Uses spline order in the energy loss calculation #1496
Conversation
…ysicsParamsOptions
…ut in the simple-driver.py
Test summary 3 844 files 5 953 suites 4m 8s ⏱️ Results for commit 9532d2f. ♻️ This comment has been updated with latest results. |
sethrj
added
enhancement
… grid created being too small
…pline template. Failing due to energy grid of size 2 in test
…s in MockProcess. Also alter scattering mock process to have 3 xs points instead of 2
| builders[ValueGridType::energy_loss] | ||
| = std::make_unique<ValueGridLogBuilder>( | ||
| applic.lower.value(), | ||
| applic.upper.value(), | ||
| VecDbl{eloss_rate.value(), eloss_rate.value()}); | ||
| VecDbl{eloss_rate.value(), eloss_rate.value(), eloss_rate.value()}); |
There was a problem hiding this comment.
At this point it might be clearer to use the size/fill constructor
| VecDbl{eloss_rate.value(), eloss_rate.value(), eloss_rate.value()}); | |
| VecDbl(3, eloss_rate.value())); |
| @@ -88,6 +88,7 @@ void from_json(nlohmann::json const& j, RunnerInput& v) | |||
| LDIO_LOAD_OPTION(max_steps); | |||
| LDIO_LOAD_REQUIRED(initializer_capacity); | |||
| LDIO_LOAD_REQUIRED(secondary_stack_factor); | |||
| LDIO_LOAD_OPTION(spline_eloss_order); | |||
There was a problem hiding this comment.
If you scroll down in the file, you'll see there's also a "save" that you should add as well.
| @@ -265,7 +268,8 @@ struct PhysicsParamsScalars | |||
| == static_cast<bool>(fixed_step_action)) | |||
| && lambda_limit > 0 && range_factor > 0 && range_factor < 1 | |||
| && safety_factor >= 0.1 | |||
| && step_limit_algorithm != MscStepLimitAlgorithm::size_; | |||
| && step_limit_algorithm != MscStepLimitAlgorithm::size_ | |||
| && spline_eloss_order > 0; | |||
There was a problem hiding this comment.
Move this check to right before lambda_limit > 0 so that the ordering is the same as the variables.
| * \c spline_eloss_order: the order of interpolation to be used for the | ||
| * spline interpolation. If it is 1, then the existing linear interpolation | ||
| * is used. If it is 2+, the spline interpolation is used for energy loss using | ||
| * the specified order. Default value is 1 |
There was a problem hiding this comment.
Reindent please (guessing this is VSCode's fault...)
| eloss = Energy{step * calc_eloss_rate(pre_step_energy)}; | ||
|
|
||
| size_type order = physics.scalars().spline_eloss_order; | ||
| CELER_ASSERT(order > 0); |
There was a problem hiding this comment.
Since we already asserted by construction that the physics is "true", let's skip this check.
| else if (order > 1) | ||
| { |
There was a problem hiding this comment.
This should probably be:
| else if (order > 1) | |
| { | |
| else | |
| { |
unless you want to add an else { CELER_ASSERT_UNREACHABLE(); } after the else.
| // Construct a grid calculator from a physics table | ||
| template<class T> | ||
| inline CELER_FUNCTION T make_calculator(ValueGridId, | ||
| const size_type order) const; |
There was a problem hiding this comment.
Generally we don't const local function arguments, particularly because this can affect Windows ABI
| const size_type order) const; | |
| size_type order) const; |
| */ | ||
| template<class T> | ||
| CELER_FUNCTION T PhysicsTrackView::make_calculator(ValueGridId id, | ||
| const size_type order) const |
There was a problem hiding this comment.
| const size_type order) const | |
| size_type order) const |
|
Great work @lebuller ! There are just a few things to look at. |
Adds an input option in the runner,
spline_eloss_order, which is passed toPhysicsParams::Inputas an option. This option is used to define the order of interpolation to be preformed incalc_mean_energy_lossinPhysicsStepUtils.hh. If this order is 1 the existing linear cross-section calculator is used and if it is greater than 1 a spline calculator of the given order is used.