[GeoMechanicsApplication] Extract computation of integration coefficients #12334
Conversation
avdg81
added
Refactor
applications/GeoMechanicsApplication/custom_elements/U_Pw_small_strain_element.cpp
Show resolved
Hide resolved
|
@markelov208 FYI, this is Annes PR for making the lists for the integration coefficients. In the end, lists were only added for the CalculateAll functions, so you could follow a similar way in the CalculateMassMatrix function (note that the signature of the CalculateIntegrationCoefficient has changed, so it's a good idea to bring your branch up to date with master when this has merged). |
Hi Gennady, good question. I have considered that, but then I realized that this is only an intermediate step. The aim is to have such a function available later, perhaps as part of an |
Hi Anne, thank you for the clarification. Can we put such a function into 'geometry.h'? The 'geometry' class has already integration points. I have personal interest in this function ;) to use it for CalculateMassMatrix; of course I can copy the statement. Kind regards, |
There was a problem hiding this comment.
The number of times you had to do the same action reveals how many copies exist within our code base. Your change looks to be consequently executed for all of them.
Remarks that I would like to make would without exception refer to the existing situation that you start improving here, so they would not be productive here.
Please go ahead and merge this, such that a next step can be taken.
| // Compute weighting coefficient for integration | ||
| Variables.IntegrationCoefficient = | ||
| this->CalculateIntegrationCoefficient(IntegrationPoints, GPoint, Variables.detJ); | ||
| Variables.IntegrationCoefficient = integration_coefficients[GPoint]; | ||
|
|
||
| Variables.IntegrationCoefficientInitialConfiguration = this->CalculateIntegrationCoefficient( | ||
| IntegrationPoints, GPoint, Variables.detJInitialConfiguration); |
There was a problem hiding this comment.
This piece of code seems to use the same function with the same input. Therefore I think that detJ == detJInitialConfiguration. In the new implementation storing the list of detJ for the integrationpoints would then avoid the second call to CalculateIntegrationCoefficient.
There was a problem hiding this comment.
I'm not sure. ElementVariables::detJ is taken from ElementVariables::detJContainer, which is computed by calling member InitializeElementVariables. ElementVariables::detJInitialConfiguration on the other hand is computed by calling member CalculateKinematics (which in turn calls member CalculateDerivativesOnInitialConfiguration). So they are probably two different things. However, it is unclear to me where ElementVariables::detJInitialConfiguration is being used and whether that is correct. To be investigated.
Calculation of the integration coefficient does not require the member function to be non-`const`. Also, these member functions were declared `virtual`, but never overridden.
To reduce the amount of duplicated code. As a result, the variables that store the result could be made `const`.
📝 Description
Extracted the computation of a vector of integration points from the integration point loops in (virtually) all implementations of
CalculateAll. Also modified a few function parameters types from reference to reference-to-const. In that way, it becomes clearer that the supplied object won't be modified by the callee.