Research map / Optimization / Nonconvex and smooth optimization
Sobolev Gradient Ascent for Optimal Transport: Barycenter Optimization and Convergence Analysis
Kaheon Kim, Bohan Zhou, Changbo Zhu, Xiaohui Chen
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 SGA instead of WDHA | Wasserstein barycenter solver · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | barycenter functional value (x10^3): 3.940 -> 3.917 | Improved |
| use SGA instead of WDHA | Wasserstein barycenter solver · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | running time (s): 625 -> 298 | Improved |
| use SGA instead of CWB | Wasserstein barycenter solver · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | running time (s): 1346 -> 298 | Improved |
| use SGA instead of DSB | Wasserstein barycenter solver · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | running time (s): 3643 -> 298 | Improved |
| use SGA instead of CWB | Wasserstein barycenter solver · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | barycenter functional value (x10^3): 4.202 -> 3.917 | Improved |
| replace constant step size 0.1 with annealing step 0.5/sqrt(t) (ablation) | WDHA · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | barycenter functional value (x10^3): 3.940 -> 4.012 | Hurt |
| replace constant step size 0.1 with annealing step 0.5/sqrt(t) (ablation) | SGA · 2D synthetic barycenter, 1024x1024 grid, w=(1/3,1/4,1/6,1/4) | barycenter functional value (x10^3): 5.923 -> 5.917 | No clear effect |
| replace constant step size 0.1 with annealing step 0.5/sqrt(t) (ablation) | WDHA · 2D synthetic barycenter, 1024x1024 grid, w=(2/3,0,0,1/3) | running time (s): 625 -> 497 | Improved |
| use SGA instead of WDHA | 3D ball-to-cube barycenter, 200x200x200 grid · 3D ball to cube interpolation | convergence: WDHA diverges while SGA gives 1.443, 1.577, 1.455 (x10^-3) | Improved |
Closest papers by meaning
- Optimal Transport Barycenter via Nonconvex-Concave Minimax Optimization (ICML 2025 · similarity 0.91)
- Computing Optimal Transport Maps and Wasserstein Barycenters Using Conditional Normalizing Flows (ICML 2025 · similarity 0.89)
- A Truncated Newton Method for Optimal Transport (ICLR 2025 · similarity 0.87)
- Energy-Guided Continuous Entropic Barycenter Estimation for General Costs (NeurIPS 2024 · similarity 0.86)
- A Provably Convergent and Practical Algorithm for Gromov–Wasserstein Optimal Transport (NeurIPS 2026 · similarity 0.86)
- On the Convergence of Projected Bures-Wasserstein Gradient Descent under Euclidean Strong Convexity (ICML 2024 · similarity 0.86)
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.
