CoupledMCMC is a BEAST 2 package, which you can install through the
package manager that comes with BEAST.
Choose CoupledMCMC
from the list of packages.
If you want to run a Standard analysis (i.e. an analysis using the Standard template for MCMC analyses), you can choose the CoupledMCMC template instead under the File => Templates => CoupledMCMC
menu. Then you can set up the analysis as you would a Standard analysis, except the MCMC tab is replaced by the CoupledMCMC tab.
If you want to run another analysis than a Standard one, you still can use CoupledMCMC per below.
After you installed the CoupledMCMC
package (version 0.1.5 or better), the MCMC2CoupledMCMC
app becomes available in the app launcher.
- Create MCMC analysis in BEAUti with any of the available templates, save as
mcmc.xml
Now there are 2 ways to proceed:
2a. from a terminal, run
/path/to/beast/bin/applauncher MCMC2CoupledMCMC -xml mcmc.xml -o mc3.xml
This creates a file
mc3.xml
containing a CoupledMCMC analysis with the same model/operators/loggers etc as themcmc.xml
analysis.2b. from BEAUti, use menu
File > Launch apps
, selectMCMC to Coupled MCMC converter
from the available apps, fill in form and click OK
In order to set up a pre-prepared xml to run with coupled MCMC, open the *.xml
and change the MCMC line in the xml.
To do so, go to the line with:
<run id="mcmc" spec="MCMC" chainLength="....." numInitializationAttempts="....">
To have a run with coupled MCMC, we have to replace that one line with:
<run id="mcmc" spec="coupledMCMC.CoupledMCMC" chainLength="100000000" storeEvery="1000000" deltaTemperature="0.025" chains="2" resampleEvery="10000">
chainLength="100000000"
defines for how many iterations the chains is rundeltaTemperature="0.025"
defines the temperature difference between the chain n and chain n-1. This value should be changed such that the acceptance probability of a swap is between 0.25 and 0.6chains="2"
defines the number of parallel chains that are run. The first chain is the one that explores the posterior just like a normal MCMC chain. All other chains are what's called heated. This means that MCMC moves of those chains have a higher probability of being accepted. While these heated chains don't explore the posterior properly, they can be used to propose new states to the one cold chain.
Müller NF, Bouckaert RR. Adaptive Metropolis-coupled MCMC for BEAST 2. PeerJ. 2020 Sep 16;8:e9473. DOI:10.7717/peerj.9473 BioRxiv preprint (2019): 603514. (abstract, pdf);
Parallel Metropolis coupled Markov chain Monte Carlo for Bayesian phylogenetic inference https://academic.oup.com/bioinformatics/article/20/3/407/186341