{"id":"arxiv/2605.03542","url":"https://arxiv.org/abs/2605.03542","tags":["math.NA","cs.LG"],"score":0,"title":"Random test functions, $H^{-1}$ norm equivalence, and stochastic variational physics-informed neural networks","domain":"math.NA","source":"arxiv","authors":["Diego Marcondes"],"reviews":0,"subject":"math.NA","summary":"The dual norm characterisation of weak solutions of second-order linear elliptic partial differential equations is mathematically natural but computationally intractable: evaluating the $H^{-1}$ norm of the residual requires a supremum over an infinite-dimensional test space. We prove that the $H^{-1}$ norm of any functional is equivalent to its expected squared evaluation against a random test function whose probability distribution depends only on the domain. Crucially, realisations of this random test function have negative Sobolev regularity for $d \\geq 2$, yet this roughness is not an…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-05-05","reactions":{}}