Faster LazyTensor mul!() for dense suboperators (#39)

Dramatically speed up operations involving LazyTensors consisting of dense factors.
* Use transpose-GEMM-transpose approach for LazyTensor operations so that BLAS can be used.
* Add limited support for summing LazyTensor operators.
* Keep old code as special case for pure sparse LazyTensors (it is faster for this case)

Co-authored-by: Ashley Milsted <[email protected]>

Author: amilsted

Project.toml

@@ -1,13 +1,15 @@
 name = "QuantumOpticsBase"
 uuid = "4f57444f-1401-5e15-980d-4471b28d5678"
-version = "v0.3.1"
+version = "0.3.1"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
 FFTW = "7a1cc6ca-52ef-59f5-83cd-3a7055c09341"
+LRUCache = "8ac3fa9e-de4c-5943-b1dc-09c6b5f20637"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
+Strided = "5e0ebb24-38b0-5f93-81fe-25c709ecae67"
 
 [compat]
 Adapt = "1, 2, 3.3"

src/QuantumOpticsBase.jl

@@ -1,6 +1,6 @@
 module QuantumOpticsBase
 
-using SparseArrays, LinearAlgebra
+using SparseArrays, LinearAlgebra, LRUCache, Strided
 import LinearAlgebra: mul!, rmul!
 
 export bases, Basis, GenericBasis, CompositeBasis, basis,
@@ -13,7 +13,8 @@ export bases, Basis, GenericBasis, CompositeBasis, basis,
         operators_sparse, SparseOperator, diagonaloperator, SparseOpType,
         operators_lazysum, LazySum,
         operators_lazyproduct, LazyProduct,
-        operators_lazytensor, LazyTensor,
+        operators_lazytensor, LazyTensor, lazytensor_use_cache, lazytensor_clear_cache,
+        lazytensor_cachesize, lazytensor_disable_cache, lazytensor_enable_cache,
         superoperators, SuperOperator, DenseSuperOperator, DenseSuperOpType,
                 SparseSuperOperator, SparseSuperOpType, spre, spost, liouvillian,
         fock, FockBasis, number, destroy, create,