{"id":"arxiv/2610.39148","url":"https://arxiv.org/abs/2610.39148","tags":["math.OC"],"score":0,"title":"Sharp approximation thresholds for power means of common-subspace volumes","domain":"math.OC","source":"arxiv","authors":["Alexey Peregudin"],"reviews":0,"subject":"math.OC","summary":"We study how well a single $k$-dimensional subspace can preserve volume from several positive semidefinite matrices at once. Each matrix is compressed to the subspace and scored by the geometric mean of the eigenvalues of that compression; the scores are aggregated by a weighted power mean of order $p \\le 1$. For every fixed $k$ and every fixed rational $p$ we determine the optimal approximation factor: it is achieved by a randomized polynomial-time algorithm, and every larger constant is NP-hard. Above a critical power ($p=-k/2$ over the reals and $p=-k$ over the complex numbers) the factor…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-29","reactions":{}}