I found inconsistent lnL results when using EX2 model (and perhaps EX3 and a few other mixture sub models) according to fixed bls (and thus also tree topology) in tree construction.
which gave me the best lnl as -19010.645 as seen in the .log file.
Then I used the following R code to calculate the lnl using the .sitelog file and got -19201.6:
a <- readLines("1.sitelh")[2]; sum(na.omit(as.numeric(strsplit(a, " ")[[1]])))
The files that i used for testing are attached here. ex2_blfix.tar.zip
I actually did some calculations using my own script and found that the sitelh file seems correct (but probably my calculation is incorrect). Would you pls mind having a look at it? Thanks!
BTW, I have another suggestion that the iqtree google grp seems unaccessible but i think there are many interesting threads there so might be worth reopening or archiving at another place. thanks!
Dear Minh,
I found inconsistent lnL results when using EX2 model (and perhaps EX3 and a few other mixture sub models) according to fixed bls (and thus also tree topology) in tree construction.
The cmd is:
which gave me the best lnl as
-19010.645as seen in the.logfile.Then I used the following R code to calculate the lnl using the
.sitelogfile and got-19201.6:The files that i used for testing are attached here. ex2_blfix.tar.zip
I actually did some calculations using my own script and found that the sitelh file seems correct (but probably my calculation is incorrect). Would you pls mind having a look at it? Thanks!
BTW, I have another suggestion that the iqtree google grp seems unaccessible but i think there are many interesting threads there so might be worth reopening or archiving at another place. thanks!
Best, Sishuo