{"id":"arxiv/2607.03692","url":"https://arxiv.org/abs/2607.03692","tags":["cs.LG","math.NA","math.ST"],"score":0,"title":"PIEFS: Physics-Informed Eigenfunction Features with Learnable Scaling","domain":"cs.LG","source":"arxiv","authors":["Varvara Nazarenko","Timur Lidzhiev","Alexander Tarakanov"],"reviews":0,"subject":"cs.LG","summary":"Spectral methods are widely used to construct representations from the geometry of data, but they often rely on a fixed kernel, graph Laplacian, or manually selected feature scaling. We propose Physics-Informed Eigenfunction Features with Learnable Scaling (PIEFS), a supervised neural representation-learning framework with a spectral inductive bias, based on a modified Dirichlet energy. In PIEFS, scalar coordinate maps are trained under empirical Gram orthogonality, a supervised linear readout, and a Dirichlet penalty in which the input gradient is transformed by a learnable metric…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-07-04","reactions":{}}