Current browse context:
math.ST
Change to browse by:
References & Citations
Mathematics > Statistics Theory
Title: How Many Modes Can a Mixture of Gaussians with Uniformly Bounded Means Have?
(Submitted on 4 May 2020 (v1), last revised 22 Oct 2020 (this version, v3))
Abstract: We show, by an explicit construction, that a mixture of univariate Gaussian densities with variance $1$ and means in $[-A,A]$ can have $\Omega(A^2)$ modes. This disproves a recent conjecture of Dytso, Yagli, Poor and Shamai \cite{DYPS20} who showed that such a mixture can have at most $O(A^{2})$ modes and surmised that the upper bound could be improved to $O(A)$. Our result holds even if an additional variance constraint is imposed on the mixing distribution. Extending the result to higher dimensions, we exhibit a mixture of Gaussians in $\mathbb{R}^{d}$, with identity covariances and means inside $[-A,A]^{d}$, that has $\Omega(A^{2d})$ modes.
Submission history
From: Navin Kashyap [view email][v1] Mon, 4 May 2020 15:49:36 GMT (126kb)
[v2] Wed, 6 May 2020 19:52:33 GMT (126kb)
[v3] Thu, 22 Oct 2020 15:19:33 GMT (129kb)
Link back to: arXiv, form interface, contact.