Killing demes not preventing "simulations did not converge" error
178 views
Skip to first unread message
Kyle Lewald
Mar 11, 2022, 7:40:23 PM3/11/22
to fastsimcoal2
Hi all, I am using fastsimcoal version 2.7.0.5 to estimate parameters of my model from some observed MAF data. The model I am using is attached below.
Basically I have 3 demes, where deme 0 and deme 1 merge into deme 2 at times T0 and T1, and have an exponential growth. deme 2 also has its own growth rate, which starts at T2. As I do not know the ordering of T0, T1, and T2, I am using the "keep" parameter to preserve the current growth rate whenever deme 0 and deme 1 merge into deme 2. I also have migration flowing between deme 0 & 2, as well as 1 & 2, so I am using a single migration matrix and "killing" demes once they have merged to prevent errors (p29 of manual).
Basically I have 3 demes, where deme 0 and deme 1 merge into deme 2 at times T0 and T1, and have an exponential growth. deme 2 also has its own growth rate, which starts at T2. As I do not know the ordering of T0, T1, and T2, I am using the "keep" parameter to preserve the current growth rate whenever deme 0 and deme 1 merge into deme 2. I also have migration flowing between deme 0 & 2, as well as 1 & 2, so I am using a single migration matrix and "killing" demes once they have merged to prevent errors (p29 of manual).
However, when running this estimation 100 times, many of my runs still return various errors, including "Simulations
did not converge. MRCA not found! (2 lineages remaining)", ""munmap_chunk():
invalid pointer", and "free():
invalid next size (fast)".
Is there something wrong with the way I have specified my .tpl or .est files? I saw this bug report (https://groups.google.com/g/fastsimcoal/c/X-W4OVr85JI/m/DAqvss6wAAAJ) that also had issues with exponential growth/migration that resulted in a "free(): invalid next size (fast)" error, but I thought it was corrected in v27.0.5. Is this the same bug?
Best,
Kyle
Kyle
0 new messages