random number bug in Pelegant 23.1.0

Moderators: michael_borland, soliday

Post Reply
michael_borland
Posts: 1933
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

random number bug in Pelegant 23.1.0

Post by michael_borland » 29 Jun 2010, 08:24

A bug was just discovered in version 23.1.0 of Pelegant related to certain random number generators. In particular, it would affect the SREFFECTS element and any elements (e.g., CSBEND, KQUAD) that have the ISR flag set to 1. This bug does not affect serial elegant or prior versions of Pelegant.

The bug is that the random number generators are identical across all processors. For a case where quantum excitation is not dominant, this won't affect results much. However, if you are, for example, tracking in a ring for many damping times, you'll get very odd results.

The attached file patches the problem for those who build from source. We'll try to post updated versions in a few days.

--Michael
Attachments
drand_oag.c
(3.17 KiB) Downloaded 803 times

michael_borland
Posts: 1933
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: random number bug in Pelegant 23.1.0

Post by michael_borland » 30 Jun 2010, 13:26

Updated versions have been posted (version 23.1.2)

Contrary to the subject of this post, the affected versions are 23.1.0 and 23.1.1

--Michael

michael_borland
Posts: 1933
Joined: 19 May 2008, 09:33
Location: Argonne National Laboratory
Contact:

Re: random number bug in Pelegant 23.1.0

Post by michael_borland » 01 Jul 2010, 09:20

Updated source and executables have been posted on our software page. The new version number is 23.1.2.

--Michael

Post Reply