<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	>

<channel>
	<title>Felix&#039; Math Place &#187; Algebraic Geometry</title>
	<atom:link href="http://math.fontein.de/category/algebraic-geometry/feed/" rel="self" type="application/rss+xml" />
	<link>http://math.fontein.de</link>
	<description>Focussed on, but not limited to Computational Number Theory</description>
	<lastBuildDate>Sat, 30 Jul 2011 12:35:49 +0000</lastBuildDate>
	<language>en</language>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.1</generator>
		<item>
		<title>A Strange Inequality.</title>
		<link>http://math.fontein.de/2010/12/09/a-strange-inequality/</link>
		<comments>http://math.fontein.de/2010/12/09/a-strange-inequality/#comments</comments>
		<pubDate>Thu, 09 Dec 2010 07:49:09 +0000</pubDate>
		<dc:creator>Felix Fontein</dc:creator>
				<category><![CDATA[Algebraic Geometry]]></category>
		<category><![CDATA[Elementary Number Theory]]></category>
		<category><![CDATA[function field]]></category>
		<category><![CDATA[genus]]></category>
		<category><![CDATA[inequality]]></category>

		<guid isPermaLink="false">http://math.fontein.de/?p=795</guid>
		<description><![CDATA[We show how to prove a number theoretic inequality, originating from geometry, using an elementary approach.]]></description>
			<content:encoded><![CDATA[<p>Today, while trying to prove a result for a preprint I&#8217;m working on, I got so frustrated that I played around with something else from that paper. I got an idea to get rid of something non-nice, namely I had an asymptotic bound <img src='http://math.fontein.de/wp-content/latex/c03/c03dae496af78ae0d7f5d7cc779b601c-T-000000-0.png' alt='O(g/n + 1)' title='O(g/n + 1)' class='latex-inline' /> for <img src='http://math.fontein.de/wp-content/latex/b7c/b7c63b141e32e1736fbfa26fb2d405ad-T-000000-0.png' alt='g, n \to \infty' title='g, n \to \infty' class='latex-inline' />, and wanted to drop the <img src='http://math.fontein.de/wp-content/latex/701/7015440590904d38e35615ff0672d21c-T-000000-0.png' alt='+1' title='+1' class='latex-inline' /> if possible. Of course, this is only possibe if <img src='http://math.fontein.de/wp-content/latex/17a/17a76855f03c4ee4d4122e82376e2aa6-T-000000-0.png' alt='g &gt; 0' title='g &gt; 0' class='latex-inline' /> and if <img src='http://math.fontein.de/wp-content/latex/d9d/d9dfc3404f34e53e8f492198434ae18c-T-000000-0.png' alt='g \ge C n' title='g \ge C n' class='latex-inline' /> for some constant <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' />.</p>
<p>So I began working this out to see how far I could get. It is rather easy to translate the whole problem into a question on some integers. Namely, assume that <img src='http://math.fontein.de/wp-content/latex/b54/b5483f2f1af0ee56acb54ef669929a57-T-000000-0.png' alt='n &gt; 1' title='n &gt; 1' class='latex-inline' /> is an integer, and we are given <img src='http://math.fontein.de/wp-content/latex/03c/03c7c0ace395d80182db07ae2c30f034-T-000000-0.png' alt='s' title='s' class='latex-inline' /> integers <img src='http://math.fontein.de/wp-content/latex/881/881c549174b656994b7492316b4f4900-T-000000-0.png' alt='1 \le n_1, \dots, n_s &lt; n' title='1 \le n_1, \dots, n_s &lt; n' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/e8d/e8dabba18a39c862da1225e7e54dee11-T-000000-0.png' alt='\gcd(n_1, \dots, n_s, n) = 1' title='\gcd(n_1, \dots, n_s, n) = 1' class='latex-inline' />. Define the quantity <img src='http://math.fontein.de/wp-content/latex/197/1977a70d1e39c9d9b0018e0939ffb752-T-000000-0.png' alt='g(n_1, \dots, n_s)' title='g(n_1, \dots, n_s)' class='latex-inline' /> as <img src='http://math.fontein.de/wp-content/latex/938/938299f5aca33a839fb4273fcac5d1ce-T-000000-0.png' alt='\displaystyle  \frac{\biggl(n - \gcd\biggl(n, \sum\limits_{i=1}^s n_i\biggr)\biggr) + \sum\limits_{i=1}^s (n - \gcd(n, n_i)) - 2 (n - 1)}{2}. ' title='\displaystyle  \frac{\biggl(n - \gcd\biggl(n, \sum\limits_{i=1}^s n_i\biggr)\biggr) + \sum\limits_{i=1}^s (n - \gcd(n, n_i)) - 2 (n - 1)}{2}. ' class='latex-displaystyle' /> One can show that this is always a non-negative integer. Now I claim that <img src='http://math.fontein.de/wp-content/latex/197/1977a70d1e39c9d9b0018e0939ffb752-T-000000-0.png' alt='g(n_1, \dots, n_s)' title='g(n_1, \dots, n_s)' class='latex-inline' /> is either 0, or <img src='http://math.fontein.de/wp-content/latex/aae/aae53e252031465f6c3dfa3966e3096d-T-000000-0.png' alt='\ge \frac{1}{6} n' title='\ge \frac{1}{6} n' class='latex-inline' />.</p>
<p>Where does this strange expression comes from? Consider the function field <img src='http://math.fontein.de/wp-content/latex/b41/b4144b8b6a1cfe3f4e313dcd86a9d330-T-000000-0.png' alt='K = \C(x, y)' title='K = \C(x, y)' class='latex-inline' /> over <img src='http://math.fontein.de/wp-content/latex/ee7/ee77cd72573eec25fba471d91befc2d2-T-000000-0.png' alt='\C' title='\C' class='latex-inline' /> defined by <img src='http://math.fontein.de/wp-content/latex/e70/e706807c511225a2a0be21d53ed20a05-T-000000-0.png' alt='\displaystyle  y^n = \prod_{i=1}^s (x - i)^{n_i}. ' title='\displaystyle  y^n = \prod_{i=1}^s (x - i)^{n_i}. ' class='latex-displaystyle' /> (In fact, one can do this over any field whose characteristic is coprime to <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' />, and which has at least <img src='http://math.fontein.de/wp-content/latex/03c/03c7c0ace395d80182db07ae2c30f034-T-000000-0.png' alt='s' title='s' class='latex-inline' /> elements. Moreover, over an algebraically closed field <img src='http://math.fontein.de/wp-content/latex/8ce/8ce4b16b22b58894aa86c421e8759df3-T-000000-0.png' alt='k' title='k' class='latex-inline' />, any function field extension <img src='http://math.fontein.de/wp-content/latex/1d6/1d6230fbaa4fffd690c4805e7d1efbaf-T-000000-0.png' alt='K / k(x)' title='K / k(x)' class='latex-inline' /> which is cyclic of order <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' />, where <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' /> is coprime to the characteristic of <img src='http://math.fontein.de/wp-content/latex/8ce/8ce4b16b22b58894aa86c421e8759df3-T-000000-0.png' alt='k' title='k' class='latex-inline' />, can be written in this form, since this is a <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Kummer_extension#Kummer_extensions">Kummer extension</a>.) One can show that this equation is irreducibe if and only if <img src='http://math.fontein.de/wp-content/latex/e8d/e8dabba18a39c862da1225e7e54dee11-T-000000-0.png' alt='\gcd(n_1, \dots, n_s, n) = 1' title='\gcd(n_1, \dots, n_s, n) = 1' class='latex-inline' />, and that the genus of this function field is given by <img src='http://math.fontein.de/wp-content/latex/197/1977a70d1e39c9d9b0018e0939ffb752-T-000000-0.png' alt='g(n_1, \dots, n_s)' title='g(n_1, \dots, n_s)' class='latex-inline' />. This also explains why we must have that <img src='http://math.fontein.de/wp-content/latex/a7f/a7fa944d04dcc2cbc9f6dba56863c3dd-T-000000-0.png' alt='g(n_1, \dots, n_s) \ge 0' title='g(n_1, \dots, n_s) \ge 0' class='latex-inline' />, since the genus is always a nonnegative integer.</p>
<p>Now we have a struggle: is it easier to show the claim using some Elementary Number Theory, or using some (advanced?) Algebraic Geometry (considering the geometric side) or Algebraic Number Theory (considering the function field side)? I don&#8217;t have any idea how to use the latter two to prove this, but I found an elementary proof.</p>
<p>First, note that if <img src='http://math.fontein.de/wp-content/latex/4c2/4c220818b042a75ab5d79e2d3d0f9bf2-T-000000-0.png' alt='s \ge 5' title='s \ge 5' class='latex-inline' />, or in case <img src='http://math.fontein.de/wp-content/latex/789/7898a97eab0982e253d5d8855e58cc87-T-000000-0.png' alt='n \ge 4' title='n \ge 4' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/a76/a7696b090f9c3147b0158aa26f6ad396-T-000000-0.png' alt='n \nmid n_1 + \dots + n_s' title='n \nmid n_1 + \dots + n_s' class='latex-inline' />, at least five of the <img src='http://math.fontein.de/wp-content/latex/fe2/fe20c1fe862c25552af640f5b3d65715-T-000000-0.png' alt='n - \gcd(n, \bullet)' title='n - \gcd(n, \bullet)' class='latex-inline' /> terms must be <img src='http://math.fontein.de/wp-content/latex/75b/75b48e308c7119284f86e8f4f73f90c1-T-000000-0.png' alt='\ge \frac{n}{2}' title='\ge \frac{n}{2}' class='latex-inline' /> since <img src='http://math.fontein.de/wp-content/latex/719/71976219fb1234a4f2ba6bef7a046183-T-000000-0.png' alt='\bullet' title='\bullet' class='latex-inline' /> is not divisible by <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/a3a/a3a13130eaca19bf284e3fdb25696749-T-000000-0.png' alt='g(n_1, \dots, n_s) \ge \frac{5 \cdot \tfrac{1}{2} n - 2 (n - 1)}{2} \ge \tfrac{1}{2} n' title='g(n_1, \dots, n_s) \ge \frac{5 \cdot \tfrac{1}{2} n - 2 (n - 1)}{2} \ge \tfrac{1}{2} n' class='latex-inline' />.</p>
<p>Moreover, if <img src='http://math.fontein.de/wp-content/latex/2ee/2ee4828bc9ebde1d178afd88d92426ff-T-000000-0.png' alt='n \mid n_1 + \dots + n_s' title='n \mid n_1 + \dots + n_s' class='latex-inline' />, we have <img src='http://math.fontein.de/wp-content/latex/bef/befe65f35d6c3d1d97c9b050549b585f-T-000000-0.png' alt='n_s \equiv -(n_1 + \dots + n_{s-1}) \pmod{n}' title='n_s \equiv -(n_1 + \dots + n_{s-1}) \pmod{n}' class='latex-inline' />, and therefore <img src='http://math.fontein.de/wp-content/latex/380/38072c76407185a5c2a5b7d1a42d698b-T-000000-0.png' alt='1 = \gcd(n_1, \dots, n_s, n) = \gcd(n_1, \dots, n_{s-1}, n)' title='1 = \gcd(n_1, \dots, n_s, n) = \gcd(n_1, \dots, n_{s-1}, n)' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/90f/90f72bbe857cfe35a85dd34015d423c8-T-000000-0.png' alt='\gcd(n, n_s) = \gcd(n, n_1 + \dots + n_{s-1})' title='\gcd(n, n_s) = \gcd(n, n_1 + \dots + n_{s-1})' class='latex-inline' />. Hence, <img src='http://math.fontein.de/wp-content/latex/a46/a4600c489f9bbe0f4f8abbd17a56eb4e-T-000000-0.png' alt='\displaystyle  g(n_1, \dots, n_s) = g(n_1, \dots, n_{s-1}). ' title='\displaystyle  g(n_1, \dots, n_s) = g(n_1, \dots, n_{s-1}). ' class='latex-displaystyle' /> Therefore, it suffices to consider the case <img src='http://math.fontein.de/wp-content/latex/a76/a7696b090f9c3147b0158aa26f6ad396-T-000000-0.png' alt='n \nmid n_1 + \dots + n_s' title='n \nmid n_1 + \dots + n_s' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/c87/c8713199d43292ddf15db62bc955b3cc-T-000000-0.png' alt='s \le 3' title='s \le 3' class='latex-inline' />.</p>
<p>For <img src='http://math.fontein.de/wp-content/latex/1ea/1ea6df72c31304a0064f61704390c3be-T-000000-0.png' alt='s = 1' title='s = 1' class='latex-inline' />, note that <img src='http://math.fontein.de/wp-content/latex/d84/d84451a2fba1beef27e6efcc78094b86-T-000000-0.png' alt='\gcd(n, n_1) = 1' title='\gcd(n, n_1) = 1' class='latex-inline' /> implies <img src='http://math.fontein.de/wp-content/latex/8bf/8bfbb56c201eb20b43709d19701d84d8-T-000000-0.png' alt='g(n_1) = 0' title='g(n_1) = 0' class='latex-inline' />. Hence, there is nothing to show.</p>
<p>For <img src='http://math.fontein.de/wp-content/latex/185/1854056973826159fd101c7787ad422e-T-000000-0.png' alt='s = 2' title='s = 2' class='latex-inline' />, let me consider three cases.</p>
<ol>
<li><img src='http://math.fontein.de/wp-content/latex/c12/c1224e21dadedfeeb5f0bb4eacc3da9f-T-000000-0.png' alt='\gcd(n, n_1) = n/p' title='\gcd(n, n_1) = n/p' class='latex-inline' /> for some prime <img src='http://math.fontein.de/wp-content/latex/838/83878c91171338902e0fe0fb97a8c47a-T-000000-0.png' alt='p' title='p' class='latex-inline' />;</li>
<li><img src='http://math.fontein.de/wp-content/latex/373/3730c34c847868e2caa71dc76f5330c6-T-000000-0.png' alt='\gcd(n, n_1) = n/p^2' title='\gcd(n, n_1) = n/p^2' class='latex-inline' /> for some prime <img src='http://math.fontein.de/wp-content/latex/838/83878c91171338902e0fe0fb97a8c47a-T-000000-0.png' alt='p' title='p' class='latex-inline' />;</li>
<li><img src='http://math.fontein.de/wp-content/latex/390/39002610ea5f3e8ed7825299c6b1d678-T-000000-0.png' alt='\gcd(n, n_1) = n/(p q)' title='\gcd(n, n_1) = n/(p q)' class='latex-inline' /> for two distinct prims <img src='http://math.fontein.de/wp-content/latex/7ce/7cea42b00df9e2ff6c5c53c857e00d02-T-000000-0.png' alt='p, q' title='p, q' class='latex-inline' />.</li>
</ol>
<p>In all three cases, one can show by analyzing several cases that the claim is true. Thus, we are only interested in the cases where no <img src='http://math.fontein.de/wp-content/latex/91c/91c2aa226eeae37b2321a26615a523c2-T-000000-0.png' alt='\gcd(n, n_i)' title='\gcd(n, n_i)' class='latex-inline' /> is of the form <img src='http://math.fontein.de/wp-content/latex/95a/95a06b23e22389ef825be575ba4def65-T-000000-0.png' alt='n/p' title='n/p' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/8db/8dbada5159159e9d72275aa3c530515d-T-000000-0.png' alt='n/p^2' title='n/p^2' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/910/9100de2a5f994871fc3f9e9719571270-T-000000-0.png' alt='n/(pq)' title='n/(pq)' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/736/736174af4f8c108d41e5d5990746da09-T-000000-0.png' alt='\gcd(n, n_i) \le \frac{1}{12}' title='\gcd(n, n_i) \le \frac{1}{12}' class='latex-inline' /> for all <img src='http://math.fontein.de/wp-content/latex/865/865c0c0b4ab0e063e5caa3387c1a8741-T-000000-0.png' alt='i' title='i' class='latex-inline' />. Since not all <img src='http://math.fontein.de/wp-content/latex/91c/91c2aa226eeae37b2321a26615a523c2-T-000000-0.png' alt='\gcd(n, n_i)' title='\gcd(n, n_i)' class='latex-inline' />&#8216;s cannot be the same &ndash; this would contradict <img src='http://math.fontein.de/wp-content/latex/b6c/b6c2c8ea24f1c20974b04323af790c84-T-000000-0.png' alt='\displaystyle  1 = \gcd(n_1, \dots, n_s, n) = \gcd(\gcd(n_1, n), \dots, \gcd(n_s, n)) ' title='\displaystyle  1 = \gcd(n_1, \dots, n_s, n) = \gcd(\gcd(n_1, n), \dots, \gcd(n_s, n)) ' class='latex-displaystyle' /> &ndash; we must have <img src='http://math.fontein.de/wp-content/latex/f32/f32fad09eeddee713b596736c1e74502-T-000000-0.png' alt=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; n - (\tfrac{1}{2} + \tfrac{1}{12} + \tfrac{1}{13}) n + 2 \ge \tfrac{1}{3} n + 2. ' title=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; n - (\tfrac{1}{2} + \tfrac{1}{12} + \tfrac{1}{13}) n + 2 \ge \tfrac{1}{3} n + 2. ' class='latex-displaystyle' /></p>
<p>Let me demonstrate how to do <img src='http://math.fontein.de/wp-content/latex/c12/c1224e21dadedfeeb5f0bb4eacc3da9f-T-000000-0.png' alt='\gcd(n, n_1) = n/p' title='\gcd(n, n_1) = n/p' class='latex-inline' />. (In fact, this case suffices if one does not wants the constant <img src='http://math.fontein.de/wp-content/latex/6c2/6c2e3e2e98abd1fd9a66519db9da8d90-T-000000-0.png' alt='\frac{1}{6}' title='\frac{1}{6}' class='latex-inline' />, but one is happy with the constant <img src='http://math.fontein.de/wp-content/latex/813/81366272f331b85cff0eba0f402d6918-T-000000-0.png' alt='\frac{1}{24}' title='\frac{1}{24}' class='latex-inline' />, since <img src='http://math.fontein.de/wp-content/latex/77f/77fb9e086d0a27450c7006aa43820972-T-000000-0.png' alt='1 - \frac{1}{4} - \frac{1}{6} - \frac{1}{2} = \frac{1}{12}' title='1 - \frac{1}{4} - \frac{1}{6} - \frac{1}{2} = \frac{1}{12}' class='latex-inline' />.) Assume that <img src='http://math.fontein.de/wp-content/latex/c12/c1224e21dadedfeeb5f0bb4eacc3da9f-T-000000-0.png' alt='\gcd(n, n_1) = n/p' title='\gcd(n, n_1) = n/p' class='latex-inline' /> for some prime <img src='http://math.fontein.de/wp-content/latex/838/83878c91171338902e0fe0fb97a8c47a-T-000000-0.png' alt='p' title='p' class='latex-inline' />. Since <img src='http://math.fontein.de/wp-content/latex/eb7/eb75f5176fa5de643d6ed66caeb75542-T-000000-0.png' alt='\displaystyle  1 = \gcd(n_1, n_2, n) = \gcd(\gcd(n, n_1), \gcd(n, n_2) = \gcd(n/p, \gcd(n, n_2)), ' title='\displaystyle  1 = \gcd(n_1, n_2, n) = \gcd(\gcd(n, n_1), \gcd(n, n_2) = \gcd(n/p, \gcd(n, n_2)), ' class='latex-displaystyle' /> we must have <img src='http://math.fontein.de/wp-content/latex/d8c/d8c09c0246703c271d7222fe23395699-T-000000-0.png' alt='\gcd(n, n_2) \mid p' title='\gcd(n, n_2) \mid p' class='latex-inline' />, with <img src='http://math.fontein.de/wp-content/latex/cc6/cc607d2522bb176a2f84dece870444fc-T-000000-0.png' alt='p^2 \nmid n' title='p^2 \nmid n' class='latex-inline' /> in case <img src='http://math.fontein.de/wp-content/latex/74d/74d9dfa5aef9d20c66bb4b5e8bebc25f-T-000000-0.png' alt='\gcd(n, n_2) = p' title='\gcd(n, n_2) = p' class='latex-inline' />. In both cases, we have <img src='http://math.fontein.de/wp-content/latex/ecb/ecbb571b4a3155f8b1339ac73088235d-T-000000-0.png' alt='\gcd(n/p, n_1 + n_2) = 1' title='\gcd(n/p, n_1 + n_2) = 1' class='latex-inline' />.</p>
<p>In case <img src='http://math.fontein.de/wp-content/latex/425/4259ffebcc4c178bab6810e6e59153ac-T-000000-0.png' alt='\gcd(n, n_2) = 1' title='\gcd(n, n_2) = 1' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/ecb/ecbb571b4a3155f8b1339ac73088235d-T-000000-0.png' alt='\gcd(n/p, n_1 + n_2) = 1' title='\gcd(n/p, n_1 + n_2) = 1' class='latex-inline' /> implies <img src='http://math.fontein.de/wp-content/latex/f91/f91a286536e6f0a389d72b866470667d-T-000000-0.png' alt='\gcd(n, n_1 + n_2) \mid p' title='\gcd(n, n_1 + n_2) \mid p' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/c4c/c4c5a434096b0cf3d9fdf75bb08c0895-T-000000-0.png' alt='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) \le n/p + 1 + p' title='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) \le n/p + 1 + p' class='latex-inline' />. In case <img src='http://math.fontein.de/wp-content/latex/74d/74d9dfa5aef9d20c66bb4b5e8bebc25f-T-000000-0.png' alt='\gcd(n, n_2) = p' title='\gcd(n, n_2) = p' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/ecb/ecbb571b4a3155f8b1339ac73088235d-T-000000-0.png' alt='\gcd(n/p, n_1 + n_2) = 1' title='\gcd(n/p, n_1 + n_2) = 1' class='latex-inline' /> implies <img src='http://math.fontein.de/wp-content/latex/437/4378105528f2810d5c615286c6a6275e-T-000000-0.png' alt='\gcd(n, n_1 + n_2) = 1' title='\gcd(n, n_1 + n_2) = 1' class='latex-inline' />. Therefore, we also have <img src='http://math.fontein.de/wp-content/latex/164/16439577b0e2334cc9544e044ba2545c-T-000000-0.png' alt='\displaystyle  \gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) \le n/p + p + 1. ' title='\displaystyle  \gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) \le n/p + p + 1. ' class='latex-displaystyle' /> This yields <img src='http://math.fontein.de/wp-content/latex/a58/a58266faafdc1670bcd365c26b4807a7-T-000000-0.png' alt=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; n - (n/p + p + 1) + 2 = n \tfrac{p - 1}{p} - p + 1 \overset{!}{\ge} \tfrac{1}{3} n. ' title=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; n - (n/p + p + 1) + 2 = n \tfrac{p - 1}{p} - p + 1 \overset{!}{\ge} \tfrac{1}{3} n. ' class='latex-displaystyle' /> The latter inequality is true if and only if <img src='http://math.fontein.de/wp-content/latex/59a/59a681c731588f3f9dfb1fdd74265980-T-000000-0.png' alt='\displaystyle  n \ge \frac{3 p (p - 1)}{2 p - 3}. ' title='\displaystyle  n \ge \frac{3 p (p - 1)}{2 p - 3}. ' class='latex-displaystyle' /> If we write <img src='http://math.fontein.de/wp-content/latex/da0/da0488488b626e03ecae4160ec7febee-T-000000-0.png' alt='n = p k' title='n = p k' class='latex-inline' />, this translates to <img src='http://math.fontein.de/wp-content/latex/7ce/7ceeb8035e8d89e78ff427d80b34dcc3-T-000000-0.png' alt='\displaystyle  p \ge 3 \frac{k - 1}{2 k - 3}; ' title='\displaystyle  p \ge 3 \frac{k - 1}{2 k - 3}; ' class='latex-displaystyle' /> if <img src='http://math.fontein.de/wp-content/latex/974/974feba567c3b9e14531067c8cbe5f2e-T-000000-0.png' alt='k \ge 2' title='k \ge 2' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/e2a/e2ada8000109a5eae216e20d39e077d0-T-000000-0.png' alt='3 \frac{k - 1}{2 k - 3} \le 3' title='3 \frac{k - 1}{2 k - 3} \le 3' class='latex-inline' />, and if <img src='http://math.fontein.de/wp-content/latex/a9a/a9ad1161fcfd3561aed2d5b45e231183-T-000000-0.png' alt='k \ge 3' title='k \ge 3' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/5e7/5e78913cac382205f9ca35c1ffd26498-T-000000-0.png' alt='3 \frac{k - 1}{2 k - 3} \le 2' title='3 \frac{k - 1}{2 k - 3} \le 2' class='latex-inline' />. Hence, the only cases were the above argument does not work are <img src='http://math.fontein.de/wp-content/latex/d48/d48bada8350b099a4f53577237a5e427-T-000000-0.png' alt='n = p' title='n = p' class='latex-inline' /> (<img src='http://math.fontein.de/wp-content/latex/5c6/5c6f937eacd3732196734c56ec527fa4-T-000000-0.png' alt='k = 1' title='k = 1' class='latex-inline' />) and <img src='http://math.fontein.de/wp-content/latex/cff/cff6a0454d14f999df05e954187d3683-T-000000-0.png' alt='n = 4' title='n = 4' class='latex-inline' /> (<img src='http://math.fontein.de/wp-content/latex/2d4/2d4dcf10084570378af72846cd24eee5-T-000000-0.png' alt='k = 2' title='k = 2' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/56e/56e0ae3293d209a8a47139f7591104db-T-000000-0.png' alt='p = 2' title='p = 2' class='latex-inline' />).</p>
<p>In case <img src='http://math.fontein.de/wp-content/latex/cff/cff6a0454d14f999df05e954187d3683-T-000000-0.png' alt='n = 4' title='n = 4' class='latex-inline' />, we must have <img src='http://math.fontein.de/wp-content/latex/6e2/6e21db0fe346f983c9c83bec3c905189-T-000000-0.png' alt='n_1 = 2' title='n_1 = 2' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/b15/b157fccf962378fac6f0096aebb75e02-T-000000-0.png' alt='n_2 \in \{ 1, 3 \}' title='n_2 \in \{ 1, 3 \}' class='latex-inline' />. In that case, <img src='http://math.fontein.de/wp-content/latex/84b/84b0256f0c4b393327aee73995768842-T-000000-0.png' alt='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) = 2 + 1 + 1 = 4' title='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) = 2 + 1 + 1 = 4' class='latex-inline' />, whence <img src='http://math.fontein.de/wp-content/latex/827/827194b330b1dd26fd0ac9531c00153f-T-000000-0.png' alt=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}={} &amp; (4 - 1) + (4 - 2) + (4 - 1) - 2 (4 - 1) = 2 \ge \tfrac{1}{3} \cdot 4. ' title=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}={} &amp; (4 - 1) + (4 - 2) + (4 - 1) - 2 (4 - 1) = 2 \ge \tfrac{1}{3} \cdot 4. ' class='latex-displaystyle' /> In case <img src='http://math.fontein.de/wp-content/latex/d48/d48bada8350b099a4f53577237a5e427-T-000000-0.png' alt='n = p' title='n = p' class='latex-inline' />, since we do by assumption <img src='http://math.fontein.de/wp-content/latex/639/63951355d27d08fe90f7c60871a2a5e5-T-000000-0.png' alt='p \nmid n_1 + n_2' title='p \nmid n_1 + n_2' class='latex-inline' />, we obtain <img src='http://math.fontein.de/wp-content/latex/ddf/ddfc4adf4076387abce8c64a7b3a47a5-T-000000-0.png' alt='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) = 1 + 1 + 1 = 3' title='\gcd(n, n_1) + \gcd(n, n_2) + \gcd(n, n_1 + n_2) = 1 + 1 + 1 = 3' class='latex-inline' />, whence <img src='http://math.fontein.de/wp-content/latex/b41/b41f6645a31135a9374db50c6a397d17-T-000000-0.png' alt='\displaystyle  (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) = n - 1 \ge \tfrac{1}{3} n ' title='\displaystyle  (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^2 (n - \gcd(n, n_i)) - 2 (n - 1) = n - 1 \ge \tfrac{1}{3} n ' class='latex-displaystyle' /> since <img src='http://math.fontein.de/wp-content/latex/e4a/e4a8215090ec712898282c6445419778-T-000000-0.png' alt='n \ge 3/2' title='n \ge 3/2' class='latex-inline' />.</p>
<p>The cases <img src='http://math.fontein.de/wp-content/latex/ac5/ac599d4ebcf60269a0b30a6e9efdc806-T-000000-0.png' alt='\gcd(n, n_1) = n / p^2' title='\gcd(n, n_1) = n / p^2' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/c54/c54179cd629058757ea08f8f3dc94e01-T-000000-0.png' alt='\gcd(n, n_1) = n / (p q)' title='\gcd(n, n_1) = n / (p q)' class='latex-inline' /> are proven analogously, with a few more case distinctions.</p>
<p>So we are left with the case <img src='http://math.fontein.de/wp-content/latex/01f/01f63e9d9053cc0b801a9f5f0f7d6c35-T-000000-0.png' alt='s = 3' title='s = 3' class='latex-inline' />. Here, one can proceed in a similar, painful way. Or one increases the constant to <img src='http://math.fontein.de/wp-content/latex/3f7/3f745284407e4a2b198f3db1bab93436-T-000000-0.png' alt='\frac{1}{12}' title='\frac{1}{12}' class='latex-inline' />, since we know that not all <img src='http://math.fontein.de/wp-content/latex/91c/91c2aa226eeae37b2321a26615a523c2-T-000000-0.png' alt='\gcd(n, n_i)' title='\gcd(n, n_i)' class='latex-inline' />&#8216;s can be <img src='http://math.fontein.de/wp-content/latex/a2f/a2f070a31330443ceb0dcf352fe50035-T-000000-0.png' alt='n/2' title='n/2' class='latex-inline' />, whence one is at least <img src='http://math.fontein.de/wp-content/latex/075/075ecf3629810c0a2743ce9c2d4a73b5-T-000000-0.png' alt='\le n/3' title='\le n/3' class='latex-inline' />. Hence, <img src='http://math.fontein.de/wp-content/latex/411/41108718cb891d427a0b8a95632a4df5-T-000000-0.png' alt=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^3 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; 4 n - 3 \cdot \tfrac{1}{2} n - \tfrac{1}{3} n - 2 n + 2 = \tfrac{1}{6} n + 2, ' title=' &amp; (n - \gcd(n, n_1 + n_2)) + \sum_{i=1}^3 (n - \gcd(n, n_i)) - 2 (n - 1) \\ {}\ge{} &amp; 4 n - 3 \cdot \tfrac{1}{2} n - \tfrac{1}{3} n - 2 n + 2 = \tfrac{1}{6} n + 2, ' class='latex-displaystyle' /> which yields the claim.</p>
<p>To sum everything up, we showed the following theorem:</p>
<blockquote class='theorem'><div class='theoremtitle'>Theorem.</div> <div class='theoremmain'><br />
Let <img src='http://math.fontein.de/wp-content/latex/b54/b5483f2f1af0ee56acb54ef669929a57-T-000000-0.png' alt='n &gt; 1' title='n &gt; 1' class='latex-inline' /> be an integer, <img src='http://math.fontein.de/wp-content/latex/9bf/9bf498c35722326495df6e4edebdfc9a-T-000000-0.png' alt='s \ge 1' title='s \ge 1' class='latex-inline' />, and <img src='http://math.fontein.de/wp-content/latex/e0a/e0a83675b10dc4a413d0969ec643b39f-T-000000-0.png' alt='n_1, \dots, n_s \in \{ 1, \dots, n - 1 \}' title='n_1, \dots, n_s \in \{ 1, \dots, n - 1 \}' class='latex-inline' /> satisfy <img src='http://math.fontein.de/wp-content/latex/e8d/e8dabba18a39c862da1225e7e54dee11-T-000000-0.png' alt='\gcd(n_1, \dots, n_s, n) = 1' title='\gcd(n_1, \dots, n_s, n) = 1' class='latex-inline' />. Then <img src='http://math.fontein.de/wp-content/latex/197/1977a70d1e39c9d9b0018e0939ffb752-T-000000-0.png' alt='g(n_1, \dots, n_s)' title='g(n_1, \dots, n_s)' class='latex-inline' />, defined as <img src='http://math.fontein.de/wp-content/latex/1b8/1b870bd65841635577fc443decda9547-T-000000-0.png' alt='\displaystyle  \frac{\biggl(n - \gcd\biggl(n, \sum\limits_{i=1}^s n_i\biggr)\biggr) + \sum\limits_{i=1}^s (n - \gcd(n, n_i)) - 2 (n - 1)}{2}, ' title='\displaystyle  \frac{\biggl(n - \gcd\biggl(n, \sum\limits_{i=1}^s n_i\biggr)\biggr) + \sum\limits_{i=1}^s (n - \gcd(n, n_i)) - 2 (n - 1)}{2}, ' class='latex-displaystyle' /> satisfies <img src='http://math.fontein.de/wp-content/latex/a7f/a7fa944d04dcc2cbc9f6dba56863c3dd-T-000000-0.png' alt='g(n_1, \dots, n_s) \ge 0' title='g(n_1, \dots, n_s) \ge 0' class='latex-inline' />, and if it is strictly larger than zero, <img src='http://math.fontein.de/wp-content/latex/5c9/5c9230ab1936ec65b7279c4f53e15d44-T-000000-0.png' alt='\displaystyle  g(n_1, \dots, n_s) \ge \frac{1}{24} n. ' title='\displaystyle  g(n_1, \dots, n_s) \ge \frac{1}{24} n. ' class='latex-displaystyle' /><br />
</div><div class='theoremqed'>□</div></blockquote>
<p>As mentioned, if one invests more work, one can actually show <img src='http://math.fontein.de/wp-content/latex/c4d/c4d4f718c23aedda1ba9177487f9269e-T-000000-0.png' alt='g(n_1, \dots, n_s) \ge \frac{1}{6} n' title='g(n_1, \dots, n_s) \ge \frac{1}{6} n' class='latex-inline' />. For my preprint though, this is not worth the trouble.</p>
]]></content:encoded>
			<wfw:commentRss>http://math.fontein.de/2010/12/09/a-strange-inequality/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
		</item>
		<item>
		<title>Diagonalizable Matrices.</title>
		<link>http://math.fontein.de/2010/01/29/diagonalizable-matrices/</link>
		<comments>http://math.fontein.de/2010/01/29/diagonalizable-matrices/#comments</comments>
		<pubDate>Fri, 29 Jan 2010 04:47:39 +0000</pubDate>
		<dc:creator>Felix Fontein</dc:creator>
				<category><![CDATA[Algebraic Geometry]]></category>
		<category><![CDATA[Linear Algebra]]></category>
		<category><![CDATA[diagonalizable matrices]]></category>
		<category><![CDATA[topological argument]]></category>
		<category><![CDATA[Zariski topology]]></category>

		<guid isPermaLink="false">http://math.fontein.de/?p=612</guid>
		<description><![CDATA[We consider the property of an n times n matrix of being diagonalizable. Is this property open in the standard topology, or the Zariski topology? The emphasis lies on the real and complex numbers, as well as on arbitrary algebraically closed fields.]]></description>
			<content:encoded><![CDATA[<p>Today, during a lecture, we were posed the question whether <img src='http://math.fontein.de/wp-content/latex/a42/a4286f037642624c98226cc35c50cf4d-T-000000-0.png' alt='D_n(K)' title='D_n(K)' class='latex-inline' />, the set of diagonalizable <img src='http://math.fontein.de/wp-content/latex/50f/50f17e5c11d610b19c0471830dc4dda1-T-000000-0.png' alt='n \times n' title='n \times n' class='latex-inline' /> matrices over an algebraically closed field <img src='http://math.fontein.de/wp-content/latex/a5f/a5f3c6a11b03839d46af9fb43c97c188-T-000000-0.png' alt='K' title='K' class='latex-inline' />, is Zariski-open, i.e. open in the <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Zariski_topology">Zariski topology</a>. This would imply that in case <img src='http://math.fontein.de/wp-content/latex/c6a/c6abdf6bf507b19d7ef73f313b4a168d-T-000000-0.png' alt='K = \C' title='K = \C' class='latex-inline' />, the set <img src='http://math.fontein.de/wp-content/latex/bef/befe9c4af38eb94d766458fbc66608ce-T-000000-0.png' alt='D_n(M)' title='D_n(M)' class='latex-inline' /> would be open and dense in <img src='http://math.fontein.de/wp-content/latex/17b/17bf321c18efd4fa053f156a2855401c-T-000000-0.png' alt='M_n(K) = \R^{n \times n}' title='M_n(K) = \R^{n \times n}' class='latex-inline' /> in the <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Standard_topology#Topology_of_Euclidean_space">standard (Euclidean) topolgy</a>.</p>

<p>Unfortunately, the answer turns out to be &ldquo;no&rdquo; for the case <img src='http://math.fontein.de/wp-content/latex/c6a/c6abdf6bf507b19d7ef73f313b4a168d-T-000000-0.png' alt='K = \C' title='K = \C' class='latex-inline' /> (as well as <img src='http://math.fontein.de/wp-content/latex/4fd/4fd3cd3d0c61b052ccb7e9c79332db0c-T-000000-0.png' alt='K = \R' title='K = \R' class='latex-inline' />):</p>

<blockquote class='example'><div class='exampletitle'>Example.</div> <div class='examplemain'>
Let <img src='http://math.fontein.de/wp-content/latex/ae2/ae22d74db6a49ed650084b282db907fd-T-000000-0.png' alt='n \ge 2' title='n \ge 2' class='latex-inline' />. Consider the matrix <img src='http://math.fontein.de/wp-content/latex/903/903808c81c5e2e7dde0ac75d985659e5-T-000000-0.png' alt='\displaystyle  A := \Matrix{ 0 &amp; \cdots &amp; 0 \\ \vdots &amp; \ddots &amp; \vdots \\ 0 &amp; \cdots &amp; 0 } \in D_n(\C), ' title='\displaystyle  A := \Matrix{ 0 &amp; \cdots &amp; 0 \\ \vdots &amp; \ddots &amp; \vdots \\ 0 &amp; \cdots &amp; 0 } \in D_n(\C), ' class='latex-displaystyle' /> as well as the sequence <img src='http://math.fontein.de/wp-content/latex/ade/ade7e2c5565e66628d7e8675180bcf1a-T-000000-0.png' alt='\displaystyle  A_m := \Matrix{ 0 &amp; 1/m &amp; 0 &amp; \cdots &amp; 0 \\ \vdots &amp; \ddots &amp; 0 &amp; \ddots &amp; \vdots \\ \vdots &amp; &amp; \ddots &amp; \ddots &amp; 0 \\ \vdots &amp; &amp; &amp; \ddots &amp; 0 \\ 0 &amp; \cdots &amp; \cdots &amp; \cdots &amp; 0 } \in M_n(\C). ' title='\displaystyle  A_m := \Matrix{ 0 &amp; 1/m &amp; 0 &amp; \cdots &amp; 0 \\ \vdots &amp; \ddots &amp; 0 &amp; \ddots &amp; \vdots \\ \vdots &amp; &amp; \ddots &amp; \ddots &amp; 0 \\ \vdots &amp; &amp; &amp; \ddots &amp; 0 \\ 0 &amp; \cdots &amp; \cdots &amp; \cdots &amp; 0 } \in M_n(\C). ' class='latex-displaystyle' /> Clearly, <img src='http://math.fontein.de/wp-content/latex/456/45622fea31e8eaa4c63b486d2606d847-T-000000-0.png' alt='\lim_{m\to\infty} A_m = A' title='\lim_{m\to\infty} A_m = A' class='latex-inline' />. Assume that <img src='http://math.fontein.de/wp-content/latex/f3c/f3c9a3aa66c63f925587b9796a6ccc1c-T-000000-0.png' alt='D_n(\C)' title='D_n(\C)' class='latex-inline' /> is open in <img src='http://math.fontein.de/wp-content/latex/8eb/8ebc72779fa3ddeccff4b3683e7e7a49-T-000000-0.png' alt='M_n(\C)' title='M_n(\C)' class='latex-inline' />; then we must have <img src='http://math.fontein.de/wp-content/latex/c11/c1173a177b4ece5b93214b5339a03721-T-000000-0.png' alt='A_m \in D_n(\C)' title='A_m \in D_n(\C)' class='latex-inline' /> for almost all <img src='' alt='Formula does not parse: m \in \IN' title='Formula does not parse: m \in \IN' class='latex-inline' />. But <img src='http://math.fontein.de/wp-content/latex/d78/d7867ddd2b2e1311b03302260b65815f-T-000000-0.png' alt='m A_m' title='m A_m' class='latex-inline' /> is in <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Jordan_canonical_form">Jordan canonical form</a>, and clearly not diagonalizable; but this means that <img src='http://math.fontein.de/wp-content/latex/622/6229582d82419733f47d5c1be8132bbd-T-000000-0.png' alt='A_m \not\in D_n(\C)' title='A_m \not\in D_n(\C)' class='latex-inline' /> for <b>all</b> <img src='' alt='Formula does not parse: m \in \IN' title='Formula does not parse: m \in \IN' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/f3c/f3c9a3aa66c63f925587b9796a6ccc1c-T-000000-0.png' alt='D_n(\C)' title='D_n(\C)' class='latex-inline' /> is not open in <img src='http://math.fontein.de/wp-content/latex/8eb/8ebc72779fa3ddeccff4b3683e7e7a49-T-000000-0.png' alt='M_n(\C)' title='M_n(\C)' class='latex-inline' />. 
</div></blockquote>

<p>But nonetheless, <img src='http://math.fontein.de/wp-content/latex/a42/a4286f037642624c98226cc35c50cf4d-T-000000-0.png' alt='D_n(K)' title='D_n(K)' class='latex-inline' /> contains a Zariski-open subset of <img src='http://math.fontein.de/wp-content/latex/542/542a168ec328abfdad97020ffca04ac2-T-000000-0.png' alt='M_n(K)' title='M_n(K)' class='latex-inline' /> in case <img src='http://math.fontein.de/wp-content/latex/a5f/a5f3c6a11b03839d46af9fb43c97c188-T-000000-0.png' alt='K' title='K' class='latex-inline' /> is algebraically closed (which implies that <img src='http://math.fontein.de/wp-content/latex/f3c/f3c9a3aa66c63f925587b9796a6ccc1c-T-000000-0.png' alt='D_n(\C)' title='D_n(\C)' class='latex-inline' /> lies dense in <img src='http://math.fontein.de/wp-content/latex/8eb/8ebc72779fa3ddeccff4b3683e7e7a49-T-000000-0.png' alt='M_n(\C)' title='M_n(\C)' class='latex-inline' />). For that recall that <img src='http://math.fontein.de/wp-content/latex/78a/78aecd4acce358d314d40e49e07f0e6b-T-000000-0.png' alt='\chi_A = \det(x E_n - A) \in K[x]' title='\chi_A = \det(x E_n - A) \in K[x]' class='latex-inline' /> is the characteristic polynomial of <img src='http://math.fontein.de/wp-content/latex/925/9254d448c174bc4ff54f942da1569e94-T-000000-0.png' alt='A \in M_n(K)' title='A \in M_n(K)' class='latex-inline' />.</p>

<blockquote class='theorem'><div class='theoremtitle'>Proposition.</div> <div class='theoremmain'>
Consider the set <img src='http://math.fontein.de/wp-content/latex/8b0/8b035a6aa72518a243012c29385c9df2-T-000000-0.png' alt='\displaystyle  V_n(K) := \{ A \in M_n(K) \mid \chi_A \text{ is squarefree } \}. ' title='\displaystyle  V_n(K) := \{ A \in M_n(K) \mid \chi_A \text{ is squarefree } \}. ' class='latex-displaystyle' /> Then <img src='http://math.fontein.de/wp-content/latex/fb4/fb4d3f53ffd38f8d31e0672def6d440c-T-000000-0.png' alt='V_n(K) \subseteq M_n(K)' title='V_n(K) \subseteq M_n(K)' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/139/139d2dfe109aee66d31d3275b7576dcd-T-000000-0.png' alt='V_n(K)' title='V_n(K)' class='latex-inline' /> is Zariski-open in <img src='http://math.fontein.de/wp-content/latex/542/542a168ec328abfdad97020ffca04ac2-T-000000-0.png' alt='M_n(K)' title='M_n(K)' class='latex-inline' />. In fact, <img src='http://math.fontein.de/wp-content/latex/139/139d2dfe109aee66d31d3275b7576dcd-T-000000-0.png' alt='V_n(K)' title='V_n(K)' class='latex-inline' /> is the complement of a hypersurface in <img src='http://math.fontein.de/wp-content/latex/542/542a168ec328abfdad97020ffca04ac2-T-000000-0.png' alt='M_n(K)' title='M_n(K)' class='latex-inline' />.
</div></blockquote>

<blockquote class='proof'><div class='prooftitle'>Proof.</div> <div class='proofmain'>
Note that in case <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' /> is squarefree, <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' /> splits into distinct linear factors since <img src='http://math.fontein.de/wp-content/latex/a5f/a5f3c6a11b03839d46af9fb43c97c188-T-000000-0.png' alt='K' title='K' class='latex-inline' /> is algebraically closed. Hence, <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' /> has <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' /> distinct eigenvalues in <img src='http://math.fontein.de/wp-content/latex/a5f/a5f3c6a11b03839d46af9fb43c97c188-T-000000-0.png' alt='K' title='K' class='latex-inline' /> and therefore one obtains <img src='http://math.fontein.de/wp-content/latex/7b8/7b8b965ad4bca0e41ab51de7b31363a1-T-000000-0.png' alt='n' title='n' class='latex-inline' /> linearly independent eigenvectors of <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' />; i.e., <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' /> is diagonalizable over <img src='http://math.fontein.de/wp-content/latex/a5f/a5f3c6a11b03839d46af9fb43c97c188-T-000000-0.png' alt='K' title='K' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/fb4/fb4d3f53ffd38f8d31e0672def6d440c-T-000000-0.png' alt='V_n(K) \subseteq M_n(K)' title='V_n(K) \subseteq M_n(K)' class='latex-inline' />.
<br />
Now we show that <img src='http://math.fontein.de/wp-content/latex/3df/3df0b13554c6005dbdff2b056f8dbc2a-T-000000-0.png' alt='M_n(K) \setminus V_n(K)' title='M_n(K) \setminus V_n(K)' class='latex-inline' /> is a hypersurface in <img src='http://math.fontein.de/wp-content/latex/542/542a168ec328abfdad97020ffca04ac2-T-000000-0.png' alt='M_n(K)' title='M_n(K)' class='latex-inline' />, i.e. there exists a polynomial <img src='http://math.fontein.de/wp-content/latex/56a/56ac75808e038ce796f26ea3bda2b70d-T-000000-0.png' alt='f \in K[x_{ij} \mid 1 \le i, j \le n]' title='f \in K[x_{ij} \mid 1 \le i, j \le n]' class='latex-inline' /> such that <img src='http://math.fontein.de/wp-content/latex/fe5/fe52143592773c9a5f8a13d1ffb73819-T-000000-0.png' alt='V_n(K) = \{ A \in M_n(K) \mid f(A) \neq 0 \}' title='V_n(K) = \{ A \in M_n(K) \mid f(A) \neq 0 \}' class='latex-inline' />. For that, consider the maps <img src='http://math.fontein.de/wp-content/latex/1ad/1ad23a33e065528a9b12163e9d5e0109-T-000000-0.png' alt='f_0, \dots, f_{n-1} : M_n(K) \to K' title='f_0, \dots, f_{n-1} : M_n(K) \to K' class='latex-inline' /> defined by <img src='http://math.fontein.de/wp-content/latex/2fc/2fc3743ab68d8d263ee7adc5d581bc4c-T-000000-0.png' alt='\chi_A = x^n + \sum_{i=0}^{n-1} f_i(A) x^i' title='\chi_A = x^n + \sum_{i=0}^{n-1} f_i(A) x^i' class='latex-inline' /> for all <img src='http://math.fontein.de/wp-content/latex/925/9254d448c174bc4ff54f942da1569e94-T-000000-0.png' alt='A \in M_n(K)' title='A \in M_n(K)' class='latex-inline' />. Obviously, these <img src='http://math.fontein.de/wp-content/latex/59b/59bdf0ba696e13164c5a926386f23cb0-T-000000-0.png' alt='f_i' title='f_i' class='latex-inline' /> must be polynomials. Next, consider the <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Discriminant#Discriminant_of_a_polynomial">discriminant</a> <img src='http://math.fontein.de/wp-content/latex/a18/a189200d28d32edecba26d05576e8848-T-000000-0.png' alt='D(\chi_A)' title='D(\chi_A)' class='latex-inline' /> of <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' />; this is a polynomial expression in the coefficients of <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' />, i.e. in <img src='http://math.fontein.de/wp-content/latex/01a/01a3a2c27fff41602fad2a0d03b3a537-T-000000-0.png' alt='1, f_0(A), \dots, f_{n-1}(A)' title='1, f_0(A), \dots, f_{n-1}(A)' class='latex-inline' />, whose value is zero if, and only if, <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' /> is squarefree. Therefore, <img src='http://math.fontein.de/wp-content/latex/879/8798931d61d33849d48fe171943a47b4-T-000000-0.png' alt='A \in V_n(K) \Leftrightarrow D(\chi_A) \neq 0' title='A \in V_n(K) \Leftrightarrow D(\chi_A) \neq 0' class='latex-inline' />. Finally, <img src='http://math.fontein.de/wp-content/latex/4c3/4c32b50564dcede9d590b62bf706b730-T-000000-0.png' alt='f := D(\chi_A)' title='f := D(\chi_A)' class='latex-inline' /> is a polynomial, whence <img src='http://math.fontein.de/wp-content/latex/fe5/fe52143592773c9a5f8a13d1ffb73819-T-000000-0.png' alt='V_n(K) = \{ A \in M_n(K) \mid f(A) \neq 0 \}' title='V_n(K) = \{ A \in M_n(K) \mid f(A) \neq 0 \}' class='latex-inline' /> is Zariski-open in <img src='http://math.fontein.de/wp-content/latex/542/542a168ec328abfdad97020ffca04ac2-T-000000-0.png' alt='M_n(K)' title='M_n(K)' class='latex-inline' />.
</div><div class='proofqed'>□</div></blockquote>

<p>Note that the situation is different over <img src='http://math.fontein.de/wp-content/latex/7b0/7b09fea775960abce9cd821eb6ee9ef7-T-000000-0.png' alt='\R' title='\R' class='latex-inline' />:</p>

<blockquote class='theorem'><div class='theoremtitle'>Proposition.</div> <div class='theoremmain'>
In the standard topology, <img src='http://math.fontein.de/wp-content/latex/053/0535d9f0691f1afb15e88809170fc1de-T-000000-0.png' alt=' &amp; \overline{D_n(\R)} = \overline{D_n(\R) \cap V_n(\R)} \\ {}={} &amp; \{ A \in M_n(\R) \mid A \text{ has only real eigenvalues } \} \\ {}={} &amp; \{ A \in M_n(\R) \mid A \text{ has a Jordan canonical form over } \R \}.' title=' &amp; \overline{D_n(\R)} = \overline{D_n(\R) \cap V_n(\R)} \\ {}={} &amp; \{ A \in M_n(\R) \mid A \text{ has only real eigenvalues } \} \\ {}={} &amp; \{ A \in M_n(\R) \mid A \text{ has a Jordan canonical form over } \R \}.' class='latex-displaystyle' />
</div></blockquote>

<blockquote class='proof'><div class='prooftitle'>Proof.</div> <div class='proofmain'>
Assume that <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' /> has at least one eigenvalue <img src='http://math.fontein.de/wp-content/latex/748/7486a0ac2f120b731ef87a64bc6fe6a8-T-000000-0.png' alt='\lambda \in \C' title='\lambda \in \C' class='latex-inline' /> with imaginary part <img src='http://math.fontein.de/wp-content/latex/7ff/7ff82e196ed9d515f4847f89f9f3a7f8-T-000000-0.png' alt='\Im \lambda \neq 0' title='\Im \lambda \neq 0' class='latex-inline' />. If <img src='http://math.fontein.de/wp-content/latex/c6f/c6fa38eb47b1af5ceded1da81ec3cc30-T-000000-0.png' alt='(A_m)_m' title='(A_m)_m' class='latex-inline' /> is a sequence of matrices with <img src='http://math.fontein.de/wp-content/latex/456/45622fea31e8eaa4c63b486d2606d847-T-000000-0.png' alt='\lim_{m\to\infty} A_m = A' title='\lim_{m\to\infty} A_m = A' class='latex-inline' />, each <img src='http://math.fontein.de/wp-content/latex/8c6/8c6eeb9d794937840bcf6a27fc52bb87-T-000000-0.png' alt='A_m' title='A_m' class='latex-inline' /> must have an eigenvalue <img src='http://math.fontein.de/wp-content/latex/228/228f560521346a911bc937e45f1f8345-T-000000-0.png' alt='\lambda_m \in \C' title='\lambda_m \in \C' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/1b8/1b8ac0d5cd709e2f9b75bb1653a8e248-T-000000-0.png' alt='\lim_{m\to\infty} \lambda_m = \lambda' title='\lim_{m\to\infty} \lambda_m = \lambda' class='latex-inline' />. But then, for infinitely many <img src='http://math.fontein.de/wp-content/latex/6f8/6f8f57715090da2632453988d9a1501b-T-000000-0.png' alt='m' title='m' class='latex-inline' />, we must have <img src='http://math.fontein.de/wp-content/latex/f03/f03a1884824c5d83c229080da149f569-T-000000-0.png' alt='\lambda_m \not\in \R' title='\lambda_m \not\in \R' class='latex-inline' /> (since <img src='http://math.fontein.de/wp-content/latex/b53/b53da9b0380ce4bef6b6184217aa15e4-T-000000-0.png' alt='\C \setminus \R' title='\C \setminus \R' class='latex-inline' /> is open), whence we cannot have <img src='http://math.fontein.de/wp-content/latex/ea4/ea426a49afd704b20090b891799d33d1-T-000000-0.png' alt='A_m \in D_n(\R)' title='A_m \in D_n(\R)' class='latex-inline' /> for infinitely many <img src='http://math.fontein.de/wp-content/latex/6f8/6f8f57715090da2632453988d9a1501b-T-000000-0.png' alt='m' title='m' class='latex-inline' />. Hence, <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' /> is not in the closure of <img src='http://math.fontein.de/wp-content/latex/6f6/6f6368817f902c0042ace3f784724ae3-T-000000-0.png' alt='D_n(\R) \cap V_n(\R)' title='D_n(\R) \cap V_n(\R)' class='latex-inline' />.
<br />
Now assume that <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' /> has only real eigenvalues. Then there exist some <img src='http://math.fontein.de/wp-content/latex/a25/a251f92bce948a3f34b374a75f02af3f-T-000000-0.png' alt='T \in GL_n(\R)' title='T \in GL_n(\R)' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/3cc/3cc5f319996834fe70a2014de317b4da-T-000000-0.png' alt='T^{-1} A T' title='T^{-1} A T' class='latex-inline' /> in Jordan canonical form. By pertubing the diagonal elements of <img src='http://math.fontein.de/wp-content/latex/3cc/3cc5f319996834fe70a2014de317b4da-T-000000-0.png' alt='T^{-1} A T' title='T^{-1} A T' class='latex-inline' /> slightly, we can obtain a sequence of matrices <img src='http://math.fontein.de/wp-content/latex/922/9225b1f3956e13dd3613f420dd9ea29e-T-000000-0.png' alt='B_m \in V_n(\R) \cap D_n(\R)' title='B_m \in V_n(\R) \cap D_n(\R)' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/7e8/7e889d6df9968b09f1d05ab78fb10ff5-T-000000-0.png' alt='\lim_{m \to \infty} B_m \to T^{-1} A T' title='\lim_{m \to \infty} B_m \to T^{-1} A T' class='latex-inline' />. But then, <img src='http://math.fontein.de/wp-content/latex/2e5/2e58b61e8e67c388f75324694b88cbe3-T-000000-0.png' alt='\lim_{m\to\infty} T B_m T^{-1} = A' title='\lim_{m\to\infty} T B_m T^{-1} = A' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/1a7/1a76754a1c4d4bba312024626d3a1752-T-000000-0.png' alt='T B_m T^{-1} \in V_n(\R) \cap D_n(\R)' title='T B_m T^{-1} \in V_n(\R) \cap D_n(\R)' class='latex-inline' /> for every <img src='http://math.fontein.de/wp-content/latex/08a/08a92a9ada8121d5a4f159211c2c06b4-T-000000-0.png' alt='m \in \N' title='m \in \N' class='latex-inline' />.
<br />
Note that this implies <img src='http://math.fontein.de/wp-content/latex/608/60859192170776c7366caca824dcdb3c-T-000000-0.png' alt='A \in \overline{V_n(\R) \cap D_n(\R)}' title='A \in \overline{V_n(\R) \cap D_n(\R)}' class='latex-inline' />; moreover, this also implies <img src='http://math.fontein.de/wp-content/latex/257/25772743c13119aa16c8154e58920206-T-000000-0.png' alt='D_n(\R) \subseteq \overline{D_n(\R) \cap V_n(\R)}' title='D_n(\R) \subseteq \overline{D_n(\R) \cap V_n(\R)}' class='latex-inline' />. Hence, the first two equalities hold. The third equality is standard.
</div><div class='proofqed'>□</div></blockquote>

<p>Also note that <img src='http://math.fontein.de/wp-content/latex/6c9/6c9ee16a1019535ad399f55ba333c514-T-000000-0.png' alt='V_n(\R) \not\subseteq D_n(\R)' title='V_n(\R) \not\subseteq D_n(\R)' class='latex-inline' /> for <img src='http://math.fontein.de/wp-content/latex/b54/b5483f2f1af0ee56acb54ef669929a57-T-000000-0.png' alt='n &gt; 1' title='n &gt; 1' class='latex-inline' />, as the example <img src='http://math.fontein.de/wp-content/latex/ff5/ff5b0dd7f179e4e78ff2e74d6244d48e-T-000000-0.png' alt='\displaystyle  \Matrix{ 0 &amp; 1 \\ -1 &amp; 0 } ' title='\displaystyle  \Matrix{ 0 &amp; 1 \\ -1 &amp; 0 } ' class='latex-displaystyle' /> (which is diagonalizable over <img src='http://math.fontein.de/wp-content/latex/ee7/ee77cd72573eec25fba471d91befc2d2-T-000000-0.png' alt='\C' title='\C' class='latex-inline' />, with eigenvalues <img src='http://math.fontein.de/wp-content/latex/365/36589c0963bd9597f22f7a7e659cae3b-T-000000-0.png' alt='\pm i' title='\pm i' class='latex-inline' />) shows. So what about <img src='http://math.fontein.de/wp-content/latex/e1a/e1a1c9b029d80b4097200c635cb336ca-T-000000-0.png' alt='\overline{V_n(\R)}' title='\overline{V_n(\R)}' class='latex-inline' />? In fact, as in the case of <img src='http://math.fontein.de/wp-content/latex/c6a/c6abdf6bf507b19d7ef73f313b4a168d-T-000000-0.png' alt='K = \C' title='K = \C' class='latex-inline' />, it turns out that <img src='http://math.fontein.de/wp-content/latex/e1a/e1a1c9b029d80b4097200c635cb336ca-T-000000-0.png' alt='\overline{V_n(\R)}' title='\overline{V_n(\R)}' class='latex-inline' /> is <img src='http://math.fontein.de/wp-content/latex/da8/da8b83193c17f7d94e908daf50d2cf99-T-000000-0.png' alt='M_n(\R)' title='M_n(\R)' class='latex-inline' />.</p>

<blockquote class='theorem'><div class='theoremtitle'>Proposition.</div> <div class='theoremmain'>
We have <img src='http://math.fontein.de/wp-content/latex/c6a/c6a75c0b817d409d61e2b780105c8e78-T-000000-0.png' alt='\displaystyle  \overline{V_n(\R)} = M_n(\R). ' title='\displaystyle  \overline{V_n(\R)} = M_n(\R). ' class='latex-displaystyle' />
</div></blockquote>

<p>For the proof, we need a little lemma.</p>

<blockquote class='theorem'><div class='theoremtitle'>Lemma.</div> <div class='theoremmain'>
Let <img src='http://math.fontein.de/wp-content/latex/36c/36c4f9876db96f5d2ffaac87599ee1cd-T-000000-0.png' alt='S := \{ f \in \R[x] \mid f \text{ is squarefree } \}' title='S := \{ f \in \R[x] \mid f \text{ is squarefree } \}' class='latex-inline' />. Then <img src='http://math.fontein.de/wp-content/latex/f58/f5860883a7cc1e5ba5543a0ce4e86315-T-000000-0.png' alt='\overline{S} = \R[x]' title='\overline{S} = \R[x]' class='latex-inline' />.
</div></blockquote>

<blockquote class='proof'><div class='prooftitle'>Proof.</div> <div class='proofmain'>
Let <img src='http://math.fontein.de/wp-content/latex/3f5/3f5f5e20608eb4d4e1bcfd6aeba85e3d-T-000000-0.png' alt='f \in \R[x]' title='f \in \R[x]' class='latex-inline' /> be an arbitrary polynomial. Write <img src='http://math.fontein.de/wp-content/latex/5b2/5b2b9e0ad217fb225eff284fb8c55b11-T-000000-0.png' alt='f = \lambda \prod_{i=1}^n p_i' title='f = \lambda \prod_{i=1}^n p_i' class='latex-inline' />, where <img src='http://math.fontein.de/wp-content/latex/399/399a2dc7632afc7e271834c5050b5e83-T-000000-0.png' alt='\lambda \in \R^*' title='\lambda \in \R^*' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/95e/95e49ce959facccf131afcdfd34179e3-T-000000-0.png' alt='p_i \in \R[x]' title='p_i \in \R[x]' class='latex-inline' /> is irreducible and monic, <img src='http://math.fontein.de/wp-content/latex/b2b/b2b7ab3a6158d153d879eaf940e1d715-T-000000-0.png' alt='1 \le i \le n' title='1 \le i \le n' class='latex-inline' />. Now the coefficients of all <img src='http://math.fontein.de/wp-content/latex/eca/eca91c83a74a2373ca5f796700e99fd3-T-000000-0.png' alt='p_i' title='p_i' class='latex-inline' />&#8216;s (except the highest coefficients) are a finite set in <img src='http://math.fontein.de/wp-content/latex/7b0/7b09fea775960abce9cd821eb6ee9ef7-T-000000-0.png' alt='\R' title='\R' class='latex-inline' /> of <img src='http://math.fontein.de/wp-content/latex/f84/f84ff19c131c304d9d8726bcacba509b-T-000000-0.png' alt='d := \sum_{i=1}^n \deg p_i' title='d := \sum_{i=1}^n \deg p_i' class='latex-inline' /> elements, whence there exists sequences <img src='http://math.fontein.de/wp-content/latex/859/8599c6b09ceedb7f85e4456292b4a456-T-000000-0.png' alt='(a_1^{(m)}, \dots, a_d^{(m)})' title='(a_1^{(m)}, \dots, a_d^{(m)})' class='latex-inline' /> with pairwise distinct <img src='http://math.fontein.de/wp-content/latex/eb0/eb052547cfbc5b27970b0218547e6838-T-000000-0.png' alt='a_1^{(m)}, \dots, a_d^{(m)}' title='a_1^{(m)}, \dots, a_d^{(m)}' class='latex-inline' /> such that <img src='http://math.fontein.de/wp-content/latex/bcb/bcbd046ce5b74105c2da1913206c6963-T-000000-0.png' alt='\lim a_i^{(m)}' title='\lim a_i^{(m)}' class='latex-inline' /> converges to one coefficent of one <img src='http://math.fontein.de/wp-content/latex/8b6/8b6f59f2af8f45b773cb64ac76c9b095-T-000000-0.png' alt='p_j' title='p_j' class='latex-inline' />. In particular, we can construct monic polynomials <img src='http://math.fontein.de/wp-content/latex/700/70046ce833e0e697754119147cd02287-T-000000-0.png' alt='p_i^{(m)} \in \R[x]' title='p_i^{(m)} \in \R[x]' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/a2b/a2bbaf9efcef3fcbf0bb9f05c9d31ca3-T-000000-0.png' alt='\deg p_i^{(m)} = \deg p_i' title='\deg p_i^{(m)} = \deg p_i' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/d50/d506fade03d64242cd035ef902ba9904-T-000000-0.png' alt='\lim_{m\to\infty} p_i^{(m)} = p_i' title='\lim_{m\to\infty} p_i^{(m)} = p_i' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/ddc/ddc6b478f84660b580c25ca3a7ffeb1e-T-000000-0.png' alt='p_i^{(m)} \neq p_j^{(m)}' title='p_i^{(m)} \neq p_j^{(m)}' class='latex-inline' /> for every <img src='http://math.fontein.de/wp-content/latex/82c/82cb38cfb7b8f079dda70c6a96f37479-T-000000-0.png' alt='i \neq j' title='i \neq j' class='latex-inline' />. Even more, we can make sure that every <img src='http://math.fontein.de/wp-content/latex/b94/b943faceb91be0079d0fab5ac40060e7-T-000000-0.png' alt='p_i^{(m)}' title='p_i^{(m)}' class='latex-inline' /> is irreducible; this enforces that <img src='http://math.fontein.de/wp-content/latex/6cf/6cfbe10fd847aea3368ad835ba3cddbb-T-000000-0.png' alt='f_m := \prod_{i=1}^n p_i^{(m)}' title='f_m := \prod_{i=1}^n p_i^{(m)}' class='latex-inline' /> is squarefree, i.e. <img src='http://math.fontein.de/wp-content/latex/92a/92abf41e3db1de5f8f3d0615526e43b4-T-000000-0.png' alt='f_m \in S' title='f_m \in S' class='latex-inline' />. Therefore, we found a sequence in <img src='http://math.fontein.de/wp-content/latex/5db/5dbc98dcc983a70728bd082d1a47546e-T-000000-0.png' alt='S' title='S' class='latex-inline' /> converging to <img src='http://math.fontein.de/wp-content/latex/8fa/8fa14cdd754f91cc6554c9e71929cce7-T-000000-0.png' alt='f' title='f' class='latex-inline' />, whence <img src='http://math.fontein.de/wp-content/latex/cbe/cbeb139f67bd6de2e5a4f4e9f2166cb0-T-000000-0.png' alt='f \in \overline{S}' title='f \in \overline{S}' class='latex-inline' />.
</div><div class='proofqed'>□</div></blockquote>

<blockquote class='proof'><div class='prooftitle'>Proof (Proof of the Proposition).</div> <div class='proofmain'>
Let <img src='http://math.fontein.de/wp-content/latex/bf0/bf0f6e691c450f02ace178a2fa662b69-T-000000-0.png' alt='A \in M_n(\R)' title='A \in M_n(\R)' class='latex-inline' /> whose characteristic polynomial <img src='http://math.fontein.de/wp-content/latex/50e/50e7626b086e5a0a454c76be2f9b7852-T-000000-0.png' alt='\chi_A' title='\chi_A' class='latex-inline' /> can be written as <img src='http://math.fontein.de/wp-content/latex/5c7/5c715121ce1f053417d5c254d4196718-T-000000-0.png' alt='\prod_{i=1}^t p_i' title='\prod_{i=1}^t p_i' class='latex-inline' />, with not necessarily distinct, but monic and irreducible polynomials <img src='http://math.fontein.de/wp-content/latex/23b/23b9daace8ca783d4c316b0cf11eb251-T-000000-0.png' alt='p_1, \dots, p_n \in \R[x]' title='p_1, \dots, p_n \in \R[x]' class='latex-inline' />. There exists a matrix <img src='http://math.fontein.de/wp-content/latex/a25/a251f92bce948a3f34b374a75f02af3f-T-000000-0.png' alt='T \in GL_n(\R)' title='T \in GL_n(\R)' class='latex-inline' /> with <img src='http://math.fontein.de/wp-content/latex/a49/a49f90a53537e4824d5705d21f2c68f6-T-000000-0.png' alt='\displaystyle  T^{-1} A T = \Matrix{ C_{p_1} &amp; &amp; 0 \\ &amp; \ddots &amp; \\ 0 &amp; &amp; C_{p_t} }, ' title='\displaystyle  T^{-1} A T = \Matrix{ C_{p_1} &amp; &amp; 0 \\ &amp; \ddots &amp; \\ 0 &amp; &amp; C_{p_t} }, ' class='latex-displaystyle' /> where <img src='http://math.fontein.de/wp-content/latex/624/62414c4de7a99ede511aa941b089db4b-T-000000-0.png' alt='C_{p_i}' title='C_{p_i}' class='latex-inline' /> is the <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Companion_matrix">companion matrix</a> of <img src='http://math.fontein.de/wp-content/latex/eca/eca91c83a74a2373ca5f796700e99fd3-T-000000-0.png' alt='p_i' title='p_i' class='latex-inline' />; this is a <a href="http://math.fontein.de/forward.php?r=http://en.wikipedia.org/wiki/Frobenius_normal_form">Frobenius normal form</a> of <img src='http://math.fontein.de/wp-content/latex/7fc/7fc56270e7a70fa81a5935b72eacbe29-T-000000-0.png' alt='A' title='A' class='latex-inline' />. Now we can find a sequence of squarefree polynomials <img src='http://math.fontein.de/wp-content/latex/700/70046ce833e0e697754119147cd02287-T-000000-0.png' alt='p_i^{(m)} \in \R[x]' title='p_i^{(m)} \in \R[x]' class='latex-inline' /> such that for every <img src='http://math.fontein.de/wp-content/latex/6f8/6f8f57715090da2632453988d9a1501b-T-000000-0.png' alt='m' title='m' class='latex-inline' />, <img src='http://math.fontein.de/wp-content/latex/c87/c87ab960a03cad32c35f1a5093f6237a-T-000000-0.png' alt='p_1^{(m)}, \dots, p_t^{(m)}' title='p_1^{(m)}, \dots, p_t^{(m)}' class='latex-inline' /> are pairwise coprime, and that <img src='http://math.fontein.de/wp-content/latex/d50/d506fade03d64242cd035ef902ba9904-T-000000-0.png' alt='\lim_{m\to\infty} p_i^{(m)} = p_i' title='\lim_{m\to\infty} p_i^{(m)} = p_i' class='latex-inline' />. Then set <img src='http://math.fontein.de/wp-content/latex/d54/d54ce44117bee7335b390512ce9e7492-T-000000-0.png' alt='\displaystyle  A_m := T \Matrix{ C_{p_1^{(m)}} &amp; &amp; 0 \\ &amp; \ddots &amp; \\ 0 &amp; &amp; C_{p_t^{(m)}} } T^{-1} \in M_n(\R); ' title='\displaystyle  A_m := T \Matrix{ C_{p_1^{(m)}} &amp; &amp; 0 \\ &amp; \ddots &amp; \\ 0 &amp; &amp; C_{p_t^{(m)}} } T^{-1} \in M_n(\R); ' class='latex-displaystyle' /> clearly, <img src='http://math.fontein.de/wp-content/latex/456/45622fea31e8eaa4c63b486d2606d847-T-000000-0.png' alt='\lim_{m\to\infty} A_m = A' title='\lim_{m\to\infty} A_m = A' class='latex-inline' />. Moreover, the characteristic polynomial of <img src='http://math.fontein.de/wp-content/latex/8c6/8c6eeb9d794937840bcf6a27fc52bb87-T-000000-0.png' alt='A_m' title='A_m' class='latex-inline' /> is given by <img src='http://math.fontein.de/wp-content/latex/112/112e6bd00ad8019212ccc2e2389407a7-T-000000-0.png' alt='\prod_{i=1}^t p_i^{(m)}' title='\prod_{i=1}^t p_i^{(m)}' class='latex-inline' />, i.e. it is squarefree by choice of the <img src='http://math.fontein.de/wp-content/latex/b94/b943faceb91be0079d0fab5ac40060e7-T-000000-0.png' alt='p_i^{(m)}' title='p_i^{(m)}' class='latex-inline' />. Therefore, <img src='http://math.fontein.de/wp-content/latex/1b7/1b7ba5d717dc90d06a73f3a0785ee1e0-T-000000-0.png' alt='A_m \in V_n(\R)' title='A_m \in V_n(\R)' class='latex-inline' /> and <img src='http://math.fontein.de/wp-content/latex/899/89915a624186203b9737daf0cf619dd8-T-000000-0.png' alt='A \in \overline{V_n(\R)}' title='A \in \overline{V_n(\R)}' class='latex-inline' />.
</div><div class='proofqed'>□</div></blockquote>
]]></content:encoded>
			<wfw:commentRss>http://math.fontein.de/2010/01/29/diagonalizable-matrices/feed/</wfw:commentRss>
		<slash:comments>0</slash:comments>
		</item>
	</channel>
</rss>

