physo
Use PhySO for symbolic regression, class symbolic regression, toolkit-level expression workflows, and benchmark problem loaders.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- physo
- license
- MIT
- metadata
- { "disco-role": "operating" }
- read_when
- Use when the user names `physo`, `PhySO`, symbolic regression, ClassSR, toolkit expression helpers, or benchmark problem loaders.
- description
- Use PhySO for symbolic regression, class symbolic regression, toolkit-level expression workflows, and benchmark problem loaders.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/install-and-smoke.md1.1 KB
- 参考文档references/repo-provenance.md1.5 KB
- 参考文档references/repo-routing-metadata.json418 B
- 参考文档references/troubleshooting.md2.0 KB
- 脚本scripts/physo_skill_doctor.py4.2 KB
03 SKILL.md 正文
PhySO
Use this skill when working with the physo package, the scientific Python symbolic-regression library.
Install and Import Check
Run scripts/physo_skill_doctor.py to confirm the installed package, key public exports, and the CPU scientific stack used by the verified baseline. The checked baseline was PhySO 1.2.0 on CPU; this skill does not claim CUDA verification. Import-time LaTeX warnings are expected when optional display tools are missing.
Route by Task
- One dataset with
X,y, units,y_weights,candidate_wrapper, run presets, logging, or Pareto inspection: usesub-skills/sr/SKILL.md. - Multiple realizations with shared class constants or realization-specific constants: use
sub-skills/class-sr/SKILL.md. - Prefix expressions, libraries, tokens,
Program/VectPrograms, random sampling, constant optimization, display, or result reloads: usesub-skills/toolkit/SKILL.md. - Feynman/Class benchmark problems, sample generation, metadata, or symbolic equivalence checks: use
sub-skills/benchmarks/SKILL.md.
Shared References
references/install-and-smoke.md: install/import check and smoke commands.references/troubleshooting.md: cross-cutting import, display, backend, and routing failures.references/repo-provenance.md: source version, dirty-state snapshot, and evidence paths.references/repo-routing-metadata.json: router placement metadata for compatible agents.
Common Starting Points
- Start with the doctor script when the environment or package import is uncertain.
- Route to a sub-skill before drafting commands or code for a user workflow.
- Use the sub-skill smoke scripts for route-specific tiny checks after the environment is known to be usable.
- If the user asks for benchmark reproduction, training-scale sweeps, or maintainer jobfiles, stop at the boundary and keep this runtime skill focused on day-to-day package workflows.
Boundaries
This skill is self-contained. It does not depend on opening the source checkout at runtime, and it does not cover repo maintenance, release automation, or long benchmark campaigns.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).