[BACK]Return to ref-hgm.html CVS log [TXT][DIR] Up to [local] / OpenXM / src / hgm / doc

Diff for /OpenXM/src/hgm/doc/ref-hgm.html between version 1.5 and 1.10

version 1.5, 2014/03/26 05:02:18 version 1.10, 2014/05/16 11:30:31
Line 12  the Holonomic Gradient Descent Method  (HGD) </h1>
Line 12  the Holonomic Gradient Descent Method  (HGD) </h1>
   
 <h2> Papers  and Tutorials</h2>  <h2> Papers  and Tutorials</h2>
 <ol>  <ol>
   <li> J.Hayakawa, A.Takemura,
   Estimation of exponential-polynomial distribution by holonomic gradient descent
   <a href="http://arxiv.org/abs/1403.7852"> arxiv:1403.7852</a>
   
   <li> C.Siriteanu, A.Takemura, S.Kuriki,
   MIMO Zero-Forcing Detection Performance Evaluation by Holonomic Gradient Method
   <a href="http://arxiv.org/abs/1403.3788"> arxiv:1403.3788</a>
   
 <li> T.Koyama,  <li> T.Koyama,
 Holonomic Modules Associated with Multivariate Normal Probabilities of Polyhedra,  Holonomic Modules Associated with Multivariate Normal Probabilities of Polyhedra,
 <a href="http://arxiv.org/abs/1311.6905"> arxiv:1311.6905 </a>  <a href="http://arxiv.org/abs/1311.6905"> arxiv:1311.6905 </a>
Line 80  Advances in Applied Mathematics 47 (2011), 639--658,
Line 88  Advances in Applied Mathematics 47 (2011), 639--658,
   
 <h2> Three Steps of HGM </h2>  <h2> Three Steps of HGM </h2>
 <ol>  <ol>
 <li> Find a holonomic system satisfied by the normalizing constant.  <li> Finding a holonomic system satisfied by the normalizing constant.
 We may use computational or theoretical methods to find it.  We may use computational or theoretical methods to find it.
 Groebner basis and related methods are used.  Groebner basis and related methods are used.
 <li> Find an initial value vector for the holonomic system.  <li> Finding an initial value vector for the holonomic system.
 This is equivalent to evaluating the normalizing constant and its derivatives  This is equivalent to evaluating the normalizing constant and its derivatives
 at a point.  at a point.
 This step is usually performed by a series expansion.  This step is usually performed by a series expansion.
 <li> Solve the holonomic system numerically. We use several methods  <li> Solving the holonomic system numerically. We use several methods
 in numerical analysis such as the Runge-Kutta method of solving  in numerical analysis such as the Runge-Kutta method of solving
 ordinary differential equations and efficient solvers of systems of linear  ordinary differential equations and efficient solvers of systems of linear
 equations.  equations.
 </ol>  </ol>
   
 <h2> Software Packages for HGM</h2>  <h2> Software Packages for HGM</h2>
   Most software packages are experimental and temporary documents are found in
   "asir-contrib manual" (auto-autogenerated part), or
   "Experimental Functions in Asir", or "miscellaneous and other documents"
   of the
   <a href="http://www.math.kobe-u.ac.jp/OpenXM/Current/doc/index-doc.html">
   OpenXM documents</a>
   or in <a href="./"> this folder</a>.
   The nightly snapshot of the asir-contrib can be found in the asir page below,
   or look up our <a href="http://www.math.sci.kobe-u.ac.jp/cgi/cvsweb.cgi/">
   cvsweb page</a>.
 <ol>  <ol>
 <li> <a href="http://www.math.kobe-u.ac.jp/OpenXM/Math/hgm/"> hgm package for R </a> for the step 3.  <li> Command line interfaces are in the folder OpenXM/src/hgm
 <li> yang (for Pfaffian systems) , nk_restriction (for D-module integrations),  in the OpenXM source tree. See <a href="http://www.math.kobe-u.ac.jp/OpenXM">
 tk_jack  (for Jack polynomials), ko_fb_pfaffian (Pfaffian system for the Fisher-Bingham system)  OpenXM distribution page </a>.
 are for the steps 1 or 2 and in the  <li> <a href="http://www.math.kobe-u.ac.jp/OpenXM/Math/hgm/"> hgm package for R </a> (hgm_*tar.gz, hgm-manual.pdf) for the step 3.
 <a href="http://www.math.kobe-u.ac.jp/Asir"> asir-contrib </a>  <li> The following packages are
 <li> nk_fb_gen_c is a package to generate a C program to perform  for the computer algebra system
 maximal Likehood estimates for the Fisher-Bingham distribution by HGD (holonomic gradient descent)  <a href="http://www.math.kobe-u.ac.jp/Asir"> Risa/Asir</a>.
 It is in the  They are in the asir-contrib collection.
 <a href="http://www.math.kobe-u.ac.jp/Asir"> asir-contrib </a>  <ul>
   <li> yang.rr (for Pfaffian systems) ,
   nk_restriction.rr (for D-module integrations),
   tk_jack.rr  (for Jack polynomials),
   ko_fb_pfaffian.rr (Pfaffian system for the Fisher-Bingham system),
   are for the steps 1 or 2.
   <li> nk_fb_gen_c.rr is a package to generate a C program to perform
   maximal Likehood estimates for the Fisher-Bingham distribution by HGD (holonomic gradient descent).
   <li> ot_hgm_ahg.rr (HGM for A-distributions, very experimental).
   </ul>
 </ol>  </ol>
   
 <h2> Programs to try examples of our papers </h2>  <h2> Programs to try examples of our papers </h2>
Line 111  It is in the 
Line 138  It is in the 
 <li> <a href="http://www.math.kobe-u.ac.jp/OpenXM/Math/Fisher-Bingham-2"> d-dimensional Fisher-Bingham System </a>  <li> <a href="http://www.math.kobe-u.ac.jp/OpenXM/Math/Fisher-Bingham-2"> d-dimensional Fisher-Bingham System </a>
 </ol>  </ol>
   
 <pre> $OpenXM: OpenXM/src/hgm/doc/ref-hgm.html,v 1.4 2014/03/24 21:03:55 takayama Exp $ </pre>  <pre> $OpenXM: OpenXM/src/hgm/doc/ref-hgm.html,v 1.9 2014/05/15 07:34:05 takayama Exp $ </pre>
 </body>  </body>
 </html>  </html>
Legend:
Removed from v.1.5  
changed lines
  Added in v.1.10
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>