Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Solving high-dimensional parabolic PDEs using the tensor train format
Lorenz Richter, Leon Sallandt, Nikolas Nüsken
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 tensor trains instead of DenseNet | HJB d=100, implicit scheme · HJB d=100 | relative error: 1.71 \text{e}^{-3} -> 5.90 \text{e}^{-5} | Improved |
| Use tensor trains instead of DenseNet | HJB d=100, implicit scheme · HJB d=100 | computation time: 44712 -> 41 | Improved |
| Use the implicit scheme instead of the explicit scheme | TT solver · HJB d=100 | relative error: 3.17 \text{e}^{-4} -> 5.90 \text{e}^{-5} | Improved |
| Use the explicit scheme instead of the implicit scheme | TT solver · HJB d=100 | computation time: 41 -> 25 | Improved |
| Use tensor trains instead of DenseNet | HJB double well, non-interacting d=50 · HJB double well, non-interacting d=50 | computation time: 1987 -> 16 | Improved |
| Use tensor trains instead of DenseNet | HJB interacting double well d=20 · HJB interacting double well d=20 | relative error: 4.24\text{e}^{-3} -> 1.52\text{e}^{-3} | Improved |
| Use tensor trains instead of DenseNet | CIR d=100 · CIR d=100 | PDE loss: 7.57\text{e}^{-3} -> 5.06\text{e}^{-4} | Improved |
| Increase the sample size from K=1000 to K=20000 (ablation) | TT implicit · PDE with unbounded solution d=10 | relative error: 1.22\text{e}^{-1} -> 6.11\text{e}^{-3} | Improved |
| Increase polynomial degree from 0 to 1 (ablation) | TT · CIR d=100 | PDE loss: 9.04\text{e}^{-2} -> 7.80\text{e}^{-4} | Improved |
| Increase polynomial degree beyond 1 (ablation) | TT · CIR d=100 | PDE loss: 7.80\text{e}^{-4} -> 1.05\text{e}^{-3}; 7.80\text{e}^{-4} -> 5.06\text{e}^{-4} | No clear effect |
| Increase polynomial degree from 0 to 1 (ablation) | TT · HJB d=100 | relative error: 5.90\text{e}^{-5} -> 4.99\text{e}^{-5} | No clear effect |
Closest papers by meaning
- P$^2$C$^2$Net: PDE-Preserved Coarse Correction Network for efficient prediction of spatiotemporal dynamics (NeurIPS 2024 · similarity 0.87)
- Learning a Neural Solver for Parametric PDEs to Enhance Physics-Informed Methods (ICLR 2025 · similarity 0.87)
- UGrid: An Efficient-And-Rigorous Neural Multigrid Solver for Linear PDEs (ICML 2024 · similarity 0.86)
- MC^2: Monte Carlo Correction for Fast Elliptic PDE Solving (NeurIPS 2026 · similarity 0.86)
- Boosting Generalization in Parametric PDE Neural Solvers through Adaptive Conditioning (NeurIPS 2024 · similarity 0.86)
- Toward Efficient Kernel-Based Solvers for Nonlinear PDEs (ICML 2025 · 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.
