bug issue

[Kiran]

Hi Mr Hey

I am trying to run the 0811 version of IMa on RedHat Linux 5.7 64 bit,
but it crashes after a day or two with an error messege that the
memory is corrupted. I am running the same input file that I used on
IMa 411.
The command line I am using is:

-q 25 -m 75 -t 0.5 -b 500000 -l 24.0 -d 50 -hfg -hk 30 -hn 30 -ha 0.95
-hb 0.99 -s 2354657 -u 2 -p25 -j23 -o outfile.txt

I am sending you the input file in your email id. I would appreciate
if you can please help finding the cause of the crash: input file,
system or the program?

Below is the error messege it generates:



glibc detected *** IMa2.20110824: malloc(): memory corruption:
0x00000000059e8bd0 ***
======= Backtrace: =========
/lib64/libc.so.6[0x2b109bf9f0fe]
/lib64/libc.so.6(__libc_calloc+0xcd)[0x2b109bfa0aad]
IMa2.20110824[0x40eb0c]
IMa2.20110824[0x41179a]
IMa2.20110824[0x415dbc]
IMa2.20110824[0x416b08]
IMa2.20110824[0x418107]
/lib64/libc.so.6(__libc_start_main+0xf4)[0x2b109bf49994]
IMa2.20110824[0x401729]
======= Memory map: ========
00400000-00469000 r-xp 00000000 00:24
24052502 /common/bin/IMa2.20110824
00668000-00669000 rwxp 00068000 00:24
24052502 /common/bin/IMa2.20110824
00669000-012cf000 rwxp 00669000 00:00 0
038db000-07765000 rwxp 038db000 00:00
0 [heap]
2b109ba8b000-2b109baa7000 r-xp 00000000 00:14
20252386 /lib64/ld-2.5.so
2b109baa7000-2b109bab1000 rwxp 2b109baa7000 00:00 0
2b109bca7000-2b109bca8000 r-xp 0001c000 00:14
20252386 /lib64/ld-2.5.so
2b109bca8000-2b109bca9000 rwxp 0001d000 00:14
20252386 /lib64/ld-2.5.so
2b109bca9000-2b109bd2b000 r-xp 00000000 00:14
20252401 /lib64/libm-2.5.so
2b109bd2b000-2b109bf2a000 ---p 00082000 00:14
20252401 /lib64/libm-2.5.so
2b109bf2a000-2b109bf2b000 r-xp 00081000 00:14
20252401 /lib64/libm-2.5.so
2b109bf2b000-2b109bf2c000 rwxp 00082000 00:14
20252401 /lib64/libm-2.5.so
2b109bf2c000-2b109c07a000 r-xp 00000000 00:14
20252393 /lib64/libc-2.5.so
2b109c07a000-2b109c27a000 ---p 0014e000 00:14
20252393 /lib64/libc-2.5.so
2b109c27a000-2b109c27e000 r-xp 0014e000 00:14
20252393 /lib64/libc-2.5.so
2b109c27e000-2b109c27f000 rwxp 00152000 00:14
20252393 /lib64/libc-2.5.so
2b109c27f000-2b109c4d8000 rwxp 2b109c27f000 00:00 0
2b109c4fe000-2b109c50b000 r-xp 00000000 00:14
20250743 /lib64/libgcc_s-4.1.2-20080825.so.1
2b109c50b000-2b109c70b000 ---p 0000d000 00:14
20250743 /lib64/libgcc_s-4.1.2-20080825.so.1
2b109c70b000-2b109c70c000 rwxp 0000d000 00:14
20250743 /lib64/libgcc_s-4.1.2-20080825.so.1
2b10a0000000-2b10a0021000 rwxp 2b10a0000000 00:00 0
2b10a0021000-2b10a4000000 ---p 2b10a0021000 00:00 0
7fffb488d000-7fffb48a2000 rw-p 7ffffffe9000 00:00
0 [stack]
7fffb4997000-7fffb499a000 r-xp 7fffb4997000 00:00
0 [vdso]
ffffffffff600000-ffffffffffe00000 ---p 00000000 00:00
0 [vsyscall]


Thanks!
Kiran

[jhey]

My first guess is that with your high migration priors you are
overrunning the migration arrays. there were some changes in the way
they were implemented in the newer version

where do you think your 4Nu values are at? and your 2NM values?
with a prior set to -q25 and -m75 you can techically accomodate a 2NM
value of 937.5 which is panmixia. In general you want to do runs
with priors on migration such that the 2NM values are not much above
1.

jhey

[Kiran]

Hi Mr. Hey,

We are trying to determine whether gene flow is necessary to explain
the
genetic similarity between two morphologically and ecologically
distinct
species. The problem is that all evidence is pointing to a very recent
divergence. The two species are pretty much genetically
indistinguishable based on mtDNA & 5 nuclear introns (Fst is low, and
for the most part, non-significant). Our gut feeling is that a recent
divergence, with no migration, is sufficient to explain the genetic
similarity, but we are hoping to formally test the hypothesis.

Based on crude estimates of population size, 4Nu should be around
500,000 individuals for species 1 & 100,000 individuals for species 2.

[jhey]

again, what do you think 4Nu is at (e.g. just using watterson's
estimator or pi) for your loci (average across loci, but not per
base pair).

then what value of an upper bound on migration (M/u) would correspond
to 2NM = 1 ? this would be useful starting point for an upper
bound on migration

jhey

[Rita]

Hi Kiran

Could you share your IMa 64-bit compiled version?

Thanks!

Rita

The email is rca...@ualg.pt