Protect against wrong primary particle mass

[atolosadelgado]

Hi,

ddsim allows to load primary particles from an external file, but it seems there is no sanity check for the mass of the particles.

Question: would be worth to check the mass of the primary particles before starting the simulation? This issue would affect the builtin particle gun, which allows to define a total energy of primary particle smaller than the mass (e.g., an alpha particle with 1 MeV).

Best,
Alvaro

[andresailer]

Doesn't this rather mean to check against insufficient total energy, and not check against the mass of the particle? For my education, what happens if one simulates the alpha with 1 MeV total energy?

I am also not sure if you refer to the particle.tbl file, which allows one to define additional particles, but it doesn't allow one to overwrite particles Geant4 already knows.

DD4hep/DDG4/plugins/Geant4ExtraParticles.cpp

Lines 110 to 112 in 4db360e

	// don't add if the particle already exists
	G4ParticleDefinition* p = theParticleTable->FindParticle(pdg);
	if ( !p ) {

There is no way for us to know what the correct mass of a particle is. If there are MC records used as input, which define include the mass of particles, we could check that that mass is consistent with the mass of the defined, but this doesn't sound like the issue you are raising here.

Cheers,
Andre

[atolosadelgado]

Hi,

For my education, what happens if one simulates the alpha with 1 MeV total energy?

ddsim still assigns a positive kinetic energy to the alpha, but well below 1 MeV. Simulation runs but the output is wrong. It can be very misleading.

we could check that that mass is consistent with the mass of the defined

Yes, that would be my point. Sorry I was not clear enough.

Best,
Alvaro

[andresailer]

Yes, that would be my point. Sorry I was not clear enough.

But my proposal would do nothing for the 1 MeV Alpha case? The mass of the Alpha isn't changed, is it?

[atolosadelgado]

The alpha at 1 MeV should still trigger an exception because the tabulated mass is bigger than the input value of E_kin+mass*c**2.

[andresailer]

What is the tabulated mass? What is E_kin and mass in the 1 MeV alpha case?
(I need to understand, or I can't even begin to implement anything to check this)

[atolosadelgado]

Geant4 maintains a table with all possible particles and their properties. Accessing the static properties of particles can be done as follows:

// Geant4 example: extended/electromagnetic/TestEm18/src/PrimaryGeneratorAction.cc
#include "G4ParticleTable.hh"
#include "G4ParticleDefinition.hh"
...
G4ParticleDefinition* particle = G4ParticleTable::GetParticleTable()->FindParticle("alpha");
G4double particle_mass = particle->GetPDGMass();
G4bool particle_stable   = particle->GetPDGStable();

The mass of the alpha particle is retrieved using the above snippet.

The DD4hep native particle gun requires the energy parameter to include the mass of the particle, so specifying 1 MeV for an alpha particle is incorrect. If the mass is not accounted for, it could lead to erroneous simulations.

Similar issues may arise when setting up the primaries from a hepmc3 file.

In my opinion, it would be safer to throw an exception when encountering such inconsistencies before starting the actual simulation.

There is a dedicated way to declare new particles for Geant4; it is not possible to run a simulation with an incorrect mass for a known particle.

[MarkusFrankATcernch]

Hi Alvaro,

I verified again in the code:

• The Geant4ParticleGenerator with the property "Energy" / "energy" implicitly uses energy as kinetic energy.
• When the G4 primary is created it has the proper energy (Geant4InputHandling.cpp:351):
  sqrt(momentum2 + mass2)

I cannot see anything wrong with that - except that the property name is perhaps not entirely detailed.
What do I miss?

It is true, if an input file contains 3-momentum and energy and at the same time the particle mass
loaded from external file is largely bigger than what was used to generate the input file, this might end up with
a computed "negative" mass or momentum.
Is this what you meant?

[atolosadelgado]

It is true, if an input file contains 3-momentum and energy and at the same time the particle mass
loaded from external file is largely bigger than what was used to generate the input file, this might end up with
a computed "negative" mass or momentum.
Is this what you meant?

Yes, that it is. Additionally, even with positive mass, there might be a mismatch between the derived mass from input 3momentum+energy and the mass value used by Geant4. For example, if the input is E=5, pc=4, the derived mass would be mc2 = 3, but the internal Geant4 value for that particle mass may be 2.5.

In case the derived mass value is smaller than 10 eV, the dynamical mass is set to zero. In case is larger by more than 10 eV than the PDG value, the input mass is taken. This behavior is implemented here.

Thank you for taking a look to this :)

[atolosadelgado]

I had a quick chat with some Geant4 people, and I will open a review of the G4PrimaryParticle class to add a sanity check there.

A piece of code may be added in the hepMC3 reader in order to check the mass consistency of the derived value from the input and the value from Geant4/PDG. What do you think?

[MarkusFrankATcernch]

Ideally the check is done in a central piece of code where the data of all readers, generators and guns are processed.
I suspect this is somewhere in Geant4InputHandling.cpp.

[andresailer]

Maybe here

DD4hep/DDG4/src/Geant4InputHandling.cpp

Line 346 in 58fa8c0

static G4PrimaryParticle* createG4Primary(const Geant4ParticleHandle p) {

[MarkusFrankATcernch]

@andresailer This is a very good place. all is present and also the particle definitions are queried from Geant4.
I will put the fix there.

[atolosadelgado]

after reading a bit more Geant4 code, I noticed that G4PrimaryParticle actually implements a check for negative masses here.

I have asked some colleagues from the FCC physics performance studies group, if implementing a check for the input particle mass would be an obstacle for their work or not (just to be sure).

[MarkusFrankATcernch]

Please see MR #1259