Problem 2307

Summary:	Exit direction of hadronic elastic collisions perpendicular to Z-axis
Product:	Geant4	Reporter:	Sergio Losilla <sergio.losilla>
Component:	processes/hadronic/processes	Assignee:	dennis.herbert.wright
Status:	RESOLVED FIXED
Severity:	major	CC:	liulixing, sergio.losilla
Priority:	P4
Version:	10.6
Hardware:	All
OS:	All
Attachments:	Histogram containing the exit angle of neutrons in a water phantom for Patch adding the removed randomization around the Z-axis

Description Sergio Losilla 2021-01-10 15:01:27 CET

Created attachment 656 [details]
Histogram containing the exit angle of neutrons in a water phantom for

Geant4 10.6 (and later) yields wrong neutron fluences, as compared to Geant4 10.5 and earlier.

The reason is a regression: the momentum change should be randomly rotated around the Z-axis, but this was removed.

The issue had been reported here:

https://geant4-forum.web.cern.ch/t/thermal-neutron-physics-drastic-change-from-10-5-to-10-6/4011

To further test the issue, I created a histogram with the output angle of neutrons with respect to the Z-axis in elastic hadronic processes (divided by sin(theta)). In 10.5 and the patched 10.6 runs, the curves are almost flat (except for the expected bias towards theta=0 since the neutrons are shot in the +Z direction). In 10.6, almost all neutrons exit in the plane perpendicular to the Z-direction. This figure is attached.

The system is a water phantom, the source consists of 10 keV neutrons in the +Z direction. Please ask if further details are needed.

Comment 1 Sergio Losilla 2021-01-10 15:02:35 CET

Created attachment 657 [details]
Patch adding the removed randomization around the Z-axis

The patch consists of selected lines from Geant4 10.5.1.

Comment 2 dennis.herbert.wright 2021-01-11 22:31:17 CET

This has already been fixed according to bug report 2290.  It will be included in the first patch to Geant4 10.7

Comment 3 Sergio Losilla 2021-01-12 00:12:27 CET

Good to know. I will add a comment in the discussion.

Comment 4 Liu Lixing 2021-07-30 06:27:00 CEST

I have updated my geant4 to geant4.10.7.2. But I haven't seen the bug fixed! Is there something wrong? Here is the part of the code in greant4.10.7.2:
  
// Check the result for catastrophic energy non-conservation
  // cannot be applied because is not guranteed that recoil 
  // nucleus is created
  // result = CheckResult(theProj, targNucleus, result);

  // directions
  G4ThreeVector indir = track.GetMomentumDirection();
  G4ThreeVector outdir = result->GetMomentumChange();

  if(verboseLevel>1) {
    G4cout << "Efin= " << result->GetEnergyChange()
	   << " de= " << result->GetLocalEnergyDeposit()
	   << " nsec= " << result->GetNumberOfSecondaries()
	   << " dir= " << outdir
	   << G4endl;
  }

  // energies  
  G4double edep = std::max(result->GetLocalEnergyDeposit(), 0.0);
  G4double efinal = std::max(result->GetEnergyChange(), 0.0);

  // primary change
  theTotalResult->ProposeEnergy(efinal);

  if(efinal > 0.0) {
    outdir.rotateUz(indir);
    theTotalResult->ProposeMomentumDirection(outdir);
  } else {
    if(part->GetProcessManager()->GetAtRestProcessVector()->size() > 0)
         { status = fStopButAlive; }
    else { status = fStopAndKill; }
    theTotalResult->ProposeTrackStatus(status);
  }