search

barbarabarbosa / beast-mcmc

Automatically exported from code.google.com/p/beast-mcmc
0 stars 0 forks source link

GMRF Skyride - "skyride.popSize" and "skyride.precision" stuck on starting values, "gmrfBlockUpdateOperator" changes are never accepted #249

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
What steps will reproduce the problem?
1. Run an xml file for a GMRF skyride analysis
2.
3.

What is the expected output? What do you see instead?

The log file should show that the "skyride.popSize" and "skyride.precision"
values are changing as the chain is running. They do not. They are fixed on
the starting values and never update.

What version of the product are you using? On what operating system?
BEAST v1.5.2 on MacBook Pro Core 2 Duo 2.33 GHz, OSX 1.6.2. Files
constructed using Beauti 1.5.2

Please provide any additional information below.

The problem seems to happen most often with small datasets. I originally
thought this was because the value for "rootHeight" on the starting tree
was <1.0. I changed this for one file, and it worked, but this solution
didn't work for other files. I am attaching two files that don't work
(asp.coi.BA.gmrf.fixed.calib.xml and x7.xml, the latter is the example file
included in the Beast distribution) and one that does work
(asp.rag1.BA.gmrf.fixed.calib.xml).

Original issue reported on code.google.com by ron.ey...@gmail.com on 18 Nov 2009 at 12:23

Attachments:

GoogleCodeExporter commented 8 years ago

Original comment by ramb...@gmail.com on 25 Nov 2009 at 11:43

GoogleCodeExporter commented 8 years ago

Original comment by ramb...@gmail.com on 25 Nov 2009 at 11:45

GoogleCodeExporter commented 8 years ago 
This appears to be a starting-value issue.  If I modify 
asp.coi.BA.gmrf.fixed.calib.xml such that the starting log 
pop-sizes all equal -10, I achieve reasonable mixing.   Guessing reasonable 
starting-values is non-trivial and 
we need to think more about this issue.

Original comment by msuch...@gmail.com on 2 Jan 2010 at 7:00

GoogleCodeExporter commented 8 years ago 
Starting values are something that need to be considered by users. This is not 
a software issue so I am closing it.

Original comment by ramb...@gmail.com on 21 Dec 2011 at 10:10

GoogleCodeExporter commented 8 years ago 
Issue 461 has been merged into this issue.

Original comment by ramb...@gmail.com on 21 Dec 2011 at 10:12