randomized_nystrom_factorization
Compute a stable randomized Nyström PSD low-rank eigendecomposition from a dense PSD matrix or matrix-vector product interface.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- randomized_nystrom_factorization
- description
- Compute a stable randomized Nyström PSD low-rank eigendecomposition from a dense PSD matrix or matrix-vector product interface.
02 资源文件 2
- 脚本scripts/nystrom.py3.2 KB
- 脚本tests/test_nystrom.py973 B
03 SKILL.md 正文
Randomized Nyström Factorization
Use this skill when you need the paper's Algorithm 2.1 component: a reproducible low-rank PSD approximation A_nys = U diag(lambda_hat) U.T for a symmetric positive-semidefinite matrix. Do not use it as a full linear solver; downstream solver skills own preconditioning and PCG.
Inputs
- PSD matrix
Aas a dense JSON/NumPy array, or a Python callable that appliesAto a matrix. - Dimension
nand sketch rankell, with1 <= ell <= n. - Random seed for reproducibility.
Outputs
U: orthonormal basis for the Nyström range.lambda_hat: nonnegative approximate eigenvalues.- Diagnostics: sketch rank, returned rank, stability shift, matvec count, and Cholesky retry count.
Workflow
- Validate matrix dimensions and rank.
- Draw a standard Gaussian sketch and orthonormalize it with thin QR.
- Form
Y = A @ Omega; this is the only matrix interaction required by the algorithm. - Add a tiny stability shift
nu * Omegabefore Cholesky. - Factor
Omega.T @ Y_shift, solve by the Cholesky factor, and take a thin SVD. - Return
Uand squared singular values minus the stability shift, clipped at zero for roundoff.
Validation
Run:
python scripts/nystrom.py --self-test
python tests/test_nystrom.py
The tests verify shapes, orthonormality, nonnegative eigenvalues, deterministic seeding, and improved approximation with larger rank.
Limitations
The routine assumes PSD-compatible input. It is intentionally small and dense-friendly for recovery experiments; large production runs should provide an efficient matvec closure and avoid forming dense matrices.
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).