Wolfram Language Paclet Repository
Community-contributed installable additions to the Wolfram Language
Wolfram`TensorNetworks`
SparseTensorNetwork  |
| | |||||
▪
▪
The hyperedges and the output specification of the input network are preserved unchanged.
▪
The dual property form is , a predicate testing whether every tensor of is already a ; is always .
tn["SparseQ"]
tn
True
▪
All paclet operations (, path finders, and so on) work transparently on sparse-stored networks; choose storage for memory savings when most tensor entries are zero.
Examples
(7)
Basic Examples
(1)
Construct a tensor network with two dense tensors and convert it to a sparse-stored tensor network:
In[1]:=
{A,B}={RandomReal[{-1,1},{3,4}],RandomReal[{-1,1},{4,5}]};
In[2]:=
tn=
[{A,B},{{1,2},{2,3}}]
 | TensorNetwork |
Out[2]=
TensorNetwork
 |  |
|
The Sparse: Yes flag in the summary box indicates that every stored tensor is now a SparseArray:
In[3]:=
sparse=
[tn]
| SparseTensorNetwork |
Out[3]=
TensorNetwork
| |
|
Convert a randomly generated matrix product state to sparse storage:
In[4]:=
 | SparseTensorNetwork |
| RandomTensorNetwork |
Out[4]=
TensorNetwork
|  |
|
Confirm that every tensor is now stored as a SparseArray:
In[5]:=
sparse["SparseQ"]
Out[5]=
True
The heads of the stored tensors are SparseArray:
In[6]:=
Head/@sparse["Tensors"]
Out[6]=
{SparseArray,SparseArray}
Scope
(4)
Applications
(1)
Properties & Relations
(1)
 |
|
 |
|
""