templatizing SPOSet
#5306
templatizing SPOSet
#5306
Conversation
|
Test this please |
| @@ -576,6 +589,7 @@ class SPOSet : public QMCTraits | |||
| IndexType OrbitalSetSize; | |||
| }; | |||
|
|
|||
| using SPOSet = SPOSetT<QMCTraits::QTBase::ValueType>; | |||
| using SPOSetPtr = SPOSet*; | |||
There was a problem hiding this comment.
For some reason I have a reflexive dislike of the SPOSetT naming and this alias trick to reduce the code change ripple. But I am accepting it because I think the reduction of files to review and improved readability everywhere the underlying type of SPOSet doesn't matter makes up for it.
There was a problem hiding this comment.
That is temporary renaming to minimize code change for an easy review. Once we touch the consumers of SPOSet. SPOSetT will be renamed back to SPOSet.
| DIM = OHMMS_DIM, | ||
| DIM_VGL = OHMMS_DIM + 2 // Value(1) + Gradients(OHMMS_DIM) + Laplacian(1) | ||
| }; | ||
| using ValueType = T; |
There was a problem hiding this comment.
As always I am not a fan of the redundant xxxType naming of type aliases and types. I'm glad to see FullPrecRealType at least dropped this. But if the ripple isn't excessive can we drop the redundant Type suffix where ever possible. I have been trying to do it when I touch code where it is possible.
There was a problem hiding this comment.
ValueType cannot be changed due to the used in derived classes. That change requires a giant renaming PR which distract the intention of this PR to focus on adjust the base class only. When touching derived classes, Value can be introduced parallel to ValueType once we finished going through all derived classes, ValueType can be removed eventually.
|
Q. Why did only SplineR2R.cpp need to be touched at this stage, and not (say) C2C? |
It is an inconsistency between R2R and C2C implementation. Will PR to address that. Prefer changing the C2C side. |
Please review the developer documentation
on the wiki of this project that contains help and requirements.
Proposed changes
Templatizing
SPOSettoSPOSetT<T>.SPOSetis aliased toSPOSetT<QMCTraits::QTBase::ValueType>.Ye's comment:
This is the beginning of an effort to allow a single build of QMCPACK with real and complex and also mixed precision.
We start from the lower level components of TWF and gradually turn the whole TWF into template.
This PR turns SPOSet template and both real and complex specialization are unconditionally instantiated. We will work on SPOSet derived classes next.
What type(s) of changes does this code introduce?
Does this introduce a breaking change?
What systems has this change been tested on?
Checklist
Update the following with a yes where the items apply. If you're unsure about any of them, don't hesitate to ask. This is
simply a reminder of what we are going to look for before merging your code.