Make use of libcommute and update C++ API accordingly #44
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This header features a new namespace, Pomerol::Operators imports a few libcommute classes and functions.
…ion` Drop the bit field based type FockState in favor of unsigned integers. All needed bit fiddling will be done by libcommute.
`HamiltonianPart` is now templated on the matrix storage type (real or complex elements), while `Hamiltonian` employs type erasure and stores parts as a vector of `std::shared_ptr<void>`. The Hamiltonian expression is not stored in `Hamiltonian`. Instead, it is only passed to `prepare()` to construct a matrix representation. Since the matrix representation is now generated by means of libcommute, a new header `LibcommuteEigen.h` has been added to teach libcommute how to use Eigen vectors and blocks as StateVector's.
Now type type erasure is used only to store the actual Eigen matrix and a reference to the `libcommute::loperator` object.
krivenko
changed the title
…st_libcommute `GF4siteTest_libcommute` fully compiles and gives correct results now.
This is the first version of Clang to offer a reasonable resolution of the C++ Core Language Defect Report 253. Earlier versions of Clang complain about the 'hc' object from libcommute. http://open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#253
krivenko
force-pushed
the
libcommute
branch
4 times, most recently
from
94096b1
to
83b3efe
Compare
Its functionality is now part of libcommute.
`LatticePresets` is a namespace with a bunch of free functions now.
… revealed bugs Also remove outdated tests `NOperatorTest` and `SzOperatorTest`.
Some of the unit tests have been removed as they would not make sense anymore.
|
Contents of the |
aeantipov
approved these changes
May 20, 2021
| @@ -50,7 +51,7 @@ class GreensFunction : public Thermal, public ComputableObject { | |||
| /** A list of pointers to parts (every part corresponds to a part of the annihilation operator | |||
| * and a part of the creation operator). | |||
| */ | |||
| std::list<GreensFunctionPart*> parts; | |||
| std::vector<GreensFunctionPart> parts; | |||
There was a problem hiding this comment.
I noticed sometimes we use a shared pointer and some time instances. What's the rationale?
There was a problem hiding this comment.
A couple of classes were storing parts in vectors of smart pointers because I forgot to change them to plain instances.
Those are updated in the latest commit.
krivenko
changed the title
This was referenced Nov 29, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There is still a lot of stuff to be done before this PR can be merged. I decided to open and update it incrementally so that we can discuss the development progress step by step.
Most of the library has already been ported, except for the classes related to the 2-particle GF and susceptibility. I have temporarily commented out compilation of all unit tests and added a new one. It is an adapted copy of
GF4siteTest, and I suggest we use it to discuss the updated API.Some highlights of the proposed changes
libcommute >= 0.6.2as an external dependency.Operatorclass and introduce a namespacePomerol::Operators, which simply imports some types and functions from libcommute.IndexClassificationis templated on the types of the indices now. This change lifts the restriction of fermionic degrees of freedom being always addressed by the(site,orbital,spin)triplets.LatticeandIndexHamiltonianhave no real use now and should go.LatticePresetsshould become a namespace with a bunch of free functions returninglibcommute::expressionobjects.Symmetrizerhas been replaced withHilbertSpace, which is responsible for automatic partitioning of the Hilbert space into sectors. No explicit integrals of motion are required by the new algorithm.StatesClassificationas they are not needed anymore (see the previous point). We may decide to put it back later on and just for informational purposes (better screen output).FieldOperator[Part]->MonomialOperator[Part]and allow for a wider class of operators: Any product ofcandc_dagwith a prefactor, i.e. a monomial as it still generates only one-to-one connections between sectors.HamiltonianPartandMonomialOperatorPartuse type erasure to switch between real and complex stored matrices at runtime.There is still a huge mess in the details of the reworked implementation and I shall start cleaning it out once we have agreed on the new API.