{"title":"On learning Gaussian multi-index models with gradient flow part I: General properties and two-timescale learning","authors":"Alberto Bietti, Joan Bruna, Loucas Pillaud-Vivien","doi":"10.1002/cpa.70006","DOIUrl":"10.1002/cpa.70006","url":null,"abstract":"<p>We study gradient flow on the multi-index regression problem for high-dimensional Gaussian data. Multi-index functions consist of a composition of an unknown low-rank linear projection and an arbitrary unknown, low-dimensional link function. As such, they constitute a natural template for feature learning in neural networks. We consider a two-timescale algorithm, whereby the low-dimensional link function is learnt with a non-parametric model infinitely faster than the subspace parametrizing the low-rank projection. By appropriately exploiting the matrix semigroup structure arising over the subspace correlation matrices, we establish global convergence of the resulting Grassmannian gradient flow dynamics, and provide a quantitative description of its associated “saddle-to-saddle” dynamics. Notably, the timescales associated with each saddle can be explicitly characterized in terms of an appropriate Hermite decomposition of the target link function.</p>","PeriodicalId":10601,"journal":{"name":"Communications on Pure and Applied Mathematics","volume":"78 12","pages":"2354-2435"},"PeriodicalIF":2.7,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144629546","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Quasi-invariance of Gaussian measures for the \u0000 \u0000 \u0000 3\u0000 d\u0000 \u0000 $3d$\u0000 energy critical nonlinear Schrödinger equation","authors":"Chenmin Sun, Nikolay Tzvetkov","doi":"10.1002/cpa.70001","DOIUrl":"10.1002/cpa.70001","url":null,"abstract":"<p>We consider the <span></span><math>\u0000 <semantics>\u0000 <mrow>\u0000 <mn>3</mn>\u0000 <mi>d</mi>\u0000 </mrow>\u0000 <annotation>$3d$</annotation>\u0000 </semantics></math> energy critical nonlinear Schrödinger equation with data distributed according to the Gaussian measure with covariance operator <span></span><math>\u0000 <semantics>\u0000 <msup>\u0000 <mrow>\u0000 <mo>(</mo>\u0000 <mn>1</mn>\u0000 <mo>−</mo>\u0000 <mi>Δ</mi>\u0000 <mo>)</mo>\u0000 </mrow>\u0000 <mrow>\u0000 <mo>−</mo>\u0000 <mi>s</mi>\u0000 </mrow>\u0000 </msup>\u0000 <annotation>$(1-Delta)^{-s}$</annotation>\u0000 </semantics></math>, where <span></span><math>\u0000 <semantics>\u0000 <mi>Δ</mi>\u0000 <annotation>$Delta$</annotation>\u0000 </semantics></math> is the Laplace operator and <span></span><math>\u0000 <semantics>\u0000 <mi>s</mi>\u0000 <annotation>$s$</annotation>\u0000 </semantics></math> is sufficiently large. We prove that the flow sends full measure sets to full measure sets. We also discuss some simple applications. This extends a previous result by Planchon-Visciglia and the second author from <span></span><math>\u0000 <semantics>\u0000 <mrow>\u0000 <mn>1</mn>\u0000 <mi>d</mi>\u0000 </mrow>\u0000 <annotation>$1d$</annotation>\u0000 </semantics></math> to higher dimensions.</p>","PeriodicalId":10601,"journal":{"name":"Communications on Pure and Applied Mathematics","volume":"78 12","pages":"2305-2353"},"PeriodicalIF":2.7,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/cpa.70001","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144500756","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}