Problem 1965 - Large memory usage for high energy primary and bug in bertini-coalescence
Summary: Large memory usage for high energy primary and bug in bertini-coalescence
Status: RESOLVED FIXED
Alias: None
Product: Geant4
Classification: Unclassified
Component: processes/hadronic/models/cascade (show other problems)
Version: 10.3
Hardware: All All
: P4 major
Assignee: Michael Kelsey
URL:
Depends on:
Blocks:
 
Reported: 2017-04-13 19:16 CEST by Andrea Dotti
Modified: 2017-06-13 18:35 CEST (History)
0 users

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this problem.
Description Andrea Dotti 2017-04-13 19:16:04 CEST 
When used at high energy the coalescence code tries combinatorics of all exiting nucleons to form light fragments.
This code has two problems: it uses very large amounts of memory when energy of the primary is very high that can be a problem in MT mode due to a support std::set where the tested combinations are stored.
In addition the code that calculates "hashes" of the candidate light fragments (to be used in aforementioned set) is bugged and the hash function has high probability of collision. The effect of the bug is to suppress light fragments production.

Fortunately after close inspection and testing we have realized the use of the support std::set and of the hash function are not needed and can simply be removed. This solves the memory and the bug at the same time.
Comment 1 Andrea Dotti 2017-04-13 19:28:25 CEST 
fix being tested in hadr-casc-V10-03-00