VQE cost functions

We provide two different cost functions for VQE. One for PyQuil’s WavefunctionSimulator, and one for Rigetti’s QVM that can also be run on the QPU.

On the Wavefunction Simulator

class vqe.cost_function.PrepareAndMeasureOnWFSim(prepare_ansatz, make_memory_map, hamiltonian, sim=None, scalar_cost_function=True, nshots=0, enable_logging=False, qubit_mapping=None, hamiltonian_is_diagonal=False)

A cost function that prepares an ansatz and measures its energy w.r.t hamiltonian on the qvm

Parameters

  • prepare_ansatz (param) – A parametric pyquil program for the state preparation

  • make_memory_map (param) – A function that creates a memory map from the array of parameters

  • hamiltonian (PauliSum) – The hamiltonian with respect to which to measure the energy.

  • sim (Optional[WavefunctionSimulator]) – A WavefunctionSimulator instance to get the wavefunction from. Automatically created, if None is passed.

  • scalar_cost_function (bool) – If True: __call__ returns only the expectation value If False: __call__ returns a tuple (exp_val, std_dev) Defaults to True.

  • nshots (int) – Number of shots to assume to simulate the sampling noise. 0 corresponds to no sampling noise added and is the default.

  • enable_logging (bool) – If true, a log is created which contains the parameter and function values at each function call. It is a list of namedtuples of the form (“x”, “fun”)

  • qubit_mapping (Optional[Dict[QubitPlaceholder, Union[Qubit, int]]]) – A mapping to fix QubitPlaceholders to physical qubits. E.g. pyquil.quil.get_default_qubit_mapping(program) gives you on.

  • hamiltonian_is_diagonal (bool) – Pass true, if the hamiltonian contains only Z terms and products thereof. This speeds up __init__ considerably. Defaults to False.

__call__(params, nshots=None)

Cost function that computes <psi|ham|psi> with psi prepared with prepare_ansatz(params).

Parameters

  • params (Union[list, ndarray]) – Parameters of the state preparation circuit.

  • nshots (Optional[int]) – Overrides nshots from __init__ if passed

Returns

Either only the cost or a tuple of the cost and the standard deviation estimate based on the samples.

Return type

float or tuple (cost, cost_stdev)

get_wavefunction(params)

Same as __call__ but returns the wavefunction instead of cost

Parameters

params (Union[List[~T], ndarray]) – Parameters of the state preparation circuit

Returns

The wavefunction prepared with parameters params

Return type

Wavefunction

On the QVM / QPU

class vqe.cost_function.PrepareAndMeasureOnQVM(prepare_ansatz, make_memory_map, hamiltonian, qvm, scalar_cost_function=True, nshots=1, base_numshots=100, qubit_mapping=None, enable_logging=False, hamiltonian_is_diagonal=False)

A cost function that prepares an ansatz and measures its energy w.r.t hamiltonian on a quantum computer (or simulator).

This cost_function makes use of pyquils parametric circuits and thus has to be supplied with a parametric circuit and a function to create memory maps that can be passed to qvm.run.

Parameters

  • prepare_ansatz (Program) – A parametric pyquil program for the state preparation

  • make_memory_map (Callable[[Iterable[+T_co]], dict]) – A function that creates a memory map from the array of parameters

  • hamiltonian (PauliSum) – The hamiltonian

  • qvm (Union[QuantumComputer, str]) – Connection the QC to run the program on OR a name string like expected by pyquil.api.get_qc

  • scalar_cost_function (bool) – If True: __call__ returns only the expectation value If False: __call__ returns a tuple (exp_val, std_dev) Defaults to True.

  • nshots (int) – Fixed multiple of base_numshots for each estimation of the expectation value. Defaults to 1

  • base_numshots (int) –

    numshots multiplier to compile into the binary. The argument nshots of

    __call__ is then a multplier of this.

  • qubit_mapping (Optional[Dict[QubitPlaceholder, Union[Qubit, int]]]) – A mapping to fix all QubitPlaceholders to physical qubits. E.g. pyquil.quil.get_default_qubit_mapping(program) gives you on.

  • enable_logging (bool) – If true, a log is created which contains the parameter and function values at each function call. It is a list of namedtuples of the form (“x”, “fun”)

__call__(params, nshots=None)
Parameters

  • params (param) – the parameters to run the state preparation circuit with

  • nshots (param) – Overrides nshots in __init__ if passed

Returns

Either only the cost or a tuple of the cost and the standard deviation estimate based on the samples.

Return type

float or tuple (cost, cost_stdev)