Tuesday, November 17, 2009

Exceeded Maxlist

I am getting a strange error when I run the 3D Correlation function. The error is from Martin White's meshing algorithm (nearmesh):

"List length exceeded MaxList=234700"

It think it happens when a point is near any mesh within the grid. But I am not sure exactly, I didn't write this section of the code, and I've never gotten this error before.

I wrote Alexia about it, and I am hoping she'll have some experience with this problem.

Below is the email I send her today:

From: Jessica Ann Kirkpatrick
Date: Tue, Nov 17, 2009 at 11:24 PM
Subject: halo files and numbers
To: Alexia Schulz

the file I am using is from here: http://mwhite.berkeley.edu/Mock/

It was under a section that used to be there called LRG but now seems to be gone. See this blog entry.

The number of halos in the file is 102496.

When I run halolist with mmin = 5e10 I get 3811293 galaxies (for photometric data set)
When I run halolist with mmin = 5e11 I get 378932 galaxies (for spectroscopic data set)

output when I run halolist is below:

In [324]: #make galaxies
In [325]: mmin=5.e10 #for gal
In [326]: h=gals.halolist(mockLRGfile,mmin)
1.30144190788 seconds wall time to readin
1.3 seconds processor time to readin
33.0253970623 seconds wall time
30.84 seconds processor time

In [327]: g=gals.galist(h,nbins,outfile=photofile)
halo number is 102496
number of halos with 1 or more gals 102496
176.569406033 seconds wall time
168.58 seconds processor time

In [328]: mmin=5.e11 #for spec
In [329]: h=gals.halolist(mockLRGfile,mmin)
10.7454509735 seconds wall time to readin
1.66 seconds processor time to readin
37.9757530689 seconds wall time
27.46 seconds processor time

In [330]: g=gals.galist(h,nbins,outfile=specfile)
halo number is 102496
number of halos with 1 or more gals 101718
27.7532730103 seconds wall time
26.72 seconds processor time

I'm getting the error:
List length exceeded MaxList=234700
on one of my runs, has this ever happened to you?

It seems to be coming from deep in the calculation in the function meshnear.

I am getting this error when I try to run on redshift slices which are further away (and thus have higher statistics).

Tomorrow the BigBOSS meeting has a break from 12-1. Can I call you sometime in that window?


I thought perhaps a problem is that the randoms are not overlaping with the data (and so the nearmesh would mess up because the randoms are not in the same region as the data).

And this seems to be part of the problem:

Need to check these to see where the problem is. I think there might be a degrees vs radians issue. I'll look at this tomorrow.

No comments:

Post a Comment