TOPAS-nBio-Reg: a regression testing system for track structure simulations in TOPAS-nBio

Abstract Objective:
To develop a regression testing system for TOPAS-nBio: a wrapper of Geant4-DNA, and the radiobiological extension of TOPAS − a Monte Carlo code for the simulation of radiation transport. This regression testing system will be made publicly available on the TOPAS-nBio GitHub page.
Approach:
A set of seven regression tests were chosen to evaluate the suite of capabilities of TOPAS-nBio from both a physical and chemical point of view. Three different versions of the code were compared: TOPAS-nBio-v2.0 (the previous version), TOPAS-nBio-v3.0 (the current public release), and TOPAS-nBio-v4.0 (the current developer version, planned for future release). The main aspects compared for each test were the differences in execution times, variations from other versions of TOPAS-nBio, and agreement with measurements/in silico data.
Main results:
Execution times of nBio-v3.0 for all physics tests were faster than those of nBio-v2.0 due to the use of a new Geant4 version. Mean point-to-point differences between TOPAS-nBio versions across all tests fell largely within 5 %. The exceptions were the radiolytic yields (G values) of H 2 and H 2 O 2 , which differed moderately (16 % and 10 % respectively) when going from nBio-v3.0 to nBio-v4.0. In all cases a good agreement with other experimental/simulated data was obtained.
Significance:
From a developer point of view, this regression testing system is essential as it allows a more rigorous reporting of the consequences of new version releases on quantities such as the LET or G values of chemical species. Furthermore, it enables us to test “pushes” made to the codebase by collaborators and contributors. From an end-user point of view, users of the software are now able to easily evaluate how changes in the source code, made for their specific application, would affect the results of known quantities.