample selection	all cases completed in autorouter run 31336663867
Solver	core
Families	default
Samples	8
Success rate	100.0%
Solved-route completion rate	100.0%
Improved rate	0.0%
Zero final max region cost rate	0.0%
Total completion time	36.525s
Avg baseline max region cost	1.739
Avg final max region cost	1.739
Avg max region delta	0.000
Avg candidate count	0.000
Avg generated candidate count	0.000
Avg duplicate candidate count	0.000
Avg iterations	329223.500
Avg route hops	5.786
Routes requiring rip	0
Avg solveGraph time	3.931s
Avg section search time	0.000s
Avg optimizeSection time	0.001s
Avg duration	4.566s
P50 duration	2.938s
P95 duration	13.398s

Sample	Status	Routes	Baseline	Final	Delta	Hops	Rips	Attempts	Iterations	Duration	Error
sample001	success	205/205	1.440	1.440	0.000	5.693	0	0	221738	2.938s
sample003	success	145/145	1.543	1.543	0.000	6.910	0	0	108848	1.664s
sample005	success	172/172	0.766	0.766	0.000	5.785	0	0	132205	1.664s
sample007	success	263/263	0.542	0.542	0.000	5.148	0	0	308095	5.737s
sample008	success	361/361	3.391	3.391	0.000	6.673	0	0	1197441	13.398s
sample009	success	256/256	1.492	1.492	0.000	5.500	0	0	155158	2.823s
sample010	success	310/310	1.278	1.278	0.000	4.739	0	0	141829	2.963s
sample011	success	289/289	3.464	3.464	0.000	5.841	0	0	368474	5.339s

PR Details

Metric	Value
Dataset	srj18
Source	dataset-srj18 committed Pipeline7 tiny-hypergraph inputs
Revision	source dataset-srj18@c0aad90256a95256fcac814f9f7da81a82a2fdea; tscircuit-autorouter@ac87ba2740c450c226ef84a5bbcef7fb31369297
Sample selection	all cases completed in autorouter run 31336663867
Solver	core
Families	default
Samples	8
Success rate	100.0%
Solved-route completion rate	100.0%
Improved rate	0.0%
Zero final max region cost rate	0.0%
Total completion time	37.510s
Avg baseline max region cost	1.739
Avg final max region cost	1.739
Avg max region delta	0.000
Avg candidate count	0.000
Avg generated candidate count	0.000
Avg duplicate candidate count	0.000
Avg iterations	329223.500
Avg route hops	5.786
Routes requiring rip	0
Avg solveGraph time	4.061s
Avg section search time	0.000s
Avg optimizeSection time	0.001s
Avg duration	4.689s
P50 duration	2.912s
P95 duration	14.050s

Sample	Status	Routes	Baseline	Final	Delta	Hops	Rips	Attempts	Iterations	Duration	Error
sample001	success	205/205	1.440	1.440	0.000	5.693	0	0	221738	2.670s
sample003	success	145/145	1.543	1.543	0.000	6.910	0	0	108848	1.840s
sample005	success	172/172	0.766	0.766	0.000	5.785	0	0	132205	1.678s
sample007	success	263/263	0.542	0.542	0.000	5.148	0	0	308095	6.025s
sample008	success	361/361	3.391	3.391	0.000	6.673	0	0	1197441	14.050s
sample009	success	256/256	1.492	1.492	0.000	5.500	0	0	155158	2.912s
sample010	success	310/310	1.278	1.278	0.000	4.739	0	0	141829	3.121s
sample011	success	289/289	3.464	3.464	0.000	5.841	0	0	368474	5.214s

Workflow: View run
Artifact: https://github.com/tscircuit/tiny-hypergraph/actions/runs/31752820717