- Table of contents
- Varying deltaInt parameter
Varying deltaInt parameter¶
This is still a work in progress. I'm attaching some preliminary plots only; documentation will be completed later.
Studying error and simulation time in terms of deltaInt
(without crossing planes)¶
- Without crossing planes, we expected beforehand no significative changes in the errors nor the simulation times.
- This plot validates this intuition.
- Note: as
G4_max_x_err
remains essentially constant, its curves are collapsed in a single one (which appears between 1 and 0.1 mm in the lefty
axis).
Studying error and simulation time in terms of deltaInt
(with crossing planes)¶
- This is a continuation of the previous experiment that instead uses a non-zero number of crossing planes.
- As expected, simulation times not only grow with the number of planes but also with lower
deltaInt
values (this was already discussed and is consistent with the plots of the next section). G4_max_x_err
does not change in any of these scenarios.
Studying simulation time in terms of deltaInt
(with crossing planes)¶
- This plot shows simulation times for six values of
deltaInt
(between 1e-10 and 1e-5) and an increasing number of planes. - It was expected to see differences in the curves, as
deltaInt
controls how accurate should be the detection of the intersection points. - In every case, the times grow linearly with the number of planes. The lower
deltaInt
, the higher the slope of the line, which is reasonable.
Studying error when detecting crossing planes¶
- This is related to this experiment.
- In this case, when
deltaInt
= 1e-7, the accuracy improves very subtly, but remains constant afterwards. That is, fordeltaInt
in [1e-10,...,1e-7], the intersection points found by Geant4 are exactly the same.