Reproduce index

Every result on the wiki that ships a reproduce command, in one table: what the command does, whether it re-runs the search or only re-checks a stored board, and roughly what it costs to run. Click a column head to reorder.

40 commands

					Pipeline
A sound colour-counting prune for break-tolerant search	seeded	search	25 s cap per A/B arm on the phase 1 grid, 300 s per arm on the certificate rows	0.056	—
Blackwood's solver, decoded and run here	heavy	search	60-120 s per run	0.033	—
CAS	seeded	search	baseline arm 45 s × 8 frames; CAS arm deterministic per-shell beam over 20 frames (the source's per-shell MIP took 25 to 45 s per frame)	0.10	cas-annular
Constraint immediacy: every fill order pays the same 480	exact	search	—	—	constraint-immediacy
CSP presets, measured	seeded	search	60 s	2.0	—
Entropy and the area law	exact	search	—	—	entropy-area-law
Forbidden patterns	exact	search	—	—	forbidden-patterns
GAUNTLET	stochastic	verify only	~5 min ALNS lift per build (exploratory run, not the standardized single-core bench)	0.67	record-boards
KEYRING	stochastic	verify only	not logged (exploratory run, not the standardized single-core bench)	—	record-boards
Linking the clues early does not help, it hurts	exact	search	—	—	clue-corridors
LODESTONE	stochastic	verify only	not logged (exploratory run, not the standardized single-core bench)	—	record-boards
Making a beam producer 10x better: the answer is width	seeded	search	12 s saturated width-law passes; 3 s per A/B arm (192 arms in the powered SMC comparison)	0.0033	beam-width-smc
McGavin's C backtracker: the throughput story, built here	heavy	search	30 s per run	0.0083	—
No forced moves	exact	search	—	—	no-forced-moves
PALIMPSEST	stochastic	verify only	~30 min × 6 seeds (exploratory run, not the standardized single-core bench)	4.0	record-boards
Piece theft, where solvers die	exact	search	—	—	piece-theft
PRIOR	stochastic	verify only	not logged (exploratory run, not the standardized single-core bench)	—	record-boards
Reading a row's future in its leftover pieces	seeded	search	about 2 minutes for the full 120-seed run, single core	0.27	—
Reference numbers	exact	search	—	—	subgrid-placement-counts
Single-core benchmark	seeded	search	60 s	2.5	—
STAGED	stochastic	verify only	not logged (exploratory run, not the standardized single-core bench)	—	record-boards
The 470 wall: a phase boundary, not an engineering limit	exact	search	—	—	the-470-wall
The board as a codeword	exact	search	—	—	permutation-code-wall
The border balance	exact	search	—	—	border-mismatch-share
The DFS study	seeded	search	60 s	0.017	—
The fluid frame	exact	search	well under a minute for the 500-frame main run on one core (; a preflight at the default scale completed in about 7 s)	0.016	—
The frame is not the basin: a different border does not open a different high board	seeded	search	—	—	frame-is-not-the-basin
The hard region you cannot design away	seeded	search	—	—	irreducible-hard-region
The hint study	seeded	search	8 s per run	0.0022	—
The JIT backtracker: portable Rust that ties hand-tuned C on hard boards	heavy	search	6–30 s per benchmark run	0.0083	—
The piece set is extreme on every measured axis	exact	search	—	—	adversarial-piece-set
The rare colors live on the frame	exact	search	—	—	rare-color-geography
The repair study	seeded	search	60 s	0.017	—
The scaling ladder	seeded	search	12 s single-core budget per (solver, instance, seed) run; 4 rungs × 8 seeds × 2 solvers = 64 runs	0.0033	scaling-ladder
The tail as its own exact problem	seeded	search	—	—	exact-tail-endgame
Tuned to the hardness peak	exact	search	—	—	phase-transition
Verhaard reimplementation	seeded	search	120 s	0.033	—
Where you place the hints beats how many	seeded	search	—	—	—
Why a faster computer doesn't help	exact	search	—	—	prune-vs-speed
Why basin-hopping looks impossible	exact	verify only	—	—	sigma-cycles