Open mariodosreis opened 2 months ago
Hi Mario,
Have you tried to download the source code from the dev
branch and use it to compile MCMCtree
?
It seems that there are some invalid memory access problems (conditional jumps depending on uninitialized values). It is possible those values are automatically initialised by Linux/Windows binaries to 0 and so there is no segfault, but arm64 initialises them differently (or is stricter) because it is a risc processor. Ziheng fixed the source code in the dev
branch but has not yet been merged with the main
branch. The bug that affects only newer Macs was related to the use of strcpy(t,s);
, which copes source string s
to target string t
, when s
and t
are the same string at the same address.
Let me know if using the compiled version using the source code in the dev
branch fixes the issue!
All the best, Sandy
I run a quick test using the dev version on both the HIV and Flu data and the error does not occur, so it appears fixed! M.
The examples in
examples/TipDate.HIV2/
andexamples/TipDate.FluH1/
do not work. When trying any of the five mcmctree control files in those directories, the following error is printed on the screen:Trace/BPT trap: 5
and the MCMC run is aborted before starting. This was tested on a MacBook air M2.
Some of the control files were tested with mcmctree 4.9j and they appear to be running correctly.