GroupedSite

Inheritance Diagram

Inheritance diagram of tenpy.networks.site.GroupedSite

Methods

GroupedSite.__init__(sites[, labels, charges])

GroupedSite.add_op(name, op[, need_JW, hc, ...])

Add one on-site operators.

GroupedSite.change_charge([new_leg_charge, ...])

Change the charges of the site (in place).

GroupedSite.from_hdf5(hdf5_loader, h5gr, subpath)

Load instance from a HDF5 file.

GroupedSite.get_hc_op_name(name)

Return the hermitian conjugate of a given operator.

GroupedSite.get_op(name)

Return operator of given name.

GroupedSite.kroneckerproduct(ops)

Return the Kronecker product \(op0 \otimes op1\) of local operators.

GroupedSite.multiply_op_names(names)

Multiply operator names together.

GroupedSite.multiply_operators(operators)

Multiply local operators (possibly given by their names) together.

GroupedSite.op_needs_JW(name)

Whether an (composite) onsite operator is fermionic and needs a Jordan-Wigner string.

GroupedSite.remove_op(name)

Remove an added operator.

GroupedSite.rename_op(old_name, new_name)

Rename an added operator.

GroupedSite.save_hdf5(hdf5_saver, h5gr, subpath)

Export self into a HDF5 file.

GroupedSite.sort_charge([bunch])

Sort the leg charges (in place).

GroupedSite.state_index(label)

Return index of a basis state from its label.

GroupedSite.state_indices(labels)

Same as state_index(), but for multiple labels.

GroupedSite.test_sanity()

Sanity check, raises ValueErrors, if something is wrong.

GroupedSite.valid_opname(name)

Check whether 'name' labels a valid onsite-operator.

Class Attributes and Properties

GroupedSite.dim

Dimension of the local Hilbert space.

GroupedSite.onsite_ops

Dictionary of on-site operators for iteration.

class tenpy.networks.site.GroupedSite(sites, labels=None, charges='same')[source]

Bases: Site

Group two or more Site into a larger one.

A typical use-case is that you want a NearestNeighborModel for TEBD although you have next-nearest neighbor interactions: you just double your local Hilbertspace to consist of two original sites. Note that this is a ‘hack’ at the cost of other things (e.g., measurements of ‘local’ operators) getting more complicated/computationally expensive.

If the individual sites indicate fermionic operators (with entries in need_JW_string), we construct the new on-site operators of site1 to include the JW string of site0, i.e., we use the Kronecker product of [JW, op] instead of [Id, op] if necessary (but always [op, Id]). In that way the onsite operators of this DoubleSite automatically fulfill the expected commutation relations. See also Fermions and the Jordan-Wigner transformation.

Parameters:
  • sites (list of Site) – The individual sites being grouped together. Copied before use if charges!='same'.

  • labels – Include the Kronecker product of each onsite operator op on sites[i] and identities on other sites with the name opname+labels[i]. Similarly, set state labels for ' '.join(state[i]+'_'+labels[i]). Defaults to [str(i) for i in range(n_sites)], which for example grouping two SpinSites gives operators name like "Sz0" and state labels like 'up_0 down_1'.

  • charges ('same' | 'drop' | 'independent') – How to handle charges, defaults to ‘same’. 'same' means that all sites have the same ChargeInfo, and the total charge is the sum of the charges on the individual sites. 'independent' means that the sites have possibly different ChargeInfo, and the charges are conserved separately, i.e., we have n_sites conserved charges. For 'drop', we drop any charges, such that the remaining legcharges are trivial. For more complex situations, you can call set_common_charges() beforehand.

n_sites

The number of sites grouped together, i.e. len(sites).

Type:

int

sites

The sites grouped together into self.

Type:

list of Site

labels

The labels using which the single-site operators are added during construction.

Type:

list of str

kroneckerproduct(ops)[source]

Return the Kronecker product \(op0 \otimes op1\) of local operators.

Parameters:

ops (list of Array) – One operator (or operator name) on each of the ungrouped sites. Each operator should have labels ['p', 'p*'].

Returns:

prod – Kronecker product \(ops[0] \otimes ops[1] \otimes \cdots\), with labels ['p', 'p*'].

Return type:

Array

add_op(name, op, need_JW=False, hc=None, permute_dense=None)[source]

Add one on-site operators.

Parameters:
  • name (str) – A valid python variable name, used to label the operator. The name under which op is added as attribute to self.

  • op (np.ndarray | Array) – A matrix acting on the local hilbert space representing the local operator. Dense numpy arrays are automatically converted to Array. LegCharges have to be [leg, leg.conj()]. We set labels 'p', 'p*'.

  • need_JW (bool) – Whether the operator needs a Jordan-Wigner string. If True, add name to need_JW_string.

  • hc (None | False | str) – The name for the hermitian conjugate operator, to be used for hc_ops. By default (None), try to auto-determine it. If False, disable adding entries to hc_ops.

  • permute_dense (bool | None) – Flag to enable/disable permutations when converting op from numpy to np_conserved arrays. If True, the operator is permuted with perm to account for permutations induced by sorting charges; False disables the permutations. By default (None), the value of used_sort_charge is used.

change_charge(new_leg_charge=None, permute=None)[source]

Change the charges of the site (in place).

Parameters:
  • new_leg_charge (LegCharge | None) – The new charges to be used. If None, use trivial charges.

  • permute (ndarray | None) – The permutation applied to the physical leg, which also gets used to adjust state_labels and perm. If you sorted the previous leg with perm_qind, new_leg_charge = leg.sort(), use old_leg.perm_flat_from_perm_qind(perm_qind). Ignored if None.

property dim

Dimension of the local Hilbert space.

classmethod from_hdf5(hdf5_loader, h5gr, subpath)[source]

Load instance from a HDF5 file.

This method reconstructs a class instance from the data saved with save_hdf5().

Parameters:
  • hdf5_loader (Hdf5Loader) – Instance of the loading engine.

  • h5gr (Group) – HDF5 group which is represent the object to be constructed.

  • subpath (str) – The name of h5gr with a '/' in the end.

Returns:

obj – Newly generated class instance containing the required data.

Return type:

cls

get_hc_op_name(name)[source]

Return the hermitian conjugate of a given operator.

Parameters:

name (str) – The name of the operator to be conjugated. Multiple operators separated by whitespace are interpreted as an operator product, exactly as get_op() does.

Returns:

hc_op_name – Operator name for the hermitian conjugate operator.

Return type:

str

get_op(name)[source]

Return operator of given name.

Parameters:

name (str) – The name of the operator to be returned. In case of multiple operator names separated by whitespace, we multiply them together to a single on-site operator (with the one on the right acting first).

Returns:

op – The operator given by name, with labels 'p', 'p*'. If name already was an npc Array, it’s directly returned.

Return type:

np_conserved

multiply_op_names(names)[source]

Multiply operator names together.

Join the operator names in names such that get_op returns the product of the corresponding operators.

Parameters:

names (list of str) – List of valid operator labels.

Returns:

combined_opname – A valid operator name Operator name representing the product of operators in names.

Return type:

str

multiply_operators(operators)[source]

Multiply local operators (possibly given by their names) together.

Parameters:

operators (list of {str | Array}) – List of valid operator names (to be translated with get_op()) or directly on-site operators in the form of npc arrays with 'p', 'p*' label. The operators are multiplied left-to-right.

Returns:

combined_operator – The product of the given operators in a left-to-right multiplication following the usual mathematical convention. For example, if operators=['Sz', 'Sp', 'Sx'], the final operator is equivalent to site.get_op('Sz Sp Sx'), with the 'Sx' operator acting first on any physical state.

Return type:

Array

property onsite_ops

Dictionary of on-site operators for iteration.

Single operators are accessible as attributes.

op_needs_JW(name)[source]

Whether an (composite) onsite operator is fermionic and needs a Jordan-Wigner string.

Parameters:

name (str) – The name of the operator, as in get_op().

Returns:

needs_JW – Whether the operator needs a Jordan-Wigner string, judging from need_JW_string.

Return type:

bool

remove_op(name)[source]

Remove an added operator.

Parameters:

name (str) – The name of the operator to be removed.

rename_op(old_name, new_name)[source]

Rename an added operator.

Parameters:
  • old_name (str) – The old name of the operator.

  • new_name (str) – The new name of the operator.

save_hdf5(hdf5_saver, h5gr, subpath)[source]

Export self into a HDF5 file.

This method saves all the data it needs to reconstruct self with from_hdf5().

This implementation saves the content of __dict__ with save_dict_content(), storing the format under the attribute 'format'.

Parameters:
  • hdf5_saver (Hdf5Saver) – Instance of the saving engine.

  • h5gr (:class`Group`) – HDF5 group which is supposed to represent self.

  • subpath (str) – The name of h5gr with a '/' in the end.

sort_charge(bunch=True)[source]

Sort the leg charges (in place).

Parameters:

bunch (bool) – Whether to also group equal charges into larger blocks (usually a good idea).

Returns:

perm – The permutation

Return type:

1D ndarray

state_index(label)[source]

Return index of a basis state from its label.

Parameters:

label (int | string) – either the index directly or a label (string) set before.

Returns:

state_index – the index of the basis state associated with the label.

Return type:

int

state_indices(labels)[source]

Same as state_index(), but for multiple labels.

test_sanity()[source]

Sanity check, raises ValueErrors, if something is wrong.

valid_opname(name)[source]

Check whether ‘name’ labels a valid onsite-operator.

Parameters:

name (str) – Label for the operator. Can be multiple operator(labels) separated by whitespace, indicating that they should be multiplied together.

Returns:

validTrue if name is a valid argument to get_op().

Return type:

bool