{"id":58,"date":"2010-06-02T08:38:49","date_gmt":"2010-06-02T06:38:49","guid":{"rendered":"http:\/\/djalil.chafai.net\/blog\/?p=58"},"modified":"2019-11-10T19:19:25","modified_gmt":"2019-11-10T18:19:25","slug":"localization-eigenvectors-heavy-tailed-random-matrices","status":"publish","type":"post","link":"https:\/\/djalil.chafai.net\/blog\/2010\/06\/02\/localization-eigenvectors-heavy-tailed-random-matrices\/","title":{"rendered":"Localization of eigenvectors: heavy tailed random matrices"},"content":{"rendered":"<p style=\"text-align: justify;\">We measure the localization of an eigenvector $v\\in\\mathbb{R}^n$, $v\\neq0$, by the norm ratio $\\sqrt{n}\\left\\Vert v\\right\\Vert_\\infty \\left\\Vert v\\right\\Vert_2^{-1}$.<\/p>\n<p style=\"text-align: justify;\">In what follows, the matrix $X$ is $n\\times n$, with i.i.d. entries and $X_{11}=\\varepsilon (U^{-1\/\\alpha}-1)$ where $\\varepsilon$ and $U$ are independent with $\\varepsilon$ symmetric Rademacher (random sign) and $U$ uniform on $[0,1]$. For every $t&gt;0$, we have $$\\mathbb{P}(|X_{11}|&gt;t)=(1+t)^{-\\alpha}.$$<\/p>\n<p style=\"text-align: justify;\">Each of the following plots corresponds to a single realization of the matrix. In each plot, each small red cirlce corresponds to a couple (eigenvalue, eigenvector). A logarithmic scale is used for the eigenvalues, while the localization is used for the eigenvectors.\u00a0 The notion of ``bulk of the spectrum'' is unclear when $\\alpha&lt;2$. Beware that we should not 100% trust numerics when dealing with heavy tailed random matrices (no truncation here).<\/p>\n<p><strong><span style=\"text-decoration: underline;\">S<\/span><span style=\"text-decoration: underline;\">ymmetric model: eigenvectors localization versus eigenvalues<\/span><\/strong> <a href=\"\/blog\/wp-content\/uploads\/heavy\/heavy.m\">(Octave code) <\/a><\/p>\n<p><img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_0.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_1.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_1.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_2.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_2.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_4.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavy_16.00_2000.jpg\" alt=\"\" \/><\/p>\n<p><strong><span style=\"text-decoration: underline;\">Full iid m<\/span><span style=\"text-decoration: underline;\">odel: eigenvectors localization versus singular values<\/span><\/strong> (<a href=\"\/blog\/wp-content\/uploads\/heavy\/heavysing.m\">Octave code)<br \/>\n<\/a><\/p>\n<p><img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_0.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_1.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_1.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_2.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_2.50_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_4.00_2000.jpg\" alt=\"\" \/><br \/>\n<img src=\"\/blog\/wp-content\/uploads\/heavy\/heavysing_16.00_2000.jpg\" alt=\"\" \/><\/p>\n<p style=\"text-align: justify;\"><strong>Note:<\/strong> This post is motivated by an animated coffee break discussion with <a title=\"Alice Guionnet\" href=\"\/scripts\/search.php\/?q=Alice+Guionnet\">Alice Guionnet<\/a> and <a title=\"Charles Bordenave\" href=\"\/scripts\/search.php\/?q=Charles+Bordenave\">Charles Bordenave<\/a> during a <a title=\"Conference on random matrices ANR GranMa Paris, June 1-4 (2010)\" href=\"\/scripts\/search.php\/?q=Conference+on+Random+Matrices+GranMA+Paris+2010+June+1+4\">conference on random matrices<\/a> (Paris, June 1-4, 2010). You may forge numerous conjectures from these computer simulations. Another interesting problem is the limiting behavior of the eigenvalues or singular values spacings when the dimension goes to infinity, and its dependency over the tail index $\\alpha$.<\/p>\n<p><strong>Recent addition to this post:<\/strong> <a href=\"\/scripts\/search.php\/?q=Sandrine+P\u00e9ch\u00e9\">Sandrine P\u00e9ch\u00e9<\/a> pointed out that conjectures have been made by the physicists Bouchaud and Cizeau in their paper <a href=\"http:\/\/link.aps.org\/doi\/10.1103\/PhysRevE.50.1810\">Theory of L\u00e9vy matrices published in Phys. Rev. E 50, 1810\u20131822 (1994)<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>We measure the localization of an eigenvector $v\\in\\mathbb{R}^n$, $v\\neq0$, by the norm ratio $\\sqrt{n}\\left\\Vert v\\right\\Vert_\\infty \\left\\Vert v\\right\\Vert_2^{-1}$. In what follows, the matrix $X$ is $n\\times&#8230;<\/p>\n<div class=\"more-link-wrapper\"><a class=\"more-link\" href=\"https:\/\/djalil.chafai.net\/blog\/2010\/06\/02\/localization-eigenvectors-heavy-tailed-random-matrices\/\">Continue reading<span class=\"screen-reader-text\">Localization of eigenvectors: heavy tailed random matrices<\/span><\/a><\/div>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"iawp_total_views":182},"categories":[1],"tags":[],"_links":{"self":[{"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/posts\/58"}],"collection":[{"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/comments?post=58"}],"version-history":[{"count":6,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/posts\/58\/revisions"}],"predecessor-version":[{"id":11762,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/posts\/58\/revisions\/11762"}],"wp:attachment":[{"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/media?parent=58"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/categories?post=58"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/djalil.chafai.net\/blog\/wp-json\/wp\/v2\/tags?post=58"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}