qml.draw¶
-
draw
(qnode, wire_order=None, show_all_wires=False, decimals=2, max_length=100, show_matrices=True, **kwargs)[source]¶ Create a function that draws the given qnode or quantum function.
- Parameters
qnode (QNode or Callable) – the input QNode or quantum function that is to be drawn.
wire_order (Sequence[Any]) – the order (from top to bottom) to print the wires of the circuit. If not provided, the wire order defaults to the device wires. If device wires are not available, the circuit wires are sorted if possible.
show_all_wires (bool) – If True, all wires, including empty wires, are printed.
decimals (int) – How many decimal points to include when formatting operation parameters.
None
will omit parameters from operation labels.max_length (int) – Maximum string width (columns) when printing the circuit
show_matrices=False (bool) – show matrix valued parameters below all circuit diagrams
- Keyword Arguments
level (None, str, int, slice) – An indication of what transforms to apply before drawing. Check
get_transform_program()
for more information on the allowed values and usage details of this argument.expansion_strategy (str) –
The strategy to use when circuit expansions or decompositions are required.
gradient
: The QNode will attempt to decompose the internal circuit such that all circuit operations are supported by the gradient method.device
: The QNode will attempt to decompose the internal circuit such that all circuit operations are natively supported by the device.
- Returns
A function that has the same argument signature as
qnode
. When called, the function will draw the QNode/qfunc.
Note
At most, one of
level
orexpansion_strategy
needs to be provided. If neither is provided,qnode.expansion_strategy
will be used instead. Users are encouraged to predominantly uselevel
, as it allows for the same values asexpansion_strategy
and offers more flexibility in choosing the desired transforms/expansions.Warning
The
expansion_strategy
argument is deprecated and will be removed in version 0.39. Use thelevel
argument instead to specify the resulting tape you want.Example
@qml.qnode(qml.device('lightning.qubit', wires=2)) def circuit(a, w): qml.Hadamard(0) qml.CRX(a, wires=[0, 1]) qml.Rot(*w, wires=[1], id="arbitrary") qml.CRX(-a, wires=[0, 1]) return qml.expval(qml.Z(0) @ qml.Z(1))
>>> print(qml.draw(circuit)(a=2.3, w=[1.2, 3.2, 0.7])) 0: ──H─╭●─────────────────────────────────────────╭●─────────┤ ╭<Z@Z> 1: ────╰RX(2.30)──Rot(1.20,3.20,0.70,"arbitrary")─╰RX(-2.30)─┤ ╰<Z@Z>
Usage Details
By specifying the
decimals
keyword, parameters are displayed to the specified precision.>>> print(qml.draw(circuit, decimals=4)(a=2.3, w=[1.2, 3.2, 0.7])) 0: ──H─╭●─────────────────────────────────────────────────╭●───────────┤ ╭<Z@Z> 1: ────╰RX(2.3000)──Rot(1.2000,3.2000,0.7000,"arbitrary")─╰RX(-2.3000)─┤ ╰<Z@Z>
Parameters can be omitted by requesting
decimals=None
:>>> print(qml.draw(circuit, decimals=None)(a=2.3, w=[1.2, 3.2, 0.7])) 0: ──H─╭●────────────────────╭●──┤ ╭<Z@Z> 1: ────╰RX──Rot("arbitrary")─╰RX─┤ ╰<Z@Z>
If the parameters are not acted upon by classical processing like
-a
, thenqml.draw
can handle string-valued parameters as well:>>> @qml.qnode(qml.device('lightning.qubit', wires=1)) ... def circuit2(x): ... qml.RX(x, wires=0) ... return qml.expval(qml.Z(0)) >>> print(qml.draw(circuit2)("x")) 0: ──RX(x)─┤ <Z>
When requested with
show_matrices=True
(the default), matrix valued parameters are printed below the circuit. Forshow_matrices=False
, they are not printed:>>> @qml.qnode(qml.device('default.qubit', wires=2)) ... def circuit3(): ... qml.QubitUnitary(np.eye(2), wires=0) ... qml.QubitUnitary(-np.eye(4), wires=(0,1)) ... return qml.expval(qml.Hermitian(np.eye(2), wires=1)) >>> print(qml.draw(circuit3)()) 0: ──U(M0)─╭U(M1)─┤ 1: ────────╰U(M1)─┤ <𝓗(M0)> M0 = [[1. 0.] [0. 1.]] M1 = [[-1. -0. -0. -0.] [-0. -1. -0. -0.] [-0. -0. -1. -0.] [-0. -0. -0. -1.]] >>> print(qml.draw(circuit3, show_matrices=False)()) 0: ──U(M0)─╭U(M1)─┤ 1: ────────╰U(M1)─┤ <𝓗(M0)>
The
max_length
keyword warps long circuits:rng = np.random.default_rng(seed=42) shape = qml.StronglyEntanglingLayers.shape(n_wires=3, n_layers=3) params = rng.random(shape) @qml.qnode(qml.device('lightning.qubit', wires=3)) def longer_circuit(params): qml.StronglyEntanglingLayers(params, wires=range(3)) return [qml.expval(qml.Z(i)) for i in range(3)]
>>> print(qml.draw(longer_circuit, max_length=60, level="device")(params)) 0: ──Rot(0.77,0.44,0.86)─╭●────╭X──Rot(0.45,0.37,0.93)─╭●─╭X 1: ──Rot(0.70,0.09,0.98)─╰X─╭●─│───Rot(0.64,0.82,0.44)─│──╰● 2: ──Rot(0.76,0.79,0.13)────╰X─╰●──Rot(0.23,0.55,0.06)─╰X─── ───Rot(0.83,0.63,0.76)──────────────────────╭●────╭X─┤ <Z> ──╭X────────────────────Rot(0.35,0.97,0.89)─╰X─╭●─│──┤ <Z> ──╰●────────────────────Rot(0.78,0.19,0.47)────╰X─╰●─┤ <Z>
The
wire_order
keyword specifies the order of the wires from top to bottom:>>> print(qml.draw(circuit, wire_order=[1,0])(a=2.3, w=[1.2, 3.2, 0.7])) 1: ────╭RX(2.30)──Rot(1.20,3.20,0.70,"arbitrary")─╭RX(-2.30)─┤ ╭<Z@Z> 0: ──H─╰●─────────────────────────────────────────╰●─────────┤ ╰<Z@Z>
If the device or
wire_order
has wires not used by operations, those wires are omitted unless requested withshow_all_wires=True
>>> empty_qfunc = lambda : qml.expval(qml.Z(0)) >>> empty_circuit = qml.QNode(empty_qfunc, qml.device('lightning.qubit', wires=3)) >>> print(qml.draw(empty_circuit, show_all_wires=True)()) 0: ───┤ <Z> 1: ───┤ 2: ───┤
Drawing also works on batch transformed circuits:
from functools import partial from pennylane import numpy as pnp @partial(qml.gradients.param_shift, shifts=[(0.1,)]) @qml.qnode(qml.device('default.qubit', wires=1)) def transformed_circuit(x): qml.RX(x, wires=0) return qml.expval(qml.Z(0))
>>> print(qml.draw(transformed_circuit)(pnp.array(1.0, requires_grad=True))) 0: ──RX(1.10)─┤ <Z> 0: ──RX(0.90)─┤ <Z>
The function also accepts quantum functions rather than QNodes. This can be especially helpful if you want to visualize only a part of a circuit that may not be convertible into a QNode, such as a sub-function that does not return any measurements.
>>> def qfunc(x): ... qml.RX(x, wires=[0]) ... qml.CNOT(wires=[0, 1]) >>> print(qml.draw(qfunc)(1.1)) 0: ──RX(1.10)─╭●─┤ 1: ───────────╰X─┤
Levels:
The
level
keyword argument allows one to select a subset of the transforms to apply on theQNode
before carrying out any drawing. Take, for example, this circuit:@qml.transforms.merge_rotations @qml.transforms.cancel_inverses @qml.qnode(qml.device("default.qubit"), diff_method="parameter-shift") def circ(weights, order): qml.RandomLayers(weights, wires=(0, 1)) qml.Permute(order, wires=(0, 1, 2)) qml.PauliX(0) qml.PauliX(0) qml.RX(0.1, wires=0) qml.RX(-0.1, wires=0) return qml.expval(qml.PauliX(0)) order = [2, 1, 0] weights = qml.numpy.array([[1.0, 20]])
One can print the circuit without any transforms applied by passing
level="top"
orlevel=0
:>>> print(qml.draw(circ, level="top")(weights, order)) 0: ─╭RandomLayers(M0)─╭Permute──X──X──RX(0.10)──RX(-0.10)─┤ <X> 1: ─╰RandomLayers(M0)─├Permute────────────────────────────┤ 2: ───────────────────╰Permute────────────────────────────┤ M0 = [[ 1. 20.]]
Or print the circuit after applying the transforms manually applied on the QNode (
merge_rotations
andcancel_inverses
):>>> print(qml.draw(circ, level="user", show_matrices=False)(weights, order)) 0: ─╭RandomLayers(M0)─╭Permute─┤ <X> 1: ─╰RandomLayers(M0)─├Permute─┤ 2: ───────────────────╰Permute─┤
To apply all of the transforms, including those carried out by the differentiation method and the device, use
level=None
:>>> print(qml.draw(circ, level=None, show_matrices=False)(weights, order)) 0: ──RY(1.00)──╭SWAP─┤ <X> 1: ──RX(20.00)─│─────┤ 2: ────────────╰SWAP─┤
Slices can also be passed to the
level
argument. So one can, for example, request that only themerge_rotations
transform is applied:>>> print(qml.draw(circ, level=slice(1, 2), show_matrices=False)(weights, order)) 0: ─╭RandomLayers(M0)─╭Permute──X──X─┤ <X> 1: ─╰RandomLayers(M0)─├Permute───────┤ 2: ───────────────────╰Permute───────┤