`pygsti.circuits.circuit`

Defines the Circuit class

Module Contents

Classes

`Circuit`	A quantum circuit.
`CompressedCircuit`	A "compressed" Circuit that requires less disk space.
`SeparatePOVMCircuit`	Separately holds a POVM-less `Circuit` object, a POVM label, and a list of effect labels.

Functions

to_label(x)

Helper function for converting x to a single Label object

Attributes

msg

pygsti.circuits.circuit.msg

pygsti.circuits.circuit.to_label(x)

Helper function for converting x to a single Label object

Parameters

xvarious: An object to (attempt to) convert into a Label.

Returns

Label

class pygsti.circuits.circuit.Circuit(layer_labels=(), line_labels='auto', num_lines=None, editable=False, stringrep=None, name='', check=True, expand_subcircuits='default', occurrence=None, compilable_layer_indices=None)

Bases: object

A quantum circuit.

A Circuit represents a quantum circuit, consisting of state preparation, gates, and measurement operations. It is composed of some number of “lines”, typically one per qubit, and stores the operations on these lines as a sequence of Label objects, one per circuit layer, whose .sslbls members indicate which line(s) the label belongs on. When a circuit is created with ‘editable=True’, a rich set of operations may be used to construct the circuit in place, after which done_editing() should be called so that the Circuit can be properly hashed as needed.

Parameters

layer_labelsiterable of Labels or str

This argument provides a list of the layer labels specifying the state preparations, gates, and measurements for the circuit. This argument can also be a Circuit or a string, in which case it is parsed as a text-formatted circuit. Internally this will eventually be converted to a list of Label objects, one per layer, but it may be specified using anything that can be readily converted to a Label objects. For example, any of the following are allowed:

[‘Gx’,’Gx’] : X gate on each of 2 layers
[Label(‘Gx’),Label(‘Gx’)] : same as above
[(‘Gx’,0),(‘Gy’,0)] : X then Y on qubit 0 (2 layers)
[[(‘Gx’,0),(‘Gx’,1)],[(‘Gy’,0),(‘Gy’,1)]] : parallel X then Y on qubits 0 & 1

line_labelsiterable, optional

The (string valued) label for each circuit line. If ‘auto’, then line_labels is taken to be the list of all state-space labels present within layer_labels. If there are no such labels (e.g. if layer_labels contains just gate names like (‘Gx’,’Gy’)), then the special value ‘*’ is used as a single line label.

num_linesint, optional

Specify this instead of line_labels to set the latter to the integers between 0 and num_lines-1.

editablebool, optional

Whether the created Circuit is created in able to be modified. If True, then you should call done_editing() once the circuit is completely assembled, as this makes the circuit read-only and allows it to be hashed.

stringrepstring, optional

A string representation for the circuit. If None (the default), then this will be generated automatically when needed. One reason you’d want to specify this is if you know of a nice compact string representation that you’d rather use, e.g. “Gx^4” instead of the automatically generated “GxGxGxGx”. If you want to initialize a Circuit entirely from a string representation you can either specify the string in as layer_labels or set layer_labels to None and stringrep to any valid (one-line) circuit string.

namestr, optional

A name for this circuit (useful if/when used as a block within larger circuits).

checkbool, optional

Whether stringrep should be checked against layer_labels to ensure they are consistent, and whether the labels in layer_labels are a subset of line_labels. The only reason you’d want to set this to False is if you’re absolutely sure stringrep and line_labels are consistent and want to save computation time.

expand_subcircuitsbool or “default”

If “default”, then the value of Circuit.default_expand_subcircuits is used. If True, then any sub-circuits (e.g. anything exponentiated like “(GxGy)^4”) will be expanded when it is stored within the created Circuit. If False, then such sub-circuits will be left as-is. It’s typically more robust to expand sub-circuits as this facilitates comparison (e.g. so “GxGx” == “Gx^2”), but in cases when you have massive exponents (e.g. “Gx^8192”) it may improve performance to set expand_subcircuits=False.

occurrencehashable, optional

A value to set as the “occurrence id” for this circuit. This value doesn’t affect the circuit an any way except by affecting it’s hashing and equivalence testing. Circuits with different occurrence ids are not equivalent. Occurrence values effectively allow multiple copies of the same ciruit to be stored in a dictionary or DataSet.

compilable_layer_indicestuple, optional

The circuit-layer indices that may be internally altered (but retaining the same target operation) and/or combined with the following circuit layer by a hardware compiler.when executing this circuit. Layers that are not “compilable” are effectively followed by a barrier which prevents the hardward compiler from restructuring the circuit across the layer boundary.

Attributes

default_expand_subcircuitsbool: By default, expand sub-circuit labels.
line_labelstuple: The line labels (often qubit labels) of this circuit.
layertuptuple: This Circuit’s layers as a standard Python tuple of layer Labels.
tuptuple: This Circuit as a standard Python tuple of layer Labels and line labels.
strstr: The Python string representation of this Circuit.

Creates a new Circuit object, encapsulating a quantum circuit.

You only need to supply the first layer_labels argument, though usually (except for just 1 or 2 qubits) you’ll want to also supply line_labels or num_lines. If you’ll be adding to or altering the circuit before using it, you should set editable=True.

Parameters

layer_labelsiterable of Labels or str

This argument provides a list of the layer labels specifying the state preparations, gates, and measurements for the circuit. This argument can also be a Circuit or a string, in which case it is parsed as a text-formatted circuit. Internally this will eventually be converted to a list of Label objects, one per layer, but it may be specified using anything that can be readily converted to a Label objects. For example, any of the following are allowed:

[‘Gx’,’Gx’] : X gate on each of 2 layers
[Label(‘Gx’),Label(‘Gx’)] : same as above
[(‘Gx’,0),(‘Gy’,0)] : X then Y on qubit 0 (2 layers)
[[(‘Gx’,0),(‘Gx’,1)],[(‘Gy’,0),(‘Gy’,1)]] : parallel X then Y on qubits 0 & 1

line_labelsiterable, optional

The (string valued) label for each circuit line. If ‘auto’, then line_labels is taken to be the list of all state-space labels present within layer_labels. If there are no such labels (e.g. if layer_labels contains just gate names like (‘Gx’,’Gy’)), then the special value ‘*’ is used as a single line label.

num_linesint, optional

Specify this instead of line_labels to set the latter to the integers between 0 and num_lines-1.

editablebool, optional

Whether the created Circuit is created in able to be modified. If True, then you should call done_editing() once the circuit is completely assembled, as this makes the circuit read-only and allows it to be hashed.

stringrepstring, optional

A string representation for the circuit. If None (the default), then this will be generated automatically when needed. One reason you’d want to specify this is if you know of a nice compact string representation that you’d rather use, e.g. “Gx^4” instead of the automatically generated “GxGxGxGx”. If you want to initialize a Circuit entirely from a string representation you can either specify the string in as layer_labels or set layer_labels to None and stringrep to any valid (one-line) circuit string.

namestr, optional

A name for this circuit (useful if/when used as a block within larger circuits).

checkbool, optional

Whether stringrep should be checked against layer_labels to ensure they are consistent, and whether the labels in layer_labels are a subset of line_labels. The only reason you’d want to set this to False is if you’re absolutely sure stringrep and line_labels are consistent and want to save computation time.

expand_subcircuitsbool or “default”

If “default”, then the value of Circuit.default_expand_subcircuits is used. If True, then any sub-circuits (e.g. anything exponentiated like “(GxGy)^4”) will be expanded when it is stored within the created Circuit. If False, then such sub-circuits will be left as-is. It’s typically more robust to expand sub-circuits as this facilitates comparison (e.g. so “GxGx” == “Gx^2”), but in cases when you have massive exponents (e.g. “Gx^8192”) it may improve performance to set expand_subcircuits=False.

occurrencehashable, optional

A value to set as the “occurrence id” for this circuit. This value doesn’t affect the circuit an any way except by affecting it’s hashing and equivalence testing. Circuits with different occurrence ids are not equivalent. Occurrence values effectively allow multiple copies of the same ciruit to be stored in a dictionary or DataSet.

compilable_layer_indicestuple, optional

The circuit-layer indices that may be internally altered (but retaining the same target operation) and/or combined with the following circuit layer by a hardware compiler.when executing this circuit. Layers that are not “compilable” are effectively followed by a barrier which prevents the hardward compiler from restructuring the circuit across the layer boundary.

property line_labels: The line labels (often qubit labels) of this circuit.

property name

The name of this circuit.

Note: the name is not a part of the hashed value. The name is used to name the CircuitLabel returned from to_label().

property occurrence: The occurrence id of this circuit.

property layertup: This Circuit’s layers as a standard Python tuple of layer Labels.

Returns

tuple

property tup: This Circuit as a standard Python tuple of layer Labels and line labels.

Returns

tuple

property compilable_layer_indices: Tuple of the layer indices corresponding to “compilable” layers.

property compilable_by_layer: Boolean array indicating whether each layer is “compilable” or not.

property str: The Python string representation of this Circuit.

Returns

str

property layerstr: Just the string representation of the circuit layers (no ‘@<line_labels>’ suffix)

property linesstr: Just the string representation of the circuit’s line labels (the ‘@<line_labels>’ suffix)

property num_lines: The number of lines in this circuit.

Returns

int

property num_layers

The number of circuit layers.

In simple circuits, this is the same as the depth (given by depth()). For circuits containing sub-circuit blocks, this gives the number of top-level layers in this circuit.

Returns

int

property depth

The circuit depth.

This is the number of layers in simple circuits. For circuits containing sub-circuit blocks, this includes the full depth of these blocks. If you just want the number of top-level layers, use num_layers().

Returns

int

property width

The circuit width.

This is the number of qubits on which the circuit acts. This includes qubits that only idle, but are included as part of the circuit according to self.line_labels.

Returns

int

property size

Returns the circuit size.

This is the sum of the sizes of all the gates in the circuit. A gate that acts on n-qubits has a size of n, with the exception of the idle which has a size of 0. Hence, the circuit is given by: size = depth * num_lines - num_1Q_idles.

Returns

int

property duration

property num_multiq_gates

The number of multi-qubit (2+ qubits) gates in the circuit.

(Note that this cannot distinguish between “true” multi-qubit gates and gate that have been defined to act on more than one qubit but that represent some tensor-product gate.)

Returns

int

default_expand_subcircuits = True

classmethod cast(obj)

Convert obj into a Circuit.

Parameters

objobject: Object to convert

Returns

Circuit

classmethod from_tuple(tup)

Creates a Circuit from a tuple

Parameters

tuptuple: The tuple to convert.

Returns

Circuit

to_label(nreps=1)

Construct and return this entire circuit as a CircuitLabel.

Note: occurrence-id information is not stored in a circuit label, so circuits that differ only in their occurence_id will return circuit labels that are equal.

Parameters

nrepsint, optional: The number of times this circuit will be repeated (CircuitLabels support exponentiation and you can specify this here).

Returns

CircuitLabel

repeat(ntimes, expand='default')

Repeat this circuit ntimes times.

Parameters

ntimesint: Number of repetitions.
expandbool or “default”, optional: When False, the returned circuit will contain a CircuitLabel encapsulating the repetitions without explicitly storing them. When True, the returned circuit will be expanded into the ntimes repetitions. “default” means to use the value in the class variable Circuit.default_expand_subcircuits.

copy(editable='auto')

Returns a copy of the circuit.

Parameters

editable{True,False,”auto”}: Whether returned copy is editable. If “auto” is given, then the copy is editable if and only if this Circuit is.

Returns

Circuit

clear(): Removes all the gates in a circuit (preserving the number of lines).

Returns

None

extract_labels(layers=None, lines=None, strict=True)

Get a subregion - a “rectangle” - of this Circuit.

This can be used to select multiple layers and/or lines of this Circuit. The strict argument controls whether gates need to be entirely within the given rectangle or can be intersecting it. If layers is a single integer then a Label is returned (representing a layer or a part of a layer), otherwise a Circuit is returned.

Parameters

layersint, slice, or list/tuple of ints: Which layers to select (the horizontal dimension of the selection rectangle). Layers are always selected by index, and this argument can be a single (integer) index - in which case a Label is returned - or multiple indices as given by a slice or list - in which case a Circuit is returned. Note that, even though we speak of a “rectangle”, layer indices do not need to be contiguous. The special value None selects all layers.
linesstr/int, slice, or list/tuple of strs/ints: Which lines to select (the vertical dimension of the selection rectangle). Lines are selected by their line-labels (elements of the circuit’s .line_labels property), which can be strings and/or integers. A single or multiple line-labels can be specified. If the line labels are integers a slice can be used, otherwise a list or tuple of labels is the only way to select multiple of them. Note that line-labels do not need to be contiguous. The special value None selects all lines.
strictbool, optional: When True, only gates lying completely within the selected region are included in the return value. If a gate straddles the region boundary (e.g. if we select just line 1 and the circuit contains “Gcnot:1:2”) then it is silently not-included in the returned label or circuit. If False, then gates which straddle the region boundary are included. Note that this may result in a Label or Circuit containing more line labels than where requested in the call to extract_labels(…)..

Returns

Label or Circuit: The requested portion of this circuit, given as a Label if layers is a single integer and as a Circuit otherwise. Note: if you want a Circuit when only selecting one layer, set layers to a slice or tuple containing just a single index.

set_labels(lbls, layers=None, lines=None)

Write lbls to the block defined by the layers and lines arguments.

Note that lbls can be anything interpretable as a Label or list of labels.

Parameters

lblsLabel, list/tuple of Labels, or Circuit: When layers is a single integer, lbls should be a single “layer label” of type Label. Otherwise, lbls should be a list or tuple of Label objects with length equal to the number of layers being set. A Circuit may also be used in this case.
layersint, slice, or list/tuple of ints: Which layers to set (the horizontal dimension of the destination rectangle). Layers are always selected by index, and this argument can be a single (integer) index or multiple indices as given by a slice or list. Note that these indices do not need to be contiguous. The special value None stands for all layers.
linesstr/int, slice, or list/tuple of strs/ints: Which lines to set (the vertical dimension of the destination rectangle). Lines are selected by their line-labels, which can be strings and/or integers. A single or multiple line-labels can be specified. If the line labels are integers a slice can be used, otherwise a list or tuple of labels is the only way to specify multiple of them. The line-labels do not need to be contiguous. The special value None stands for all lines, and in this case new lines will be created if there are new state-space labels in lbls (when lines is not None an error is raised instead).

Returns

None

insert_idling_layers(insert_before, num_to_insert, lines=None)

Inserts into this circuit one or more idling (blank) layers, returning a copy.

By default, complete layer(s) are inserted. The lines argument allows you to insert partial layers (on only a subset of the lines).

Parameters

insert_beforeint: The layer index to insert the new layers before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The special value None inserts at the end.
num_to_insertint: The number of new layers to insert.
linesstr/int, slice, or list/tuple of strs/ints, optional: Which lines should have new layers (blank circuit space) inserted into them. A single or multiple line-labels can be specified, similarly as in extract_labels(). The default value None stands for all lines.

Returns

Circuit

insert_idling_layers_inplace(insert_before, num_to_insert, lines=None)

Inserts into this circuit one or more idling (blank) layers.

By default, complete layer(s) are inserted. The lines argument allows you to insert partial layers (on only a subset of the lines).

Parameters

insert_beforeint: The layer index to insert the new layers before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The special value None inserts at the end.
num_to_insertint: The number of new layers to insert.
linesstr/int, slice, or list/tuple of strs/ints, optional: Which lines should have new layers (blank circuit space) inserted into them. A single or multiple line-labels can be specified, similarly as in extract_labels(). The default value None stands for all lines.

Returns

None

insert_labels_into_layers(lbls, layer_to_insert_before, lines=None)

Inserts into this circuit the contents of lbls into new full or partial layers, returning a copy.

By default, complete layer(s) are inserted. The lines argument allows you to insert partial layers (on only a subset of the lines).

Parameters

lblslist/tuple of Labels, or Circuit: The full or partial layer labels to insert. The length of this list, tuple, or circuit determines the number of layers which are inserted.
layer_to_insert_beforeint: The layer index to insert lbls before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The special value None inserts at the end.
linesstr/int, slice, or list/tuple of strs/ints, optional: Which lines should have lbls inserted into them. Currently this can only be a larger set than the set of line labels present in lbls (in future versions this may allow filtering of lbls). value None stands for all lines.

Returns

Circuit

insert_labels_into_layers_inplace(lbls, layer_to_insert_before, lines=None)

Inserts into this circuit the contents of lbls into new full or partial layers.

By default, complete layer(s) are inserted. The lines argument allows you to insert partial layers (on only a subset of the lines).

Parameters

lblslist/tuple of Labels, or Circuit: The full or partial layer labels to insert. The length of this list, tuple, or circuit determines the number of layers which are inserted.
layer_to_insert_beforeint: The layer index to insert lbls before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The special value None inserts at the end.
linesstr/int, slice, or list/tuple of strs/ints, optional: Which lines should have lbls inserted into them. Currently this can only be a larger set than the set of line labels present in lbls (in future versions this may allow filtering of lbls). value None stands for all lines.

Returns

None

insert_idling_lines(insert_before, line_labels)

Insert one or more idling (blank) lines into this circuit, returning a copy.

Parameters

insert_beforestr or int: The line label to insert new lines before. The special value None inserts lines at the bottom of this circuit.
line_labelslist or tuple: A list or tuple of the new line labels to insert (can be integers and/or strings).

Returns

Circuit

insert_idling_lines_inplace(insert_before, line_labels)

Insert one or more idling (blank) lines into this circuit.

Parameters

insert_beforestr or int: The line label to insert new lines before. The special value None inserts lines at the bottom of this circuit.
line_labelslist or tuple: A list or tuple of the new line labels to insert (can be integers and/or strings).

Returns

None

insert_labels_as_lines_inplace(lbls, layer_to_insert_before=None, line_to_insert_before=None, line_labels='auto')

Inserts into this circuit the contents of lbls into new lines.

By default, lbls is inserted at the beginning of the new lines(s). The layer_to_insert_before argument allows you to insert lbls beginning at a layer of your choice.

Parameters

lblslist/tuple of Labels, or Circuit: A list of layer labels to insert as new lines. The state-space (line) labels within lbls must not overlap with that of this circuit or an error is raised. If lbls contains more layers than this circuit currently has, new layers are added automatically.
layer_to_insert_beforeint: The layer index to insert lbls before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The default value of None inserts at the beginning.
line_to_insert_beforestr or int: The line label to insert the new lines before. The default value of None inserts lines at the bottom of the circuit.
line_labelslist, tuple, or “auto”: The labels of the new lines being inserted. If “auto”, then these are inferred from lbls.

Returns

None

insert_labels_as_lines(lbls, layer_to_insert_before=None, line_to_insert_before=None, line_labels='auto')

Inserts into this circuit the contents of lbls into new lines, returning a copy.

By default, lbls is inserted at the beginning of the new lines(s). The layer_to_insert_before argument allows you to insert lbls beginning at a layer of your choice.

Parameters

lblslist/tuple of Labels, or Circuit: A list of layer labels to insert as new lines. The state-space (line) labels within lbls must not overlap with that of this circuit or an error is raised. If lbls contains more layers than this circuit currently has, new layers are added automatically.
layer_to_insert_beforeint: The layer index to insert lbls before. Can be from 0 (insert at the beginning) to len(self)-1 (insert at end), and negative indexing can be used to insert relative to the last layer. The default value of None inserts at the beginning.
line_to_insert_beforestr or int: The line label to insert the new lines before. The default value of None inserts lines at the bottom of the circuit.
line_labelslist, tuple, or “auto”: The labels of the new lines being inserted. If “auto”, then these are inferred from lbls.

Returns

None

clear_labels(layers=None, lines=None, clear_straddlers=False)

Removes all the gates within the given circuit region. Does not reduce the number of layers or lines.

Parameters

layersint, slice, or list/tuple of ints: Defines the horizontal dimension of the region to clear. See extract_labels() for details.
linesstr/int, slice, or list/tuple of strs/ints: Defines the vertical dimension of the region to clear. See extract_labels() for details.
clear_straddlersbool, optional: Whether or not gates which straddle cleared and non-cleared lines should be cleared. If False and straddling gates exist, an error will be raised.

Returns

None

delete_layers(layers=None)

Deletes one or more layers from the circuit.

Parameters

layersint, slice, or list/tuple of ints: The layer index or indices to delete. See extract_labels() for details.

Returns

None

delete_lines(lines, delete_straddlers=False)

Deletes one or more lines from the circuit.

Parameters

linesstr/int, slice, or list/tuple of strs/ints: The line label(s) to delete. See extract_labels() for details.
delete_straddlersbool, optional: Whether or not gates which straddle deleted and non-deleted lines should be removed. If False and straddling gates exist, an error will be raised.

Returns

None

to_pythonstr(op_labels)

Convert this circuit to an “encoded” python string.

In the returned string each operation label is represented as a single character, starting with ‘A’ and continuing down the alphabet. This can be useful for processing operation sequences using python’s string tools (regex in particular).

Parameters

op_labelstuple: An iterable containing at least all the layer-Labels that appear in this Circuit, and which will be mapped to alphabet characters, beginning with ‘A’.

Returns

string: The converted operation sequence.

Examples

(‘Gx’,’Gx’,’Gy’,’Gx’) => “AABA”

classmethod from_pythonstr(python_string, op_labels)

Decode an “encoded string” into a Circuit.

Create a Circuit from a python string where each operation label is represented as a single character, starting with ‘A’ and continuing down the alphabet. This performs the inverse of to_pythonstr().

Parameters

python_stringstring: string whose individual characters correspond to the operation labels of a operation sequence.
op_labelstuple: tuple containing all the operation labels that will be mapped from alphabet characters, beginning with ‘A’.

circuitA Circuit object: The circuit to be tensored.
line_orderList, optional: A list of all the line labels specifying the order of the circuit in the updated circuit. If None, the lines of circuit are added below the lines of this circuit. Note that, for many purposes, the ordering of lines of the circuit is irrelevant.

Returns

Circuit

replace_layer_with_circuit_inplace(circuit, j)

Replaces the j-th layer of this circuit with circuit.

Parameters

circuitCircuit: The circuit to insert
jint: The layer index to replace.

Returns

None

replace_layer_with_circuit(circuit, j)

Replaces the j-th layer of this circuit with circuit, returning a copy.

Parameters

circuitCircuit: The circuit to insert
jint: The layer index to replace.

Returns

Circuit

replace_gatename_inplace(old_gatename, new_gatename)

Changes the name of a gate throughout this Circuit.

Note that the name is only a part of the label identifying each gate, and doesn’t include the lines (qubits) a gate acts upon. For example, the “Gx:0” and “Gx:1” labels both have the same name but act on different qubits.

Parameters

old_gatenamestr: The gate name to replace.
new_gatenamestr: The name to replace old_gatename with.

Returns

None

replace_gatename(old_gatename, new_gatename)

Returns a copy of this Circuit except that old_gatename is changed to new_gatename.

Note that the “name” is only a part of the “label” identifying each gate, and doesn’t include the lines (qubits) a gate acts upon. For example, the “Gx:0” and “Gx:1” labels both have the same name but act on different qubits.

Parameters

old_gatenamestr: The gate name to replace.
new_gatenamestr: The name to replace old_gatename with.

Returns

Circuit

replace_gatename_with_idle_inplace(gatename)

Treats a given gatename as an idle gate throughout this Circuit.

This effectively removes this gate name from the circuit, and replaces a layer containing only this gate name with an idle layer.

Parameters

gatenamestr: The gate name to replace.

Returns

None

replace_gatename_with_idle(gatename)

Returns a copy of this Circuit with a given gatename treated as an idle gate.

This effectively removes this gate name from the circuit, and replaces a layer containing only this gate name with an idle layer.

Parameters

gatenamestr: The gate name to replace.

Returns

Circuit

replace_layer(old_layer, new_layer)

Returns a copy of this Circuit except that old_layer is changed to new_layer.

Parameters

old_layerstr or Label: The layer to find.
new_layerstr or Label: The layer to replace found layers with.

Returns

Circuit

replace_layers_with_aliases(alias_dict)

Performs a find and replace using layer aliases.

Returns a copy of this Circuit except that it’s layers that match keys of alias_dict are replaced with the corresponding values.

Parameters

alias_dictdict: A dictionary whose keys are layer Labels (or equivalent tuples or strings), and whose values are Circuits.

Returns

Circuit

change_gate_library(compilation, allowed_filter=None, allow_unchanged_gates=False, depth_compression=True, one_q_gate_relations=None)

Re-express a circuit over a different model.

Parameters

compilationdict or CompilationLibrary.

If a dictionary, the keys are some or all of the gates that appear in the circuit, and the values are replacement circuits that are normally compilations for each of these gates (if they are not, the action of the circuit will be changed). The circuits need not be on all of the qubits, and need only satisfy the requirements of the insert_circuit method. There must be a key for every gate except the self.identity gate, unless allow_unchanged_gates is False. In that case, gate that aren’t a key in this dictionary are left unchanged.

If a CompilationLibrary, this will be queried via the retrieve_compilation_of() method to find compilations for all of the gates in the circuit. So this CompilationLibrary must contain or be able to auto-generate compilations for the requested gates, except when allow_unchanged_gates is True. In that case, gates that a compilation is not returned for are left unchanged.

allowed_filterdict or set, optional

Specifies which gates are allowed to be used when generating compilations from compilation. Can only be not None if compilation is a CompilationLibrary. If a dict, keys must be gate names (like “Gcnot”) and values QubitGraph objects indicating where that gate (if it’s present in the library) may be used. If a set, then it specifies a set of qubits and any gate in the current library that is confined within that set is allowed. If None, then all gates within the library are allowed.

allow_unchanged_gatesbool, optional

Whether to allow some gates to remain unchanged, and therefore to be absent from compilation. When True such gates are left alone; when False an error is raised if any such gates are encountered.

depth_compressionbool, optional

Whether to perform depth compression after changing the gate library. If one_q_gate_relations is None this will only remove idle layers and compress the circuit by moving everything as far forward as is possible without knowledge of the action of any gates other than self.identity. See the depth_compression method for more details. Under most circumstances this should be true; if it is False changing gate library will often result in a massive increase in circuit depth.

one_q_gate_relationsdict, optional

mapperdict or function: A dictionary whose keys are the existing self.line_labels values and whose value are the new labels, or a function which takes a single (existing line-label) argument and returns a new line-label.

Returns

Circuit

reorder_lines_inplace(order)

Reorders the lines (wires/qubits) of the circuit.

Note that the ordering of the lines is unimportant for most purposes.

Parameters

orderlist: A list containing all of the circuit line labels (self.line_labels) in the order that the should be converted to.

Returns

None

reorder_lines(order)

Reorders the lines (wires/qubits) of the circuit, returning a copy.

Note that the ordering of the lines is unimportant for most purposes.

Parameters

orderlist: A list containing all of the circuit line labels (self.line_labels) in the order that the should be converted to.

Returns

Circuit

idling_lines(idle_layer_labels=None)

Returns the line labels corresponding to idling lines.

Parameters

idle_layer_labelsiterable, optional: A list or tuple of layer-labels that should be treated as idle operations, so their presence will not disqualify a line from being “idle”. E.g. [“Gi”] will cause “Gi” layers to be considered idle layers.

Returns

tuple

delete_idling_lines_inplace(idle_layer_labels=None)

Removes from this circuit all lines that are idling at every layer.

Parameters

idle_layer_labelsiterable, optional: A list or tuple of layer-labels that should be treated as idle operations, so their presence will not disqualify a line from being “idle”. E.g. [“Gi”] will cause “Gi” layers to be considered idle layers.

Returns

None

delete_idling_lines(idle_layer_labels=None)

Removes from this circuit all lines that are idling at every layer, returning a copy.

Parameters

idle_layer_labelsiterable, optional: A list or tuple of layer-labels that should be treated as idle operations, so their presence will not disqualify a line from being “idle”. E.g. [“Gi”] will cause “Gi” layers to be considered idle layers.

Returns

Circuit

replace_with_idling_line_inplace(line_label, clear_straddlers=True)

Converts the specified line to an idling line, by removing all its gates.

If there are any multi-qubit gates acting on this line, this function will raise an error when clear_straddlers=False.

Parameters

line_labelstr or int: The label of the line to convert to an idling line.
clear_straddlersbool, optional: Whether or not gates which straddle the line_label should also be cleared. If False and straddling gates exist, an error will be raised.

Returns

None

reverse_inplace(): Reverses the order of the circuit.

Returns

None

delete_idle_layers_inplace()

Deletes all layers in this circuit that contain no gate operations.

One of the steps of the depth_compression() method.

Returns

bool: False if the circuit is unchanged, and True otherwise.

compress_depth_inplace(one_q_gate_relations=None, verbosity=0)

Compresses the depth of this circuit using very simple re-write rules.

If one_q_gate_relations is provided, all sequences of 1-qubit gates in the circuit are compressed as far as is possible using only the pair-wise combination rules provided by this dict (see below).
All gates are shifted forwarded as far as is possible without any knowledge of what any of the gates are.
All idle-only layers are deleted.

Parameters

one_q_gate_relationsdict

Keys that are pairs of strings, corresponding to 1-qubit gate names, with values that are a single string, also corresponding to a 1-qubit gate name. Whenever a 1-qubit gate with name name1 is followed in the circuit by a 1-qubit gate with name2 then, if one_q_gate_relations[name1,name2] = name3, name1 -> name3 and name2 -> self.identity, the identity name in the circuit. Moreover, this is still implemented when there are self.identity gates between these 1-qubit gates, and it is implemented iteratively in the sense that if there is a sequence of 1-qubit gates with names name1, name2, name3, … and there are relations for all of (name1,name2) -> name12, (name12,name3) -> name123 etc then the entire sequence of 1-qubit gates will be compressed into a single possibly non-idle 1-qubit gate followed by idle gates in place of the previous 1-qubit gates.

If a QubitProcessorSpec object has been created for the gates/device in question, the QubitProcessorSpec.oneQgate_relations is the appropriate (and auto-generated) one_q_gate_relations.

Note that this function will not compress sequences of 1-qubit gates that cannot be compressed by independently inspecting sequential non-idle pairs (as would be the case with, for example, Gxpi Gzpi Gxpi Gzpi, if the relation did not know that (Gxpi,Gzpi) -> Gypi, even though the sequence is the identity).

verbosityint, optional

int

num_nq_gates(nq)

The number of nq-qubit gates in the circuit.

(Note that this cannot distinguish between “true” nq-qubit gates and gate that have been defined to act on nq qubits but that represent some tensor-product gate.)

Parameters

nqint: The qubit-count of the gates to count. For example, if nq == 3, this function returns the number of 3-qubit gates.

Returns

int

format_display_str(width=80)

Formats a string for displaying this circuit suject to a maximum width.

Parameters

widthint, optional: The maximum width in characters. If the circuit is longer than this width it is wrapped using multiple lines (like a musical score).

Returns

str

convert_to_cirq(qubit_conversion, wait_duration=None, gatename_conversion=None, idle_gate_name='Gi')

Converts this circuit to a Cirq circuit.

Parameters

qubit_conversiondict: Mapping from qubit labels (e.g. integers) to Cirq qubit objects.
wait_durationcirq.Duration, optional: If no gatename_conversion dict is given, the idle operation is not converted to a gate. If wait_diration is specified and gatename_conversion is not specified, then the idle operation will be converted to a cirq.WaitGate with the specified duration.
gatename_conversiondict, optional: If not None, a dictionary that converts the gatenames in the circuit to the Cirq gates that will appear in the Cirq circuit. If only standard pyGSTi names are used (e.g., ‘Gh’, ‘Gp’, ‘Gcnot’, ‘Gcphase’, etc) this dictionary need not be specified, and an automatic conversion to the standard Cirq names will be implemented.
idle_gate_namestr, optional: Name to use for idle gates. Defaults to ‘Gi’

Returns

A Cirq Circuit object.

classmethod from_cirq(circuit, qubit_conversion=None, cirq_gate_conversion=None, remove_implied_idles=True, global_idle_replacement_label='auto')

Converts and instantiates a pyGSTi Circuit object from a Cirq Circuit object.

Parameters

circuitcirq Circuit: The cirq Circuit object to parse into a pyGSTi circuit.
qubit_conversiondict, optional (default None): A dictionary specifying a mapping between cirq qubit objects and pyGSTi qubit labels (either integers or strings). If None, then a default mapping is created.
cirq_gate_conversiondict, optional (default None): If specified a dictionary with keys given by cirq gate objects, and values given by pygsti gate names which overrides the built-in conversion dictionary used by default.
remove_implied_idlesbool, optional (default True): A flag indicating whether to remove explicit idles that are part of a circuit layer containing other explicitly specified gates (i.e., whether to abide by the normal pyGSTi implicit idle convention).
global_idle_replacement_labelstring or Label or None, optional (default ‘auto’): An option specified for the handling of global idle layers. If None, no replacement of global idle layers is performed and a verbatim conversion from the cirq layer is performed. If the string ‘auto’, then the behavior is to replace global idle layers with the gate label Label(()), which is the special syntax for the global idle layer, stylized typically as ‘[]’. If another string then replace with a gate label with the specified name acting on all of the qubits appearing in the cirq circuit. If a Label object, use this directly, this does not check for compatibility so it is up to the user to ensure the labels are compatible.

Returns

pygsti_circuit: A pyGSTi Circuit instance equivalent to the specified Cirq one.

convert_to_quil(num_qubits=None, gatename_conversion=None, qubit_conversion=None, readout_conversion=None, block_between_layers=True, block_idles=True, gate_declarations=None)

Converts this circuit to a quil string.

Parameters

num_qubitsint, optional: The number of qubits for the quil file. If None, then this is assumed to equal the number of line labels in this circuit.
gatename_conversiondict, optional: A dictionary mapping gate names contained in this circuit to the corresponding gate names used in the rendered quil. If None, a standard set of conversions is used (see standard_gatenames_quil_conversions()).
qubit_conversiondict, optional: If not None, a dictionary converting the qubit labels in the circuit to the desired qubit labels in the quil output. Can be left as None if the qubit labels are either (1) integers, or (2) of the form ‘Qi’ for integer i. In this case they are converted to integers (i.e., for (1) the mapping is trivial, for (2) the mapping strips the ‘Q’).
readout_conversiondict, optional: If not None, a dictionary converting the qubit labels mapped through qubit_conversion to the bit labels for readot. E.g. Suppose only qubit 2 (on Rigetti hardware) is in use. Then the pyGSTi string will have only one qubit (labeled 0); it will get remapped to 2 via qubit_conversion={0:2}. At the end of the quil circuit, readout should go recorded in bit 0, so readout_conversion = {0:0}. (That is, qubit with pyGSTi label 0 gets read to Rigetti bit 0, even though that qubit has Rigetti label 2.)
block_between_layersbool, optional: When True, add in a barrier after every circuit layer. Including such “pragma” blocks can be important for QCVV testing, as this can help reduce the “behind-the-scenes” compilation (beyond necessary conversion to native instructions) experience by the circuit.
block_idlesbool, optional: In the special case of global idle gates, pragma-block barriers are inserted even when block_between_layers=False. Set block_idles=False to disable this behavior, whcih typically results in global idle gates being removed by the compiler.
gate_declarationsdict, optional: If not None, a dictionary that provides unitary maps for particular gates that are not already in the quil syntax.

Returns

str: A quil string.

convert_to_openqasm(num_qubits=None, standard_gates_version='u3', gatename_conversion=None, qubit_conversion=None, block_between_layers=True, block_between_gates=False, include_delay_on_idle=True, gateargs_map=None)

Converts this circuit to an openqasm string.

Parameters

num_qubitsint, optional: The number of qubits for the openqasm file. If None, then this is assumed to equal the number of line labels in this circuit.
versionstring, optional: Either ‘u3’ or ‘x-sx-rz’. Specifies the naming convention for the QASM gates. With ‘u3’, all single-qubit gates are specified in terms of the ‘u3’ gate, used by IBM and QisKit until ~2021 (see the qasm_u3 function). With ‘x-sx-rz’, all single-gates are specified in terms of ‘x’ (an x pi rotation), ‘sx’ (an x pi/2 rotation) and ‘rz’ (a parameterized rotation around z by an angle theta).
gatename_conversiondict, optional: If not None, a dictionary that converts the gatenames in the circuit to the gatenames that will appear in the openqasm output. If only standard pyGSTi names are used (e.g., ‘Gh’, ‘Gp’, ‘Gcnot’, ‘Gcphase’, etc) this dictionary need not be specified, and an automatic conversion to the standard openqasm names will be implemented.
qubit_conversiondict, optional: If not None, a dictionary converting the qubit labels in the circuit to the desired qubit labels in the openqasm output. Can be left as None if the qubit labels are either (1) integers, or (2) of the form ‘Qi’ for integer i. In this case they are converted to integers (i.e., for (1) the mapping is trivial, for (2) the mapping strips the ‘Q’).
block_between_layersbool, optional: When True, add in a barrier after every circuit layer. Including such barriers can be important for QCVV testing, as this can help reduce the “behind-the-scenes” compilation (beyond necessary conversion to native instructions) experience by the circuit.
block_between_gates: bool, optional: When True, add in a barrier after every gate (effectively serializing the circuit). Defaults to False.
include_delay_on_idle: bool, optional: When True, includes a delay operation on implicit idles in each layer, as per Qiskit’s OpenQASM 2.0 convention after the deprecation of the id operation. Defaults to True, which is commensurate with legacy usage of this function. However, this can now be set to False to avoid this behaviour if generating actually valid OpenQASM (with no opaque delay instruction) is desired.
gateargs_mapdict, optional: If not None, a dict that maps strings (representing pyGSTi standard gate names) to functions that map the parameters of a pyGSTi gate to a string to be combined with the QASM name to specify the specific gate, in QASM. If only standard pyGSTi names are used (e.g., ‘Gh’, ‘Gzr’, ‘Gczr, etc) or none of the gates are parameterized, this dictionary need not be specified, and an automatic conversion to the standard openqasm format will be implemented.

Returns

str: An openqasm string.

simulate(model, return_all_outcomes=False)

Compute the outcome probabilities of this Circuit using model as a model for the gates.

The order of the outcome strings (e.g., ‘0100’) is w.r.t. to the ordering of the qubits in the circuit. That is, the ith element of the outcome string corresponds to the qubit with label self.line_labels[i].

Parameters

modelModel: A description of the gate and SPAM operations corresponding to the labels stored in this Circuit. If this model is over more qubits than the circuit, the output will be the probabilities for the qubits in the circuit marginalized, if possible over the other qubits. But, the simulation is over the full set of qubits in the model, and so the time taken for the simulation scales with the number of qubits in the model. For models where “spectator” qubits do not affect the qubits in this circuit (such as with perfect gates), more efficient simulations will be obtained by first creating a model only over the qubits in this circuit.
return_all_outcomesbool, optional: Whether to include outcomes in the returned dictionary that have zero probability. When False, the threshold for discarding an outcome as z ero probability is 10^-12.

Returns

probsdictionary: A dictionary with keys equal to all (return_all_outcomes is True) or possibly only some (return_all_outcomes is False) of the possible outcomes, and values that are float probabilities.

done_editing()

Make this circuit read-only, so that it can be hashed (e.g. used as a dictionary key).

This is done automatically when attempting to hash a Circuit for the first time, so there’s calling this function can usually be skipped (but it’s good for code clarity).

Returns

None

expand_instruments_and_separate_povm(model, observed_outcomes=None)

Creates a dictionary of SeparatePOVMCircuit objects from expanding the instruments of this circuit.

Each key of the returned dictionary replaces the instruments in this circuit with a selection of their members. (The size of the resulting dictionary is the product of the sizes of each instrument appearing in this circuit when observed_outcomes is None). Keys are stored as SeparatePOVMCircuit objects so it’s easy to keep track of which POVM outcomes (effects) correspond to observed data. This function is, for the most part, used internally to process a circuit before computing its outcome probabilities.

Parameters

modelModel

The model used to provide necessary details regarding the expansion, including:

default SPAM layers
definitions of instrument-containing layers
expansions of individual instruments and POVMs

Returns

OrderedDict: A dict whose keys are SeparatePOVMCircuit objects and whose values are tuples of the outcome labels corresponding to this circuit, one per POVM effect held in the key.

class pygsti.circuits.circuit.CompressedCircuit(circuit, min_len_to_compress=20, max_period_to_look_for=20)

Bases: object

A “compressed” Circuit that requires less disk space.

The tuple of circuit layers is compressed using a custom algorithm which looks for repeated portions of the circuit.

One place where CompressedCircuit objects can be useful is when saving large lists of long operation sequences in some non-human-readable format (e.g. pickle). CompressedCircuit objects cannot be used in place of Circuit objects within pyGSTi, and so are not useful when manipulating and running algorithms which use operation sequences.

Parameters

circuitCircuit: The operation sequence object which is compressed to create a new CompressedCircuit object.
min_len_to_compressint, optional: The minimum length string to compress. If len(circuit) is less than this amount its tuple is returned.
max_period_to_look_forint, optional: The maximum period length to use when searching for periodic structure within circuit. Larger values mean the method takes more time but could result in better compressing.

Create a new CompressedCircuit object

Parameters

circuitCircuit: The operation sequence object which is compressed to create a new CompressedCircuit object.
min_len_to_compressint, optional: The minimum length string to compress. If len(circuit) is less than this amount its tuple is returned.
max_period_to_look_forint, optional: The maximum period length to use when searching for periodic structure within circuit. Larger values mean the method takes more time but could result in better compressing.

expand(): Expands this compressed operation sequence into a Circuit object.

Returns

Circuit

static compress_op_label_tuple(circuit, min_len_to_compress=20, max_period_to_look_for=20)

Compress a operation sequence.

The result is tuple with a special compressed- gate-string form form that is not useable by other GST methods but is typically shorter (especially for long operation sequences with a repetative structure) than the original operation sequence tuple.

Parameters

circuittuple of operation labels or Circuit: The operation sequence to compress.
min_len_to_compressint, optional: The minimum length string to compress. If len(circuit) is less than this amount its tuple is returned.
max_period_to_look_forint, optional: The maximum period length to use when searching for periodic structure within circuit. Larger values mean the method takes more time but could result in better compressing.

Returns

tuple: The compressed (or raw) operation sequence tuple.

static expand_op_label_tuple(compressed_op_labels)

Expand a compressed tuple (created with compress_op_label_tuple()) into a tuple of operation labels.

Parameters

compressed_op_labelstuple: a tuple in the compressed form created by compress_op_label_tuple(…).

Returns

tuple: A tuple of operation labels specifying the uncompressed operation sequence.

class pygsti.circuits.circuit.SeparatePOVMCircuit(circuit_without_povm, povm_label, effect_labels)

Bases: object

Separately holds a POVM-less Circuit object, a POVM label, and a list of effect labels.

This is often used to hold “expanded” circuits whose instrument labels have been replaced with specific instrument members and whose POVMs have simillarly been “expanded” except that we keep the entire expanded POVM together in this one data structure. (There’s no especially good reason for this other than practicality - that since almost all circuits end with a POVM, holding each POVM outcome (effect) separately would be very wasteful.

property full_effect_labels

pygsti.circuits.circuit

Module Contents

Classes

Functions

Attributes

Parameters

Returns

Parameters

Attributes

Parameters

Returns

Returns

Returns

Returns

Returns

Returns

Returns

Returns

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Parameters

Returns

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Examples

Parameters

Returns

Examples

Parameters

Returns

Parameters

Returns

Returns

Returns

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

Parameters

Returns

`pygsti.circuits.circuit`