Huge Markov chains

Article View: sci.math.num-analysis

Article #185

Back to sci.math.num-analysis Previous Next

#185

From: Nicola Muscettola
Date: Tue, 04 Aug 1992 15:16

15 lines
1247 bytes

I am looking at a problem that requires to find the stationary distribution
of a discrete Markov chain with order 10e+04 states with a transition
matrix that is only about 4% full (and that, in fact, could be easily generated
at run time, given the very regular structure of the chain).

I would appreciate if anybody had information on fast numerical
methods to solve this sort of problem.

I would appreciate e-mail responses. I will post an edited summary of
the responses on the boards.

-- Nicola Muscettola
   The Robotics Institute
   Carnegie Mellon University
   Pittsburgh, PA 15217

Technical Details

Message-ID: <1992Aug04.151602.5141@cs.cmu.edu>

Path:

arc1-ca.pugleaf.net!sparky!uunet!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!crabapple.srv.cs.cmu.edu!ISL1.RI.CMU.EDU!mus

Show Full Headers

Complete Article Headers

Path: arc1-ca.pugleaf.net!sparky!uunet!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!crabapple.srv.cs.cmu.edu!ISL1.RI.CMU.EDU!mus
From: mus+@RI.CMU.EDU (Nicola Muscettola)
Newsgroups: sci.math.num-analysis
Subject: Huge Markov chains
Keywords: discrete markov chains, stationary distribution, eigensystems solution
Message-ID: <1992Aug04.151602.5141@cs.cmu.edu>
Date: 4 Aug 92 15:16:02 GMT
Article-I.D.: cs.1992Aug04.151602.5141
Organization: School of Computer Science, Carnegie Mellon
Lines: 15
Nntp-Posting-Host: isl1.ri.cmu.edu
To: outbound+netnews.sci.math.num-analysis@andrew.cmu.edu, outbound+netnews.sci.math.stat@andrew.cmu.edu