|Message: Re: Irreproducible results||Not Logged In (login)|
Click on the Forum title, e.g. on the "Forums by Category" page, to read a sequence of postings to the Forum and its threads all in one page. If you are only interested in one thread or the thread following a specific posting, click the thread or the posting, which takes you to a smaller page, which contains only the part you are interested in and may be easier to navigate.
Messages are "chained" if there are only replies at the first level, i.e. 1/1.html, 1/1/1.html etc. In case of "chained" messages the message number is replaced by the icon and there is no indentation.
Inline: Display the subject line only or also the text of the posting(s); for the choice "All" the "Outline" choices are switched off.
|1||0||1||no text / full text of posting|
|2||1||All||text for level 1 only / text for All postings|
Outline: Choose the depth of the posting thread, successive toggle controls provide increasing detail.
|1||2||1||2 levels / 1 level (original posting)|
|2||3||2||3 levels / 2 levels|
|3||3||All||3 levels / all levels (all postings)|
Dear Tom, we have been evaluating your bug report carefully. We have decided to mark it as "won't fix" because we believe this is not the source of the irreproducibility you describe (in addition the change you propose will introduce more non-reproducibility according to our definition, some details at: https://twiki.cern.ch/twiki/bin/view/Geant4/MultiThreadingTaskForce#Reproducibility).
I understand you are using G4 10.1 without multi-threading. This configuration is checked periodically for reproducibility, and as I mentioned, the main production physics lists (with the exclusion of Radioactive Decay and NeutronHP, on both of which we are working) are reproducible: we check both strong and weak reproducibility (e.g. starting again job with the same random number and reproducing a specific event in the middle of a job) and we do not see problems.
I also understand that you are using your own MPI wrapper and the reproducibility issue you report is between an MPI-enabled and an MPI-disabled application (i.e. comparing two jobs running in two different conditions and with different execution paths). I would suggest to first verify and exclude that the problem is in the specific application code: if you still consider the problem to be on our side you should provide a receipt that shows a G4 stand-alone example with breaking of either strong or weak reproducibility.