Research map / Optimization / Optimizers
Coordinate-wise Power Method
Qi Lei, Kai Zhong, Inderjit S Dhillon
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use CPM instead of PM | power method · soc-Pokec | CPU time: 118 -> 26.5 | Improved |
| Use SGCD instead of PM | power method · soc-Pokec | CPU time: 118 -> 5.2 | Improved |
| Use SGCD instead of PM | power method · web-Stanford | CPU time: 8.15 -> 0.54 | Improved |
| Use SGCD instead of Lanczos | Lanczos method · soc-Pokec | CPU time: 14.2 -> 5.2 | Improved |
| Use CPM instead of PM | power method · com-Orkut | CPU time: 109.6 -> 31.5 | Improved |
| Use SGCD instead of CPM | CPM · ego-Gplus | CPU time: 0.57 -> 0.61 | Hurt |
| Use SGCD instead of PM | power method · ego-Gplus | CPU time: 0.99 -> 0.61 | Improved |
| Use VRPCA instead of PM | power method · com-Orkut | CPU time: 109.6 -> 189.7 | Hurt |
| Increase eigenvalue ratio lambda2/lambda1 toward 1 | CPM and SGCD vs PM · synthetic dense n=5000 | FLOPs | Improved |
| Increase matrix dimension n | CPM and SGCD · synthetic dense | convergence time | No clear effect |
| Use SGCD instead of Lanczos | Lanczos method · synthetic dense | CPU time | Improved |
| Use PM instead of CPM/SGCD when PM converges under 0.1 second | CPM and SGCD · synthetic dense | CPU time | No clear effect |
Closest papers by meaning
- Accelerating Greedy Coordinate Descent Methods (ICML 2018 · similarity 0.86)
- Approximate Steepest Coordinate Descent (ICML 2017 · similarity 0.85)
- Coordinate Descent Methods for DC Minimization: Optimality Conditions and Global Convergence (AAAI 2023 · similarity 0.85)
- Scale Invariant Power Iteration (JMLR 2023 · similarity 0.85)
- Even Faster Accelerated Coordinate Descent Using Non-Uniform Sampling (ICML 2016 · similarity 0.85)
- Stability-based Generalization Analysis of Randomized Coordinate Descent for Pairwise Learning (AAAI 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
