Qiskit is an open-source framework for working with noisy quantum computers at the level of pulses, circuits, and algorithms.
Qiskit is made up elements that work together to enable quantum computing. This element is Aqua (Algorithms for QUantum computing Applications) providing a library of cross-domain algorithms upon which domain-specific applications can be built.
Aqua includes domain application support for:
Note: the chemistry module was the first domain worked on. Aqua version 0.7.0 introduced a new optimization module for solving quadratic problems. At the time of writing the other domains have some logic in them but are not as fully realised. Future work is expected to build out functionality in all application areas.
Aqua was designed to be extensible, and uses a framework where algorithms and support objects used by algorithms, such as optimizers, variational forms, and oracles etc, are derived from a defined base class for the type. These along with other building blocks provide a means for end-users and developers alike to have flexibility and facilitate building and experimenting with different configurations and capability.
Note: Aqua provides some classical algorithms that take the same input data as quantum algorithms solving the same problem. For instance a Hamiltonian operator input to VQE can be used as an input to the NumPyEigensolver. This may be useful for near-term quantum experiments, for problems that can still be solved classically, as their outcome can be easily compared against a classical equivalent since the same input data can be used.
We encourage installing Qiskit via the pip tool (a python package manager), which installs all Qiskit elements, including Aqua.
pip install qiskitpip will handle all dependencies automatically and you will always install the latest (and well-tested) version.
If you want to work on the very latest work-in-progress versions, either to try features ahead of their official release or if you want to contribute to Aqua, then you can install from source. To do this follow the instructions in the documentation.
Note: there some optional packages that can be installed such as IBM CPLEX for Aqua and ab-initio chemistry libraries/programs. Refer to Optional Install information in the sections below.
Note: tutorials are undergoing revision and re-organization. Hence you may notice some content you will see referenced is under legacy_tutorials pending such re-work.
The qiskit.aqua package contains the core cross-domain algorithms and supporting logic to run
these on a quantum backend, whether a real device or simulator.
Note: while the packages below can be installed directly by pip install, e.g. pip install cplex
by doing so via the Aqua extra_requires, in this case pip install qiskit-aqua[cplex] will ensure
that a version compatible with Qiskit is installed.
- IBM CPLEX may be installed
to allow the use of the
ClassicalCPLEXclassical solver algorithm.pip install qiskit-aqua[cplex]may be used to install the community version. - PyTorch, may be installed either using command
pip install qiskit-aqua[torch]to install the package or refer to PyTorch getting started. PyTorch being installed will enable the neural networksPyTorchDiscriminatorcomponent to be used with the QGAN algorithm. - CVXPY, may be installed using command
pip install qiskit-aqua[cvpxy]to enable use of theQSVMand the classicalSklearnSVMalgorithms.
Now that Qiskit is installed, it's time to begin working with Aqua.
Let's try an experiment using Grover's algorithm to find a solution for a
Satisfiability (SAT) problem.
$ python
from qiskit import Aer
from qiskit.aqua.components.oracles import LogicalExpressionOracle
from qiskit.aqua.algorithms import Grover
sat_cnf = """
c Example DIMACS 3-sat
p cnf 3 5
-1 -2 -3 0
1 -2 3 0
1 2 -3 0
1 -2 -3 0
-1 2 3 0
"""
backend = Aer.get_backend('qasm_simulator')
oracle = LogicalExpressionOracle(sat_cnf)
algorithm = Grover(oracle)
result = algorithm.run(backend)
print(result.assignment)The code above demonstrates how Grover’s search algorithm can be used with the
LogicalExpressionOracle to find one satisfying assignment
for the Satisfiability (SAT) problem instance encoded in the
DIMACS CNF format.
The input string sat_cnf corresponds to the following Conjunctive Normal
Form (CNF):
(¬x1 ∨ ¬x2 ∨ ¬x3) ∧ (x1 ∨ ¬x2 ∨ x3) ∧ (x1 ∨ x2 ∨ ¬x3) ∧ (x1 ∨ ¬x2 ∨ ¬x3) ∧ (¬x1 ∨ x2 ∨ x3)
The Python code above prints out one possible solution for this CNF.
For example, output 1, -2, 3 indicates that logical expression
(x1 ∨ ¬x2 ∨ x3)
satisfies the given CNF.
Jupyter notebooks containing further examples, for Qiskit Aqua, may be found here in the following Qiskit GitHub repositories at qiskit-tutorials/legacy_tutorials/aqua and qiskit-community-tutorials/aqua.
The qiskit.chemistry package supports problems including ground state energy computations,
excited states and dipole moments of molecule, both open and closed-shell.
The code comprises chemistry drivers, which when provided with a molecular configuration will return one and two-body integrals as well as other data that is efficiently computed classically. This output data from a driver can then be used as input to the chemistry module that contains logic which is able to translate this into a form that is suitable for quantum algorithms. The conversion first creates a FermionicOperator which must then be mapped, e.g. by a Jordan Wigner mapping, to a qubit operator in readiness for the quantum computation.
To run chemistry experiments using Qiskit's chemistry module, it is recommended that you install a classical computation chemistry software program/library interfaced by Qiskit. Several, as listed below, are supported, and while logic to interface these programs is supplied by the chemistry module via the above pip installation, the dependent programs/libraries themselves need to be installed separately.
- Gaussian 16™, a commercial chemistry program
- PSI4, a chemistry program that exposes a Python interface allowing for accessing internal objects
- PySCF, an open-source Python chemistry program
- PyQuante, a pure cross-platform open-source Python chemistry program
A useful functionality integrated into Qiskit's chemistry module is its ability to serialize a file in hierarchical Data Format 5 (HDF5) format representing all the output data from a chemistry driver.
The HDF5 driver accepts such HDF5 files as input so molecular experiments can be run, albeit on the fixed data as stored in the file. As such, if you have some pre-created HDF5 files from created from Qiskit Chemistry, you can use these with the HDF5 driver even if you do not install one of the classical computation packages listed above.
A few sample HDF5 files for different are provided in the chemistry folder of the Qiskit Community Tutorials repository. This HDF5 Driver tutorial contains further information about creating and using such HDF5 files.
Now that Qiskit is installed, it's time to begin working with the chemistry module. Let's try a chemistry application experiment using VQE (Variational Quantum Eigensolver) algorithm to compute the ground-state (minimum) energy of a molecule.
from qiskit.chemistry import FermionicOperator
from qiskit.chemistry.drivers import PySCFDriver, UnitsType
from qiskit.aqua.operators import Z2Symmetries
# Use PySCF, a classical computational chemistry software
# package, to compute the one-body and two-body integrals in
# molecular-orbital basis, necessary to form the Fermionic operator
driver = PySCFDriver(atom='H .0 .0 .0; H .0 .0 0.735',
unit=UnitsType.ANGSTROM,
basis='sto3g')
molecule = driver.run()
num_particles = molecule.num_alpha + molecule.num_beta
num_spin_orbitals = molecule.num_orbitals * 2
# Build the qubit operator, which is the input to the VQE algorithm in Aqua
ferm_op = FermionicOperator(h1=molecule.one_body_integrals, h2=molecule.two_body_integrals)
map_type = 'PARITY'
qubit_op = ferm_op.mapping(map_type)
qubit_op = Z2Symmetries.two_qubit_reduction(qubit_op, num_particles)
num_qubits = qubit_op.num_qubits
# setup a classical optimizer for VQE
from qiskit.aqua.components.optimizers import L_BFGS_B
optimizer = L_BFGS_B()
# setup the initial state for the variational form
from qiskit.chemistry.components.initial_states import HartreeFock
init_state = HartreeFock(num_spin_orbitals, num_particles)
# setup the variational form for VQE
from qiskit.circuit.library import TwoLocal
var_form = TwoLocal(num_qubits, ['ry', 'rz'], 'cz', initial_state=init_state)
# setup and run VQE
from qiskit.aqua.algorithms import VQE
algorithm = VQE(qubit_op, var_form, optimizer)
# set the backend for the quantum computation
from qiskit import Aer
backend = Aer.get_backend('statevector_simulator')
result = algorithm.run(backend)
print(result.eigenvalue.real)The program above uses a quantum computer to calculate the ground state energy of molecular Hydrogen, H2, where the two atoms are configured to be at a distance of 0.735 angstroms. The molecular input specification is processed by PySCF driver and data is output that includes one- and two-body molecular-orbital integrals. From the output a fermionic-operator is created which is then parity mapped to generate a qubit operator. Parity mappings allow a precision-preserving optimization that two qubits can be tapered off; a reduction in complexity that is particularly advantageous for NISQ computers.
The qubit operator is then passed as an input to the Variational Quantum Eigensolver (VQE) algorithm, instantiated with a classical optimizer and a RyRz variational form (ansatz). A Hartree-Fock initial state is used as a starting point for the variational form.
The VQE algorithm is then run, in this case on the Qiskit Aer statevector simulator backend.
Here we pass a backend but it can be wrapped into a QuantumInstance, and that passed to the
run instead. The QuantumInstance API allows you to customize run-time properties of the backend,
such as the number of shots, the maximum number of credits to use, settings for the simulator,
initial layout of qubits in the mapping and the Terra PassManager that will handle the compilation
of the circuits. By passing in a backend as is done above it is internally wrapped into a
QuantumInstance and is a convenience when default setting suffice.
Jupyter notebooks containing further chemistry examples may be found in the following Qiskit GitHub repositories at qiskit-tutorials/legacy_tutorials/aqua/chemistry and qiskit-community-tutorials/chemistry.
The qiskit.finance package contains uncertainty components for stock/securities problems,
Ising translators for portfolio optimizations and data providers to source real or random data to
finance experiments.
Now that Qiskit is installed, it's time to begin working with the finance module. Let's try an experiment using Amplitude Estimation algorithm to evaluate a fixed income asset with uncertain interest rates.
import numpy as np
from qiskit import BasicAer
from qiskit.aqua.algorithms import AmplitudeEstimation
from qiskit.aqua.components.uncertainty_models import MultivariateNormalDistribution
from qiskit.finance.components.uncertainty_problems import FixedIncomeExpectedValue
# Create a suitable multivariate distribution
mvnd = MultivariateNormalDistribution(num_qubits=[2, 2],
low=[0, 0], high=[0.12, 0.24],
mu=[0.12, 0.24], sigma=0.01 * np.eye(2))
# Create fixed income component
fixed_income = FixedIncomeExpectedValue(mvnd, np.eye(2), np.zeros(2),
cash_flow=[1.0, 2.0], c_approx=0.125)
# Set number of evaluation qubits (samples)
num_eval_qubits = 5
# Construct and run amplitude estimation
algo = AmplitudeEstimation(num_eval_qubits, fixed_income)
result = algo.run(BasicAer.get_backend('statevector_simulator'))
print('Estimated value:\t%.4f' % result.estimation)
print('Probability: \t%.4f' % result.max_probability)When running the above the estimated value result should be 2.46 and probability 0.8487.
Jupyter notebooks containing further finance examples may be found in the following Qiskit GitHub repositories at qiskit-tutorials/legacy_tutorials/finance and qiskit-community-tutorials/finance.
The qiskit.ml package simply contains sample datasets at present. qiskit.aqua has some
classification algorithms such as QSVM and VQC (Variational Quantum Classifier), where this data
can be used for experiments, and there is also QGAN (Quantum Generative Adversarial Network)
algorithm.
Now that Qiskit is installed, it's time to begin working with Machine Learning. Let's try an experiment using VQC (Variational Quantum Classified) algorithm to train and test samples from a data set to see how accurately the test set can be classified.
from qiskit import BasicAer
from qiskit.aqua import QuantumInstance, aqua_globals
from qiskit.aqua.algorithms import VQC
from qiskit.aqua.components.optimizers import COBYLA
from qiskit.aqua.components.feature_maps import RawFeatureVector
from qiskit.ml.datasets import wine
from qiskit.circuit.library import TwoLocal
seed = 1376
aqua_globals.random_seed = seed
# Use Wine data set for training and test data
feature_dim = 4 # dimension of each data point
_, training_input, test_input, _ = wine(training_size=12,
test_size=4,
n=feature_dim)
feature_map = RawFeatureVector(feature_dimension=feature_dim)
vqc = VQC(COBYLA(maxiter=100),
feature_map,
TwoLocal(feature_map.num_qubits, ['ry', 'rz'], 'cz', reps=3),
training_input,
test_input)
result = vqc.run(QuantumInstance(BasicAer.get_backend('statevector_simulator'),
shots=1024, seed_simulator=seed, seed_transpiler=seed))
print('Testing accuracy: {:0.2f}'.format(result['testing_accuracy']))Jupyter notebooks containing further Machine Learning examples may be found in the following Qiskit GitHub repositories at qiskit-tutorials/legacy_tutorials/aqua/machine_learning and qiskit-tutorials/legacy_tutorials/aqua/finance/machine_learning and qiskit-community-tutorials/machine_learning.
The qiskit.optimization package covers the whole range from high-level modeling of optimization
problems, with automatic conversion of problems to different required representations, to a suite
of easy-to-use quantum optimization algorithms that are ready to run on classical simulators,
as well as on real quantum devices via Qiskit.
This optimization module enables easy, efficient modeling of optimization problems using docplex. A uniform interface as well as automatic conversion between different problem representations allows users to solve problems using a large set of algorithms, from variational quantum algorithms, such as the Quantum Approximate Optimization Algorithm QAOA, to Grover Adaptive Search using the GroverOptimizer leveraging fundamental algorithms provided by Aqua. Furthermore, the modular design of the optimization module allows it to be easily extended and facilitates rapid development and testing of new algorithms. Compatible classical optimizers are also provided for testing, validation, and benchmarking.
- IBM CPLEX may be installed using
pip install qiskit-aqua[cplex]to allow the use of theCplexOptimzerclassical solver algorithm, as well as enabling the reading ofLPfiles.
Now that Qiskit is installed, it's time to begin working with the optimization module. Let's try an optimization experiment to compute the solution of a Max-Cut. The Max-Cut problem can be formulated as quadratic program, which can be solved using many several different algorithms in Qiskit. In this example, the MinimumEigenOptimizer is employed in combination with the Quantum Approximate Optimization Algorithm (QAOA) as minimum eigensolver routine.
import networkx as nx
import numpy as np
from qiskit.optimization import QuadraticProgram
from qiskit.optimization.algorithms import MinimumEigenOptimizer
from qiskit import BasicAer
from qiskit.aqua.algorithms import QAOA
from qiskit.aqua.components.optimizers import SPSA
# Generate a graph of 4 nodes
n = 4
graph = nx.Graph()
graph.add_nodes_from(np.arange(0, n, 1))
elist = [(0, 1, 1.0), (0, 2, 1.0), (0, 3, 1.0), (1, 2, 1.0), (2, 3, 1.0)]
graph.add_weighted_edges_from(elist)
# Compute the weight matrix from the graph
w = nx.adjacency_matrix(graph)
# Formulate the problem as quadratic program
problem = QuadraticProgram()
_ = [problem.binary_var('x{}'.format(i)) for i in range(n)] # create n binary variables
linear = w.dot(np.ones(n))
quadratic = -w
problem.maximize(linear=linear, quadratic=quadratic)
# Fix node 0 to be 1 to break the symmetry of the max-cut solution
problem.linear_constraint([1, 0, 0, 0], '==', 1)
# Run quantum algorithm QAOA on qasm simulator
spsa = SPSA(max_trials=250)
backend = BasicAer.get_backend('qasm_simulator')
qaoa = QAOA(optimizer=spsa, p=5, quantum_instance=backend)
algorithm = MinimumEigenOptimizer(qaoa)
result = algorithm.solve(problem)
print(result) # prints solution, x=[1, 0, 1, 0], the cost, fval=4Learning path notebooks may be found in the optimization tutorials section of the documentation and are a great place to start.
Jupyter notebooks containing further examples, for the optimization module, may be found in the following Qiskit GitHub repositories at qiskit-tutorials/legacy_tutorials/aqua/optimization and qiskit-tutorials/legacy_tutorials/aqua/finance/optimization and qiskit-community-tutorials/optimization.
You can also use Qiskit to execute your code on a real quantum chip. In order to do so, you need to configure Qiskit to use the credentials in your IBM Quantum Experience account. For more detailed information refer to the relevant instructions in the Qiskit Terra GitHub repository .
If you'd like to contribute to Qiskit, please take a look at our contribution guidelines. This project adheres to Qiskit's code of conduct. By participating, you are expected to uphold this code.
We use GitHub issues for tracking requests and bugs. Please join the Qiskit Slack community and use the Aqua Slack channel for discussion and simple questions. For questions that are more suited for a forum, we use the Qiskit tag in Stack Overflow.
Now you're set up and ready to check out some of the other examples from the Qiskit Tutorials repository, that are used for the IBM Quantum Experience, and from the Qiskit Community Tutorials.
Aqua was inspired, authored and brought about by the collective work of a team of researchers. Aqua continues to grow with the help and work of many people, who contribute to the project at different levels. If you use Qiskit, please cite as per the provided BibTeX file.
Please note that if you do not like the way your name is cited in the BibTex file then consult the information found in the .mailmap file.
This project uses the Apache License 2.0.
However there is some code that is included under other licensing as follows:
- The Gaussian 16 driver in
qiskit.chemistrycontains work licensed under the Gaussian Open-Source Public License.