{"id":808,"date":"2016-01-03T15:54:05","date_gmt":"2016-01-03T14:54:05","guid":{"rendered":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/?page_id=808"},"modified":"2022-11-19T15:25:56","modified_gmt":"2022-11-19T14:25:56","slug":"ideal-models-ii","status":"publish","type":"page","link":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/?page_id=808","title":{"rendered":"Ideal models II"},"content":{"rendered":"<p>Recall that an ideal domain is a dcpo where every non-finite element is maximal.\u00a0 All ideal domains are algebraic and first-countable.\u00a0 The notion, and all the results presented here, are due to Keye Martin [1].<\/p>\n<p>Last time, we have seen that every completely metrizable space <em>X<\/em> has an ideal model, that is, that <em>X<\/em> can be embedded into an ideal domain <em>Y <\/em>in such a way that we can equate <em>X<\/em> with the subspace of maximal elements of Y.<\/p>\n<p>We have also seen the converse to that: if <em>X<\/em> is a metrizable space with an ideal model, then <em>X<\/em> is completely metrizable.\u00a0 The proof made use of Choquet-completeness, and eventually relied on the fact that, if <em>X<\/em> has an ideal model <em>Y<\/em>, then it has one in which it arises as a G<sub>\u03b4<\/sub> subset of <em>Y<\/em>.\u00a0 Keye Martin shows that the set of maximal elements of an ideal domain <em>Y<\/em> is <em>always<\/em> a G<sub>\u03b4<\/sub> subset of <em>Y<\/em>.\u00a0 The proof is pretty technical.\u00a0 We shall be happy to replace <em>Y<\/em> by another ideal domain <em>Y<\/em>&#8216; whose set of maximal elements will again be\u00a0<em>X<\/em>, but will be more easily seen to be a G<sub>\u03b4<\/sub> subset of <em>Y<\/em>&#8216;.<\/p>\n<h2>First-countable continuous posets<\/h2>\n<p>Recall that <em>Y<\/em>, being an ideal domain, is first-countable.\u00a0 We claim that this implies that every element <em>x<\/em> of <em>X<\/em> = Max <em>Y<\/em> (the set of maximal elements of <em>Y<\/em>) is the supremum of a countable chain of finite elements of <em>Y<\/em>.\u00a0 In fact, we have the more general result.<\/p>\n<p><strong>Lemma<\/strong>. A continuous poset <em>Z<\/em> is first-countable in its Scott topology if and only if every element is the supremum of a <em>countable chain<\/em> of elements way-below it.<\/p>\n<p>This is similar to Norberg&#8217;s Lemma (Lemma 7.7.13 in the <a href=\"https:\/\/www.cambridge.org\/gb\/knowledge\/isbn\/item7069109\/Non-Hausdorff%20Topology%20and%20Domain%20Theory\/?site_locale=en_GB\">book<\/a>) that a continuous poset is second-countable in its Scott topology if and only if it has a countable basis.<\/p>\n<p><strong>Proof.<\/strong> If every element <em>z<\/em> is the supremum of a countable chain (<em>z<sub>n<\/sub><\/em>)<em><sub>n \u2208 <strong>N<\/strong><\/sub><\/em> of elements way-below z, then the family of open subsets \u219f<em>z<sub>n<\/sub><\/em>, <em>n<\/em> \u2208 <strong>N<\/strong>, is a countable basis of open neighborhoods of <em>z<\/em>: for every open neighborhood <em>U<\/em> of <em>z<\/em>, some\u00a0<em>z<sub>n<\/sub><\/em> is in <em>U<\/em>, and then \u219f<em>z<sub>n<\/sub><\/em> is an open subset of U that contains <em>z<\/em>.<\/p>\n<p>Conversely, if <em>Z<\/em> is first-countable in its Scott topology, then every element <em>z<\/em> has a countable basis of open neighborhoods <em>U<sub>n<\/sub><\/em>, <em>n<\/em> \u2208 <strong>N<\/strong>, and we can even assume that they form a decreasing chain (Exercise 4.7.14). Since <em>Z<\/em> is a continuous poset, <em>z<\/em> is the supremum of a directed family of elements way-below <em>z<\/em>, and one of them, call it <em>z<\/em><sub>0<\/sub>, is in <em>U<\/em><sub>0<\/sub>.\u00a0 The open subset \u219f<em>z<\/em><sub>0<\/sub> contains <em>z<\/em>, and by the defining property of a basis of open neighborhoods, it must therefore contain some <em>U<sub>n<\/sub><\/em>.\u00a0 Up to some reindexing, imagine that it contains <em>U<\/em><sub>1<\/sub>.\u00a0 We repeat the argument, and find an element\u00a0<em>z<\/em><sub>1<\/sub> way-below <em>z<\/em>, and (up to some reindexing again) \u219f<em>z<\/em><sub>1<\/sub> contains <em>U<\/em><sub>2<\/sub>.\u00a0 Continuing this way, we build <em>z<\/em><sub>2<\/sub>, &#8230;, <em>z<sub>n<\/sub><\/em>, &#8230; , and so on.\u00a0 They are all way-below\u00a0<em>z<\/em>,\u00a0<em>z<sub>n<\/sub><\/em> is in <em>U<sub>n<\/sub><\/em>, and \u219f<em>z<sub>n<\/sub><\/em> contains <em>U<sub>n+1<\/sub><\/em>.\u00a0 To show that the supremum of that chain equals <em>z<\/em>, it suffices to show that every open neighborhood of <em>z<\/em> contains some <em>z<sub>n<\/sub><\/em>.\u00a0 That is clear, since that open neighborhood must contain some <em>U<sub>n<\/sub><\/em>, by the defining property of bases of open neighborhoods.\u00a0 \u2610<\/p>\n<p>By similar arguments, we also obtain:<\/p>\n<p><strong>Lemma<\/strong>. An algebraic poset <em>Z<\/em> is first-countable in its Scott topology if and only if every element is the supremum of a <em>countable chain<\/em> of finite elements below it.<\/p>\n<h2>Tweaking an ideal domain into a convenient one<\/h2>\n<p>Let us return to the ideal domain <em>Y<\/em>, with set of maximal elements <em>X<\/em>.\u00a0 As a consequence of the above lemma, every element <em>x<\/em> of <em>X<\/em> is the supremum of a countable chain of finite elements\u00a0x[<em>n<\/em>] way-below <em>x<\/em>, <em>n<\/em> \u2208 <strong>N<\/strong>.<\/p>\n<p>We use the Axiom of Choice implicitly here: for each <em>x<\/em> in <em>X<\/em>, we fix a countable chain of finite elements x[<em>n<\/em>] whose supremum equals <em>x<\/em>.<\/p>\n<p>We extend the notation <em>x<\/em>[<em>n<\/em>] to the case where <em>n<\/em>=+\u221e, and let <em>x<\/em>[+\u221e] = <em>x<\/em>.<\/p>\n<p>Define a new poset <em>Y<\/em>&#8216; as follows.\u00a0 The elements of <em>Y<\/em>&#8216; are pairs (<em>x<\/em>, <em>n<\/em>) where <em>x<\/em> is in <em>X<\/em> and <em>n<\/em> is in <strong>N<\/strong> U {+\u221e}.\u00a0 Reserve the notation \u2264 for the ordering on <em>Y<\/em> (or for the ordering on natural numbers <em>n<\/em>).\u00a0 The ordering \u2264&#8217; on <em>Y<\/em>&#8216; is given by:<\/p>\n<ul>\n<li>for all <em>n<\/em>, <em>n<\/em>&#8216; in <strong>N<\/strong> U {+\u221e}, (<em>x<\/em>, <em>n<\/em>) \u2264&#8217; (<em>x<\/em>&#8216;, <em>n<\/em>&#8216;) if and only if <em>n<\/em>\u2264<em>n<\/em>&#8216; and <em>x<\/em>[<em>n<\/em>] \u2264 <em>x&#8217;<\/em>[<em>n<\/em>&#8216;].<\/li>\n<\/ul>\n<p>In other words, we compare elements (<em>x<\/em>, <em>n<\/em>) as though they really denote <em>x<\/em>[<em>n<\/em>], except that we first compare their level <em>n<\/em>: an element (<em>x<\/em>, <em>n<\/em>) can only be below (<em>x<\/em>&#8216;, <em>n<\/em>&#8216;) if its level <em>n<\/em> is below the level <em>n<\/em>&#8216; of (<em>x<\/em>&#8216;, <em>n<\/em>&#8216;).\u00a0 We naturally equate\u00a0<em>x<\/em> in\u00a0<em>X<\/em> with the element (<em>x<\/em>, +\u221e).<\/p>\n<p><strong>Lemma.<\/strong> Every directed family (<em>x<sub>i<\/sub><\/em>, <em>n<sub>i<\/sub><\/em>)<em><sub>i \u2208 I<\/sub><\/em> in <em>Y<\/em>&#8216; has a supremum, and it is equal to (sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>], sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>).<\/p>\n<p><strong>Proof.<\/strong> Let (<em>x<\/em>, <em>n<\/em>) = (sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>], sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>).\u00a0 This is clearly an upper bound of the family, and if (<em>x<\/em>&#8216;, <em>n<\/em>&#8216;) is any other upper bound, then <em>n<sub>i<\/sub><\/em>\u2264<em>n<\/em>&#8216; for every <em>i<\/em>, so <em>n<\/em>=sup<sub>i<\/sub><em><em><sub> \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>\u2264n<\/em>&#8216;, and <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>]\u2264<em>x<\/em>&#8216; for every <em>i<\/em>, so <em>x<\/em>=sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>]\u2264<em>x<\/em>&#8216;.\u00a0 \u2610<\/p>\n<p><strong>Lemma.<\/strong> For every <em>n<\/em> in <strong>N<\/strong>, for every <em>x<\/em> in <em>X<\/em>, (<em>x<\/em>, <em>n<\/em>) is a finite element of <em>Y<\/em>&#8216;.<\/p>\n<p><strong>Proof.<\/strong> Assume (<em>x<\/em>, <em>n<\/em>) is below some directed supremum (sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>], sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>), i.e., for some monotone net (<em>x<sub>i<\/sub><\/em>, <em>n<sub>i<\/sub><\/em>)<em><sub>i \u2208 I<\/sub><\/em> in <em>Y<\/em>&#8216;.\u00a0 Since <em>n<\/em> is finite, <em>n<\/em>\u2264<em>n<sub>i<\/sub><\/em> for <em>i<\/em> large enough.\u00a0 By restricting to those indices <em>i<\/em> such that<em> n<\/em>\u2264<em>n<sub>i<\/sub><\/em>, we can therefore assume that <em>n<\/em>\u2264<em>n<sub>i<\/sub><\/em> for every\u00a0<em>i<\/em> in <em>I<\/em>.<\/p>\n<p>If sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>=+\u221e, then (<em>x<\/em>, <em>n<\/em>) \u2264&#8217; (sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>], sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>) is equivalent to <em>x<\/em>[<em>n<\/em>] \u2264 sup<em><sub>i \u2208 I<\/sub><\/em> <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>], in which case <em>x<\/em>[<em>n<\/em>] \u2264 <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>] for <em>i<\/em> large enough, since <em>x<\/em>[<em>n<\/em>] is finite in <em>Y<\/em>.\u00a0 Then (<em>x<\/em>, <em>n<\/em>) \u2264&#8217; (<em>x<sub>i<\/sub><\/em>, <em>n<sub>i<\/sub><\/em>).<\/p>\n<p>If sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>&lt;+\u221e, then let <em>n<\/em>&#8216;=sup<em><sub>i \u2208 I<\/sub><\/em> <em>n<sub>i<\/sub><\/em>.\u00a0 For <em>i<\/em> large enough, <em>n<sub>i<\/sub><\/em>=<em>n<\/em>&#8216;.\u00a0 By reindexing again, we may assume that <em>n<sub>i<\/sub><\/em>=<em>n<\/em>&#8216; for every <em>i<\/em>. Observe that (<em>x<sub>i<\/sub><\/em>, <em>n<sub>i<\/sub><\/em>)\u2264&#8217;\u00a0(<em>x<sub>j<\/sub><\/em>, <em>n<sub>j<\/sub><\/em>) if and only if <em>x<sub>i<\/sub><\/em> \u2264 <em>x<sub>j<\/sub><\/em> (since <em>n<sub>i<\/sub><\/em>=<em>n<sub>j<\/sub><\/em>=<em>n<\/em>&#8216;) if and only if\u00a0<em>x<sub>i<\/sub><\/em> = <em>x<sub>j<\/sub><\/em> (since the elements of <em>X<\/em> are maximal in <em>Y<\/em>, hence incomparable or equal).\u00a0 Since (<em>x<sub>i<\/sub><\/em>, <em>n<sub>i<\/sub><\/em>)<em><sub>i \u2208 I<\/sub><\/em> is directed (and <em>n<sub>i<\/sub><\/em>=<em>n<\/em>&#8216; for every <em>i<\/em> in <em>I<\/em>), all its elements are equal, hence also equal to their supremum.\u00a0 The fact that <em>x<\/em>[<em>n<\/em>] \u2264 <em>x<sub>i<\/sub><\/em>[<em>n<sub>i<\/sub><\/em>] for some <em>i<\/em> in I (in fact, all), is then obvious.\u00a0 \u2610<\/p>\n<p><strong>Lemma.<\/strong> For every <em>x<\/em> in <em>X<\/em>, (<em>x<\/em>, <em>+\u221e<\/em>) is a maximal element of <em>Y<\/em>&#8216;, and is not finite.<\/p>\n<p><strong>Proof.<\/strong> It is maximal: if (<em>x<\/em>, <em>+\u221e<\/em>) \u2264&#8217; (<em>x<\/em>&#8216;, <em>n<\/em>&#8216;), then <em>n<\/em>&#8216;=<em>+\u221e<\/em> and <em>x<\/em>=<em>x<\/em>[<em>+\u221e<\/em>] \u2264 <em>x<\/em>&#8216;[<em>+\u221e<\/em>]=<em>x<\/em>&#8216;, from which we obtain <em>x<\/em>=<em>x&#8217;<\/em>, since any two comparable elements in <em>X<\/em>, being maximal in <em>Y<\/em>, must be equal.<\/p>\n<p>By definition, (<em>x<\/em>, <em>+\u221e<\/em>) is the supremum of the directed family (<em>x<\/em>, <em>n<\/em>), <em>n<\/em> in <strong>N<\/strong>.\u00a0 Since (<em>x<\/em>, <em>+\u221e<\/em>) \u2264&#8217; (<em>x<\/em>, <em>n<\/em>) for no <em>n<\/em> in <strong>N<\/strong>, (<em>x<\/em>, <em>+\u221e<\/em>) is not finite.\u00a0 \u2610<\/p>\n<h2>Finding <em>X<\/em> as a G<sub>\u03b4<\/sub> subset of <em>Y<\/em>&#8216;<\/h2>\n<p>We are almost through.\u00a0 With <em>Y<\/em>&#8216; constructed as above, for every <em>n<\/em> in <strong>N<\/strong>, the set <em>U<sub>n<\/sub><\/em> defined as the union of the sets \u2191(<em>x<\/em>, <em>n<\/em>), <em>x<\/em> in <em>X<\/em>, is a union of open sets, hence is open.\u00a0 This is the set of elements &#8220;at level <em>n<\/em> or higher&#8221;.<\/p>\n<p>Every element <em>x<\/em> of <em>X<\/em>, equated with (<em>x<\/em>, <em>+\u221e<\/em>), is in every <em>U<sub>n<\/sub><\/em>.\u00a0 Conversely, any element that is in the intersection of the sets <em>U<sub>n<\/sub><\/em> must have a level larger than any natural number, hence be of the form (<em>x<\/em>, <em>+\u221e<\/em>).<\/p>\n<p>This shows that <em>X<\/em>, the set of maximal elements of <em>Y<\/em>&#8216;, is a countable intersection of open subsets, namely:<\/p>\n<p><strong>Proposition.<\/strong>\u00a0 <em>X<\/em> is a G<sub>\u03b4<\/sub> subset of <em>Y<\/em>&#8216;.<\/p>\n<p>We have already seen <a title=\"Ideal models I\" href=\"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/?page_id=681\">last time<\/a> that this was the final touch to the following theorem, due to Keye Martin [1].\u00a0 The key is that <em>Y<\/em>&#8216;, as an algebraic, hence continuous dcpo, is Choquet-complete, that a G<sub>\u03b4<\/sub> subset of a Choquet-complete space is itself Choquet-complete, and that a Choquet-complete metrizable space is completely metrizable.<\/p>\n<p><strong>Theorem (Martin [1]).<\/strong> The metrizable spaces that have an ideal model are exactly the completely metrizable spaces.<\/p>\n<p>Next time, if all goes well, I&#8217;ll explain the connection there is to so-called <em>remainders<\/em> [2].<\/p>\n<p style=\"text-align: right;\">\u2014 <a href=\"https:\/\/www.lsv.ens-paris-saclay.fr\/~goubault\/?l=en\" rel=\"attachment wp-att-993\">Jean Goubault-Larrecq<\/a>\u00a0(January 3rd, 2016)<img loading=\"lazy\" decoding=\"async\" class=\"wp-image-993 alignright\" src=\"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/wp-content\/uploads\/2016\/08\/jgl-2011.png\" alt=\"jgl-2011\" width=\"32\" height=\"44\" \/><\/p>\n<ol>\n<li>Keye Martin. \u00a0<a title=\"Ideal models of spaces\" href=\"https:\/\/www.sciencedirect.com\/science\/article\/pii\/S0304397502006989\">Ideal models of spaces<\/a>. \u00a0<a title=\"Go to Theoretical Computer Science on ScienceDirect\" href=\"https:\/\/www.sciencedirect.com\/science\/journal\/03043975\">Theoretical Computer Science<\/a>,\u00a0<a title=\"Go to table of contents for this volume\/issue\" href=\"https:\/\/www.sciencedirect.com\/science\/journal\/03043975\/305\/1\">Volume 305, Issues 1\u20133<\/a>, 18 August 2003, Pages 277\u2013297.<\/li>\n<li>Hoffmann, R.-E. On the sobrification remainder <sup>s<\/sup><em>X<\/em> \u2212 <em>X<\/em> . Pacific Journal of Mathematics, 83(1), 1979, pages 145\u2013156.<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"<p>Recall that an ideal domain is a dcpo where every non-finite element is maximal.\u00a0 All ideal domains are algebraic and first-countable.\u00a0 The notion, and all the results presented here, are due to Keye Martin [1]. Last time, we have seen &hellip; <a href=\"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/?page_id=808\">Continue reading <span class=\"meta-nav\">&rarr;<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"open","ping_status":"open","template":"","meta":{"_crdt_document":"","footnotes":""},"class_list":["post-808","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/pages\/808","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=808"}],"version-history":[{"count":25,"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/pages\/808\/revisions"}],"predecessor-version":[{"id":5952,"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=\/wp\/v2\/pages\/808\/revisions\/5952"}],"wp:attachment":[{"href":"https:\/\/projects.lsv.ens-paris-saclay.fr\/topology\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=808"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}