MrBayes stop with no reason

[ziy900409]

Hi I'm running MrBayes on Windows. The software were stop at certain generations (usually beyond 50,000,000 generation), then i restart and continuous previous data (mcmc append = yes) and than the software were continuous 20000~50000 generations than it stop again.

I tried to used different version (3.2.7 and 3.2.6) and different computer (cpu i7-4790 and i7-9700k) with same data (ntax = 207, nchar = 4177, 12 partitions ), but it still happened.

Although the software stop, it still can see MrBayes used cpu power on Task Manager.

Hope someone give me any idea. Thank you.

[ronquist]

A possible reason is that the convergence diagnostic is running out of memory. You can run the convergence diagnostics after the run has completed instead, by relaunching mrbayes and then running the sumt and sump commands on your output files. Just use mcmc mcmcdiagn=no during the analysis run, and see if it helps. If it does not, then please rerun your analysis with a fixed seed and swapseed. You do this by running set seed=1 swapseed=1, before you read in the data and run your analysis. Make sure you can reliably repeat the bug, and then share the datafile with us, and we will try to locate the problem.

[ziy900409]

Thanks for your suggestion.

I tried your suggestion, both of them look didn't work. When analysis larger than 7 milion generations, the analysis were freeze. But, there has strange point is if i using mpi version, the analysis were execute smoothly to finish. Could you provide me email and let me send you the information?

I using windows version, Mac version and linux version (ubuntu 16) all with 16 Gb ram. The analysis run independent 3 time in windows 10 and mac 10.15.5, and i setting 3 different files. (1) mcmc mcmcdiagn=no (run larger 7 milion generations, the analysis were freeze (2) set seed=1 swapseed=1 (can run to 8.5 milion gererations before the analysis freeze (3) set seed=1 swapseed=1, mcmc mcmcdiagn=no (result between (1) and (2)

MrBayes setting

[ypchan]

@ziy900409 I am facing same problem using MrBayes v3.2.7 MPI version. I set the generation to 100000000, it always stop at certain generations(usually far smaller than 100000000). Evne I restart the analysis using appen=yes, but it will stop again. If I want to complete my analysis, I need restart several times. Could your experience?

Thanks

[ziy900409]

@ypchan I meet the same problem, and I still can't fix it. I success running whole process once time. Then I used BEAST to analysis and compare the result. The result has high consistency, therefore I using BEAST for after analysis.

[tbauma1]

Same problem, though it seems to be stopping at random points. I am trying to run the tutorial and provided dataset and it keeps stopping. This is the Win32 3.2.7 version. The last time I used the software was July of 2019, with no problems.

UPDATE: Reloaded the program, and now runs fine.