# -*- coding: utf-8 -*-
# This code is part of Qiskit.
#
# (C) Copyright IBM 2017, 2018.
#
# This code is licensed under the Apache License, Version 2.0. You may
# obtain a copy of this license in the LICENSE.txt file in the root directory
# of this source tree or at http://www.apache.org/licenses/LICENSE-2.0.
#
# Any modifications or derivative works of this code must retain this
# copyright notice, and modified files need to carry a notice indicating
# that they have been altered from the originals.
"""Choose a Layout by finding the most connected subset of qubits."""
import numpy as np
import scipy.sparse as sp
import scipy.sparse.csgraph as cs
from qiskit.transpiler.layout import Layout
from qiskit.transpiler.basepasses import AnalysisPass
from qiskit.transpiler.exceptions import TranspilerError
[docs]class DenseLayout(AnalysisPass):
"""Choose a Layout by finding the most connected subset of qubits.
This pass associates a physical qubit (int) to each virtual qubit
of the circuit (Qubit).
Note:
Even though a 'layout' is not strictly a property of the DAG,
in the transpiler architecture it is best passed around between passes
by being set in `property_set`.
"""
def __init__(self, coupling_map, backend_prop=None):
"""DenseLayout initializer.
Args:
coupling_map (Coupling): directed graph representing a coupling map.
backend_prop (BackendProperties): backend properties object
"""
super().__init__()
self.coupling_map = coupling_map
self.backend_prop = backend_prop
self.cx_mat = None
self.meas_arr = None
self.num_cx = 0
self.num_meas = 0
[docs] def run(self, dag):
"""Run the DenseLayout pass on `dag`.
Pick a convenient layout depending on the best matching
qubit connectivity, and set the property `layout`.
Args:
dag (DAGCircuit): DAG to find layout for.
Raises:
TranspilerError: if dag wider than self.coupling_map
"""
num_dag_qubits = sum([qreg.size for qreg in dag.qregs.values()])
if num_dag_qubits > self.coupling_map.size():
raise TranspilerError('Number of qubits greater than device.')
# Get avg number of cx and meas per qubit
ops = dag.count_ops()
if 'cx' in ops.keys():
self.num_cx = ops['cx']
if 'measure' in ops.keys():
self.num_meas = ops['measure']
# Compute the sparse cx_err matrix and meas array
device_qubits = self.coupling_map.size()
if self.backend_prop:
rows = []
cols = []
cx_err = []
for edge in self.coupling_map.get_edges():
for gate in self.backend_prop.gates:
if gate.qubits == edge:
rows.append(edge[0])
cols.append(edge[1])
cx_err.append(gate.parameters[0].value)
break
else:
continue
self.cx_mat = sp.coo_matrix((cx_err, (rows, cols)),
shape=(device_qubits,
device_qubits)).tocsr()
# Set measurement array
meas_err = []
for qubit_data in self.backend_prop.qubits:
for item in qubit_data:
if item.name == 'readout_error':
meas_err.append(item.value)
break
else:
continue
self.meas_arr = np.asarray(meas_err)
best_sub = self._best_subset(num_dag_qubits)
layout = Layout()
map_iter = 0
for qreg in dag.qregs.values():
for i in range(qreg.size):
layout[qreg[i]] = int(best_sub[map_iter])
map_iter += 1
self.property_set['layout'] = layout
def _best_subset(self, num_qubits):
"""Computes the qubit mapping with the best connectivity.
Args:
num_qubits (int): Number of subset qubits to consider.
Returns:
ndarray: Array of qubits to use for best connectivity mapping.
"""
if num_qubits == 1:
return np.array([0])
if num_qubits == 0:
return []
device_qubits = self.coupling_map.size()
cmap = np.asarray(self.coupling_map.get_edges())
data = np.ones_like(cmap[:, 0])
sp_cmap = sp.coo_matrix((data, (cmap[:, 0], cmap[:, 1])),
shape=(device_qubits, device_qubits)).tocsr()
best = 0
best_map = None
best_error = np.inf
best_sub = None
# do bfs with each node as starting point
for k in range(sp_cmap.shape[0]):
bfs = cs.breadth_first_order(sp_cmap, i_start=k, directed=False,
return_predecessors=False)
connection_count = 0
sub_graph = []
for i in range(num_qubits):
node_idx = bfs[i]
for j in range(sp_cmap.indptr[node_idx],
sp_cmap.indptr[node_idx + 1]):
node = sp_cmap.indices[j]
for counter in range(num_qubits):
if node == bfs[counter]:
connection_count += 1
sub_graph.append([node_idx, node])
break
if self.backend_prop:
curr_error = 0
# compute meas error for subset
avg_meas_err = np.mean(self.meas_arr)
meas_diff = np.mean(
self.meas_arr[bfs[0:num_qubits]])-avg_meas_err
if meas_diff > 0:
curr_error += self.num_meas*meas_diff
cx_err = np.mean([self.cx_mat[edge[0], edge[1]]
for edge in sub_graph])
if self.coupling_map.is_symmetric:
cx_err /= 2
curr_error += self.num_cx*cx_err
if connection_count >= best and curr_error < best_error:
best = connection_count
best_error = curr_error
best_map = bfs[0:num_qubits]
best_sub = sub_graph
else:
if connection_count > best:
best = connection_count
best_map = bfs[0:num_qubits]
best_sub = sub_graph
# Return a best mapping that has reduced bandwidth
mapping = {}
for edge in range(best_map.shape[0]):
mapping[best_map[edge]] = edge
new_cmap = [[mapping[c[0]], mapping[c[1]]] for c in best_sub]
rows = [edge[0] for edge in new_cmap]
cols = [edge[1] for edge in new_cmap]
data = [1]*len(rows)
sp_sub_graph = sp.coo_matrix((data, (rows, cols)),
shape=(num_qubits, num_qubits)).tocsr()
perm = cs.reverse_cuthill_mckee(sp_sub_graph)
best_map = best_map[perm]
return best_map