4.2. tnpy.tsdrg.TensorTree#

class tnpy.tsdrg.TensorTree(mpo)[source]#

Bases: object

The bottom-up binary tree where each node contains a tensor.

Parameters

mpo (MatrixProductOperator) – The matrix product operator.

__init__(mpo)[source]#

The bottom-up binary tree where each node contains a tensor.

Parameters

mpo (tnpy.operators.MatrixProductOperator) – The matrix product operator.

Methods

__init__(mpo)

The bottom-up binary tree where each node contains a tensor.

check_root()

Accessory decorator for checking the existence of root in TensorTree.

common_ancestor(node_id1, node_id2[, lowest])

Find all common ancestor of two given nodes in the order starting from the root.

find_path(node_id[, return_itself])

Find the path from the root to the targeted node node_id.

fuse(left_id, right_id, new_id, data)

Fuse two nodes into one and assign the given data one it.

plot([view])

Plot the tree with graphviz.Digraph.

tensor_network([node_ids, conj, ...])

Construct the qtn.TensorNetwork object from the given list of node_ids.

Attributes

horizon

The horizon is a moving line to track the node ids in the current renormalized system.

leaves

A dictionary to leaves, i.e. matrix product operator.

n_layers

Compute the number of layers in this tree, leaves included.

n_leaves

The number of leaves, namely the system size.

n_nodes

Total number of nodes in tree, including leaves.

root

The root of tree.

root_id

The ID of root.
class Syntax(node='Node', conj_node='ConjNode', level_idx='LevelIdx')[source]#

Bases: object

The name of indices follows a standardized syntax.

Examples

  • For open index: Node<node_id>LevelIdx<nth_idx>

  • For connected index: Node<node1_id>-Node<node2_id>

  • In the presence of conjugated node, replace Node with ConjNode

Parameters

  • node (str) –

  • conj_node (str) –

  • level_idx (str) –

Return type

None

node: str = 'Node'#
conj_node: str = 'ConjNode'#
level_idx: str = 'LevelIdx'#
__init__(node='Node', conj_node='ConjNode', level_idx='LevelIdx')#
Parameters

  • node (str) –

  • conj_node (str) –

  • level_idx (str) –

Return type

None

__init__(mpo)[source]#

The bottom-up binary tree where each node contains a tensor.

Parameters

mpo (tnpy.operators.MatrixProductOperator) – The matrix product operator.

property root_id: int | None#

The ID of root.

Returns

None if the tree is not constructed yet.

property root: tnpy.tsdrg.Node#

The root of tree.

Raises

KeyError – If the tree is not constructed yet.

property leaves: Dict[int, tnpy.tsdrg.Node]#

A dictionary to leaves, i.e. matrix product operator.

property horizon: List[int]#

The horizon is a moving line to track the node ids in the current renormalized system. This can be useful during the bottom-up construction of the tree, until there is only one node on the horizon, that is the root.

Returns

The horizon as a list of node ids.

property n_nodes: int#

Total number of nodes in tree, including leaves.

property n_leaves: int#

The number of leaves, namely the system size.

fuse(left_id, right_id, new_id, data)[source]#

Fuse two nodes into one and assign the given data one it. Meanwhile, the horizon will be updated accordingly. Note that the data should be provided from outer scope, and this method is only responsible to create new node in the tree.

Parameters

  • left_id (int) – The ID of left node to be fused.

  • right_id (int) – The ID of right node to be fused.

  • new_id (int) – The ID of new node.

  • data (numpy.ndarray) – Data that will be assigned to the new node.

check_root()[source]#

Accessory decorator for checking the existence of root in TensorTree.

Returns:

Parameters

func (Callable) –

Return type

Callable

property n_layers: int#

Compute the number of layers in this tree, leaves included.

tensor_network(node_ids=None, conj=False, mangle_outer=True, with_leaves=False)[source]#

Construct the qtn.TensorNetwork object from the given list of node_ids. Physical quantities often don’t require all nodes, and one could cooperate this method with find_path() or common_ancestor().

Parameters

  • node_ids (Sequence[int]) – The id of nodes that will be selected from the tree.

  • conj (bool) – Take a conjugate on this network. This will change the indices on physical bonds, as well as the internal indices within the network. This will be an inplace operation.

  • mangle_outer (bool) – For conjugate network only, namely when conj is True. If this parameter is set to False, the indices on physical bonds will remain untouched. This can be useful for taking inner product on the state, i.e. \(\langle\psi|\psi\rangle\).

  • with_leaves (bool) – Whether the leaves, that is, the MPOs, should be taken into account in this network.

Return type

quimb.tensor.tensor_core.TensorNetwork

Returns:

find_path(node_id, return_itself=False)[source]#

Find the path from the root to the targeted node node_id.

Parameters

  • node_id (int) – The id of node we are looking for.

  • return_itself (bool) – Should the target node be included in returning path.

Returns

ID of nodes on the path from the root to the targeted node.

Raises

KeyError – If no path can be found to the given node_id.

Return type

List[int]

common_ancestor(node_id1, node_id2, lowest=False)[source]#

Find all common ancestor of two given nodes in the order starting from the root.

Parameters

  • node_id1 (int) –

  • node_id2 (int) –

  • lowest (bool) –

Return type

Union[int, List[int]]

Returns:

plot(view=False)[source]#

Plot the tree with graphviz.Digraph.

Parameters

view (bool) – View the plot in an interactive window.

Returns

The Digraph object.

Return type

graphviz.graphs.Digraph