v0.16.0 error: crashed in MultiNest (corrupted double-linked list)

[kuanghan]

Hi Jacopo,

Here is the tar ball containing the log file, catalog, and filter files from my latest BEAGLE run of version 0.16.0. I think these are enough to recreate the error. Thanks!

BEAGLE_kuang_170803.zip

[jacopo-chevallard]

I haven't tested, but I think that the culprit is this line in the parameter file

SF_PARAMETER  = name:specific_sfr   type:fitted   order_priority:2  prior:distribution:uniform  prior:range:[-10.,10.]

since you allow specific star formation rates of 10^10 yr^-1, which is way too much ! ;)

can you try running Beagle changing the prior to a more normal range

SF_PARAMETER  = name:specific_sfr   type:fitted   order_priority:2  prior:distribution:uniform  prior:range:[-10.,-7.]

let me know if it solves!

[kuanghan]

Thanks for spotting the sSFR range! What was I thinking :/ Well, after changing sSFR prior to range between -10. and -7., now I get a different error that seems to complain about the wavelength range of templates and filters not matching each other.

run_beagle_a2744.log.txt

[jacopo-chevallard]

I think that the problem is in the filter transmission curves and wl range of the templates. By default, HST curves include a lot of points with zero transmission, since they are all given on the same wl array, independently from the non-zero transmission region of each filter.

You can appreciate this by comparing the HST curves (for instance for F435W) in the EAZY_FILTERS.RES file with those in the EAZY_FILTERS.RES.fits included in your zip archive.

I would suggest you to trim the curves to reasonable wl ranges (excluding the wl bins with near-to-zero or negative transmission) before creating your filter throughputs FITS file, and to comment the following line in the parameter file

SHRINK TEMPLATES WL RANGE = 900 25000

[kuanghan]

Hi Jacopo,

Thanks for the tips, BEAGLE has been running since last Friday. Do you know how long it might take to run on a catalog of ~ 2000 objects? My macbook has Intel i7 2.9 GHz with 8 GB memory. I'm just wondering if I should kill the job now and reduce the number of parameters I want to fit.

[jacopo-chevallard]

Great, closing this, I'll answer by mail your other question!