pxbd time

[fephyfofum]

pxbd time

[josephwb]

Death was cast to an int in processing cmd line args; changed to float. I don't think this was the problem responsible for neg branch lengths, but I cannot recreate the problem now. However, do get seg fault on some parameter combinations (e.g. ./pxbd -e 10 -b 0.5 -t 1). Perhaps allow -e OR -t, not both.

[blackrim]

agreed about OR. hm that could be it. lets close and open again when the problem comes back.

[blackrim]

Still getting a bunch of negative branch lengths

[josephwb]

Do you have an example that will generate this (b, d, e/t, and seed)?

[blackrim]

how do you mean? birth death extant time and seed? can only do birth death
time seed or birth death extant seed
should be pxbd -b 1 -d 0.1 -t 5 -s 1 or pxbd -b 1 -d 0.1 -e 5 -s 1

On Tue, Jun 23, 2015 at 10:55 AM, Joseph W. Brown [email protected]
wrote:

Do you have an example that will generate this (b, d, e/t, and seed)?

—
Reply to this email directly or view it on GitHub
#2 (comment).

[josephwb]

Yeah, the "e/t" was supposed to mean e or t.

I just meant a combination that generates negative branch lengths so it could be identified and fixed.

[blackrim]

Oh, I see. I don't have an example that does that.

On Tue, Jun 23, 2015 at 11:27 AM, Joseph W. Brown [email protected]
wrote:

Yeah, the "e/t" was supposed to mean e or t.

I just meant a combination that generates negative branch lengths so it
could be identified and fixed.

—
Reply to this email directly or view it on GitHub
#2 (comment).

[josephwb]

Ok, I can set up a loop to try and find an example.

[josephwb]

I tried a bunch of iterations of:

./pxbd -b 0.5 -d 0.45 -e 10000 -o oot.tre && grep ":-" oot.tre

(including different parameter values) and cannot recreate the problem.

[josephwb]

pxbd now has a nreps option:

./pxbd -h
Birth-death simulator.

Usage: pxbd [OPTION]... 

 -e, --extant=INT    number of extant species, alt to time
 -t, --time=INT      depth of the tree, alt to extant
 -b, --birth=DOUBLE  birth rate, default=1
 -d, --death=DOUBLE  death rate, default=0
 -n, --nreps         number of replicates, default=1
 -o, --outf=FILE     output file, stout otherwise
 -s, --showd         show dead taxa
 -x, --seed=INT      random number seed, clock otherwise
     --help          display this help and exit
     --version       display version and exit

This should help in the search for negative edge lengths. However, I've simulated a ton and have not hit anything yet.

[josephwb]

I think this is behaving correctly. Closing.