nystrom_pcg_solver
Solve regularized PSD systems with randomized Nyström preconditioned conjugate gradients and log convergence evidence.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- nystrom_pcg_solver
- description
- Solve regularized PSD systems with randomized Nyström preconditioned conjugate gradients and log convergence evidence.
02 资源文件 2
- 脚本scripts/pcg.py5.0 KB
- 脚本tests/test_pcg.py795 B
03 SKILL.md 正文
Nyström PCG Solver
Use this skill to reproduce the paper's Algorithm 5.1 on a PSD linear system (A + mu I)x=b. It should call or mirror the generated Nyström factorization and preconditioner contracts rather than bypassing them.
Inputs
- Dense PSD matrix or matrix-vector product interface.
- Right-hand side
b. - Regularization
mu > 0. - Sketch rank
ell, tolerance, maximum iterations, and random seed.
Outputs
- Approximate solution.
- Residual history, relative residual history, and iteration count.
- Mechanism diagnostics: factorization executed, preconditioner applications, CG step count, and convergence flag.
Workflow
- Build Nyström factors with
randomized_nystrom_factorization. - Build an inverse preconditioner action with
nystrom_preconditioner. - Run left-preconditioned CG using preconditioned residual inner products.
- Optionally run ordinary CG as a baseline under the same tolerance.
- Save traces for downstream recovery evaluation.
Validation
Run:
python scripts/pcg.py --self-test
python tests/test_pcg.py
The tests use a deterministic ill-conditioned PSD system and require Nyström PCG to converge in fewer iterations than ordinary CG.
Limitations
This dense script is designed for bounded recovery experiments. For large paper-scale systems, replace dense arrays with matvec closures while preserving the same logging contract.
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).