Search & Candidates from the CLI¶
Generate a Candidate Pool¶
rank42.candidate_generate creates a normal Rank Hunter Candidate Pool in the database.
sage -python -m rank42.candidate_generate \
--db rank42.db \
--plugin mestre_sextuple_rank11 \
--pool-name mestre-scan-01 \
--a-min -500 \
--a-max 500 \
--b-max 500 \
--engine sieve \
--top 1000
Useful options include --variant, --chart, staged prime bounds, sampled generation, cache control and JSONL export.
Write a raw candidate file¶
rank42.search runs the low-level family scorer and writes JSONL.
sage -python -m rank42.search \
--family my_family \
--a-min -1000 \
--a-max 1000 \
--b-max 500 \
--engine sieve \
--top 500 \
--out candidates.jsonl
Use candidate_generate when you want the candidates stored as a Rank Hunter Pool.
Geometry-first Family hunt¶
rank42.geometry_search can start a new command-line Family hunt and create its Campaign automatically.
sage -python -m rank42.geometry_search \
--project-root . \
--db rank42.db \
--plugin my_family \
--target-rank 20 \
--ratpoints vendor/ratpoints/ratpoints
It combines staged candidate scoring, exact family baseline/torsion work, bounded upper-bound gates, geometry, point search and certification.
Run a saved Pipeline¶
This is the main command behind current Pipeline execution.
Search an external record¶
sage -python -m rank42.record_search \
--db rank42.db \
--source icarm \
--id 302 \
--stages 1000,10000,100000
Search a quartic directly¶
sage -python -m rank42.quartic_search \
--db rank42.db \
--curve-id 123 \
--coefficients 1,0,-4,0,1 \
--height 100000 \
--timeout 300
For the persisted Quartics workbench modes, use rank42.quartic_workbench_runner.
Compatibility commands¶
These modules still exist so old jobs can resume, but they reject new standalone work unless their hidden compatibility/resume flag is used:
rank42.general_huntrank42.auto_searchrank42.fixed_curve_searchrank42.auto_analyze
For new work, use Candidate Pools, Pipelines, geometry_search, or the current Analysis commands.