search

JifengJiang / beast-mcmc

Automatically exported from code.google.com/p/beast-mcmc
GNU Lesser General Public License v2.1
0 stars 0 forks source link

Error random local clock #733

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
Hi

I'm using BEAST 1.8 and it works fine when I use a strickt clock or lognormal 
relaxed clock. However, when I want to use the random local clock I get an 
error (see below)

Does anyone now why this happens and how to fix the problem?

Thanks

May 27, 2014 10:19:09 AM dr.inference.markovchain.MarkovChain runChain
SEVERE: State was not correctly calculated after an operator move.
Likelihood evaluation: -47803.09839741052
Full Likelihood evaluation: -47368.10911721017
Operator: scaleOperator(localClock.relativeRates [0.75, 1.3333333333333333] 
scale(localClock.relativeRates)
May 27, 2014 10:19:09 AM dr.inference.markovchain.MarkovChain runChain
SEVERE: State was not correctly calculated after an operator move.
Likelihood evaluation: -47737.04515278441
Full Likelihood evaluation: -47301.47682139139
Operator: scaleOperator(localClock.relativeRates [0.75, 1.3333333333333333] 
scale(localClock.relativeRates)
1.0426166666666667 minutes 
Exception in thread "Thread-6" java.lang.RuntimeException: One or more 
evaluation errors occurred during the test phase of this
run. These errors imply critical errors which may produce incorrect
results.
    at dr.inference.markovchain.MarkovChain.runChain(Unknown Source)
    at dr.inference.mcmc.MCMC.chain(Unknown Source)
    at dr.inference.mcmc.MCMC.run(Unknown Source)
    at java.lang.Thread.run(Thread.java:695)

Original issue reported on code.google.com by fdigiall...@gmail.com on 27 May 2014 at 12:49

GoogleCodeExporter commented 8 years ago 
Unable to reproduce error and fix unless a (minimal) XML and seed are provided.

Original comment by msuch...@gmail.com on 5 Aug 2014 at 1:44

GoogleCodeExporter commented 8 years ago 
I think this is the same as Issue 756. It is a problem with multi-threading 
when using certain clock models on multiple partitions (including codon 
partitioning). Interim solution is to use '-threads 1' on the command line but 
this will be less efficient on multi-core machines.

Original comment by ramb...@gmail.com on 3 Feb 2015 at 1:24