Computation of loss fraction values is not reproducible
Bug #1168272 reported by
matley
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenQuake Engine |
Fix Released
|
Medium
|
Michele Simionato | ||
Bug Description
Computing several times the loss fractions with the configuration (and seeds) might give different results.
We have probably missed some "ORDER BY" clause that changes the calls ordering (and then the results).
This is an issue in QA tests for probabilistic calculators where we do not want to wait for convergence such that we strongly rely on randomly generated numbers (so on the seed).
If we don't want to fix this problem, we might need to update those QA tests by making them using precomputed hazard.
| Changed in oq-engine: | |
| importance: | Undecided → Medium |
Revision history for this message
| Michele Simionato (michele-simionato) wrote | #1 |
| Changed in oq-engine: | |
| status: | New → Fix Committed |
| milestone: | none → 1.0.1 |
| Changed in oq-engine: | |
| status: | Fix Committed → Fix Released |
| assignee: | nobody → Michele Simionato (michele-simionato) |
To post a comment you must log in.
Is this still relevant?