The geant4 library

This library is used to store and analyse the events generated in a Geant4 simulation, it defines and stores the particle generator and simulation conditions, such as the details of the physics list used during the Monte Carlo.

Please have a look to this library contribution guide before pushing changes to this repository.

Lindhard for contained, stopped recoils

TRestGeant4QuenchingProcess has an opt-in gas model that evaluates the existing Lindhard formula once at a recoil track's initial kinetic energy E0. The same factor q(E0, A, Z) multiplies every local deposit on that track. Subdividing the same deposits therefore preserves the total visible energy, apart from floating point rounding. The original lindhard model still evaluates the factor per hit using its hadronic target information; auto, Birks, and existing configurations keep their historical behavior.

<TRestGeant4QuenchingProcess>
    <parameter name="applyToHitEnergies" value="true"/>
    <volume name="gasVolume" model="lindhard-stopped-recoil"/>
</TRestGeant4QuenchingProcess>

Use this option on unmodified deposited hit energies, before converting them to electrons or applying any other ionization-yield correction. A and Z come from the recoil's ground-state nuclear PDG code in the simulation metadata; a proton uses A=Z=1. Hadronic target isotope fields are not used. Electromagnetic tracks, neutrons, and other non-nuclear particles are unchanged by this model. A neutron event can consequently contain quenched recoil tracks and unquenched electromagnetic secondary tracks. This is an approximation, not a measured yield curve for any particular gas mixture.

The selected recoil must have a saved Init state, remain in one selected physical volume, end with zero saved kinetic energy, and have no explicit children (neither saved secondary IDs nor tracks with that parent ID). Each saved kinetic energy loss must match the corresponding local deposit, and the sum must match E0. The numerical tolerance is 8 * epsilon(float) * E0, because REST stores deposited and kinetic energies as floats. Missing metadata, unknown or excited nuclear identity, escaping tracks, secondary-producing tracks, or energy that does not close are rejected with an exception identifying the event and track; there is no fallback to another recoil model.

Simulation metadata must have storeTracks=true, registerEmptyTracks=true, removeUnwantedTracks=false, no kill volumes, and storage probability one for all geometry volumes. Complete volume storage prevents an unrecorded exit and re-entry from looking contained. For example, retain the full geometry when configuring the simulation:

<parameter name="storeTracks" value="true"/>
<parameter name="registerEmptyTracks" value="true"/>
<detector activateAllVolumes="true">
    <removeUnwantedTracks enabled="false"/>
    <volume name="gasVolume" sensitive="true" chance="1"/>
</detector>

REST does not store separate NIEL values, terminal track status, or material phase in these histories. The checks establish eligibility from the recorded steps; the user selects the gas volumes. Filtered historical files may be unsupported. With applyToHitEnergies=false, only event energy accounting is converted; keep it true when the next process generates charge from the hit energies. An attempt to convert already quenched hit energies again fails the kinetic-loss closure check.

The synthetic ProcessEvent tests in test/src/Geant4QuenchingProcesses.cxx cover deposit subdivision, recoil identity, unchanged electromagnetic/neutron hits, event energy accounting, historical Lindhard, and the unsupported-history checks. They require no transport simulation or external ROOT data file.