MrBayes is a program for Bayesian inference and model choice across a wide range of phylogenetic and evolutionary models. For documentation and downloading the program, please see the home page:
I'm trying to append my files to add more generations to my run. I noticed that even though MrBayes is saying that it's using samples from the previous analysis, the log likelihood values of the last generation in the previous run do not correspond to the starting values of the next run. For example, the previous run stopped at generation 296520000 with the following values:
When I tried to rerun the analysis, the log files started at different values:
Using samples up to generation 296520000 from previous analysis.
Initial log likelihoods and log prior probs for run 1:
Chain 1 -- -50739.775080 -- -1083.588239
Chain 2 -- -50788.915885 -- -1150.966939
Chain 3 -- -50813.104108 -- -1160.667864
Chain 4 -- -50832.016053 -- -1159.569183
There are 4 more chains on the other processor
Using a relative burnin of 25.0 % for diagnostics
Chain results (continued from previous run; 300000000 generations requested):
296521000 -- [-50736.351] (-50792.065) (-50805.992) (-50806.636) [...4 remote chains...] -- 4:49:55
296522000 -- [-50745.050] (-50771.276) (-50812.554) (-50801.063) [...4 remote chains...] -- 5:18:49
296523000 -- [-50739.118] (-50770.495) (-50809.854) (-50773.944) [...4 remote chains...] -- 5:28:23
296524000 -- [-50757.580] (-50776.478) (-50806.958) (-50773.996) [...4 remote chains...] -- 5:18:38
296525000 -- [-50762.601] (-50770.814) (-50801.990) (-50770.029) [...4 remote chains...] -- 5:24:20
Average standard deviation of split frequencies: 0.088097
296526000 -- [-50762.517] (-50773.486) (-50777.522) (-50764.763) [...4 remote chains...] -- 7:23:54
296527000 -- [-50772.201] (-50772.520) (-50800.832) (-50780.271) [...4 remote chains...] -- 7:09:59
296528000 -- [-50755.485] (-50767.253) (-50813.479) (-50791.390) [...4 remote chains...] -- 6:52:18
296529000 -- [-50761.428] (-50758.620) (-50819.173) (-50791.919) [...4 remote chains...] -- 6:44:57
296530000 -- [-50762.210] (-50763.378) (-50828.884) (-50795.448) [...4 remote chains...] -- 6:33:16
Average standard deviation of split frequencies: 0.088092
If I look at the run1.p file, it shows that the cold chain value for generation 296520000 was -3.523446e+04 and for generation 296530000 was -5.076221e+04. I find it weird that the value went down that fast.
I'm using the mpi version for running this analysis. Any possible reason why this is happening?
I'm trying to append my files to add more generations to my run. I noticed that even though MrBayes is saying that it's using samples from the previous analysis, the log likelihood values of the last generation in the previous run do not correspond to the starting values of the next run. For example, the previous run stopped at generation 296520000 with the following values:
When I tried to rerun the analysis, the log files started at different values:
If I look at the run1.p file, it shows that the cold chain value for generation 296520000 was -3.523446e+04 and for generation 296530000 was -5.076221e+04. I find it weird that the value went down that fast.
I'm using the mpi version for running this analysis. Any possible reason why this is happening?
Thanks!