EM Physics: results, publications, validations

 

Validation and verification

Electromagnetic processes can be described theoretically, especially at high energies. In that case precision of simulation depends on a choice of implementation method and an important part of validation is the direct comparison between simulation results and theory predictions.

At lower energies (below 1 MeV) general theory fail, because it is necessary to describe wave function of atomic electrons in media. So, different phenomenological approaches are applied. Cross sections, stopping powers and other physical data are provided in the evaluated databases. To be included inside simulation software a parameterization of data is required. In that case verification means a comparison between simulation results and specific or evaluated data.

Simulation of electromagnetic processes depends on tracking method via cuts and step limitation. Control on results as a function of these parameters is also part of the validation process.

Thus, simulation precision depends on chosen theoretical models, parameterization methods, tracking parameters and software implementation. All these aspects of the Standard EM packages are described in details in the Geant4 Physics Reference Manual.

Regular regression tests for each new Geant4 release are performed. For that a special software working in automatic mode have been developed. Comparisons also performed between different Geant4 models if they have common applicability energy range.

Validation strategy

EM physics groups use following strategy of verification and validation:

  • unit tests by developers;
  • small statistic tests using EM testing suite;
  • large statistic tests using EM testing suite for important modifications;
  • nightly tests by Geant4 system testing team;
  • SQA and benchmarking by Geant4 system testing team;
  • monthly reference versions are validated using EM testing suite and Geant-val;
  • specialized verification and validation focused on publication;
  • user verification and validation;