Pith. sign in

REVIEW 3 cited by

Quantifying Training Difficulty and Accelerating Convergence in Neural Network-Based PDE Solvers

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.06308 v1 pith:NWJCISXB submitted 2024-10-08 math.NA cs.LGcs.NA

classification math.NAcs.LGcs.NA
keywords trainingconvergenceneuraldifficultyeffectiveinitializationnetwork-basedrank
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Neural network-based methods have emerged as powerful tools for solving partial differential equations (PDEs) in scientific and engineering applications, particularly when handling complex domains or incorporating empirical data. These methods leverage neural networks as basis functions to approximate PDE solutions. However, training such networks can be challenging, often resulting in limited accuracy. In this paper, we investigate the training dynamics of neural network-based PDE solvers with a focus on the impact of initialization techniques. We assess training difficulty by analyzing the eigenvalue distribution of the kernel and apply the concept of effective rank to quantify this difficulty, where a larger effective rank correlates with faster convergence of the training error. Building upon this, we discover through theoretical analysis and numerical experiments that two initialization techniques, partition of unity (PoU) and variance scaling (VS), enhance the effective rank, thereby accelerating the convergence of training error. Furthermore, comprehensive experiments using popular PDE-solving frameworks, such as PINN, Deep Ritz, and the operator learning framework DeepOnet, confirm that these initialization techniques consistently speed up convergence, in line with our theoretical findings.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Spectral connvergece of random feature method in one dimension

    math.NA 2025-07 conditional novelty 7.0 of 10

    For one-dimensional second-order elliptic PDEs, the Random Feature Method achieves spectral convergence for smooth solutions, but its feature matrix is exponentially ill-conditioned.

  2. Complex Physics-Informed Neural Network

    cs.LG 2025-02 conditional novelty 4.0 of 10

    A single-layer PINN using the trainable Cauchy activation function matches or beats larger multi-layer PINN baselines on smooth analytic PDE benchmarks.

  3. Learn Singularly Perturbed Solutions via Homotopy Dynamics

    cs.LG 2025-02 conditional novelty 4.0 of 10

    A homotopy continuation method that starts training at a large PDE parameter and tracks the solution to small values improves neural network solvers for singularly perturbed problems.

Pith tools