Java Modelling Tools / Bugs / #57 Small perturbation errors in JMVA

#57 Small perturbation errors in JMVA

Milestone: JMVA

Status: closed-fixed

Owner: Giuliano Casale

Labels: None

Priority: 5

Updated: 2022-12-22

Created: 2015-07-12

Creator: Giuliano Casale

Private: No

Normalizing constant algorithms use small perturbation at scale 1e-6 that produce results that differ at this scale with the MVA results. JMVA could use a smaller scale to avoid this inconsistency.

Discussion

Giuliano Casale - 2016-04-03

status: open --> open-fixed
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Giuliano Casale - 2016-04-03

Scale 1e-5 seems to address the issue for RECAL and CoMoM. Reduction of random demands precision to 1e-3 makes it more unlikely that the problem is noticed by end users.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Giuliano Casale - 2016-07-30

Group: common --> JMVA
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Giuliano Casale - 2022-12-22

status: open-fixed --> closed-fixed
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Giuliano Casale - 2022-12-22

Only CoMoM is available as of 1.2.3-BETA1 so the issue is no longer open.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Small perturbation errors in JMVA

Group

Searches

Help

#57 Small perturbation errors in JMVA

Discussion