seemps.analysis.comptree.mps_chain_tree

seemps.analysis.comptree.mps_chain_tree(chain_tree: ChainTree, allowed_support: TypeAliasForwardRef('~seemps.typing.Vector') | None = None) → SparseMPS

Compute the MPS representation of a multivariate function encoded as a ChainTree.

Returns a SparseMPS, whose cores are highly sparse and represented as collections of CSR matrices. This construction is particularly efficient for functions with a reduced support, which are recompressed with a right-to-left binning sweep. Source: https://arxiv.org/abs/2206.03832

Parameters:

chain_treeChainTree

Chain-like computation-tree representation of the target function.

allowed_supportVector, optional

Discrete set of admissible function outputs (e.g., [0, 1] for a Heaviside function).

Returns:

SparseMPS

Sparse MPS approximation of the target multivariate function.