New method accelerates eigenvalue dataset generation using Chebyshev subspace filter

Researchers have developed a new method called the Sorting Chebyshev Subspace Filter (SCSF) to accelerate the generation of eigenvalue datasets for training neural eigenvalue methods. This approach leverages similarities between operators, a factor previously overlooked, to reduce redundant computations. SCSF utilizes fast Fourier transform sorting and a Chebyshev subspace filter to reuse eigenpairs from solved problems, achieving up to a 3.5 times speedup compared to traditional numerical solvers. AI

IMPACT Accelerates data generation for neural eigenvalue methods, potentially reducing training costs and time.

RANK_REASON Academic paper detailing a novel method for accelerating data generation for neural eigenvalue methods.

Read on arXiv cs.LG →

• arXiv
• Chebyshev subspace filter
• Hong Wang
• Sorting Chebyshev Subspace Filter

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →