cuquantum.densitymat.OperatorSpectrumSolver

class cuquantum.densitymat.OperatorSpectrumSolver(operator: Operator, which: str, hermitian: bool, config: Optional[Union[OperatorSpectrumConfig, dict]] = None)

Eigenvalue solver for computing the spectrum (eigenvalues and eigenstates) of quantum operators.

This class provides an interface for finding eigenvalues and eigenstates of Operator objects using iterative methods.

Parameters

• operator – The quantum operator whose eigenvalues and eigenstates are to be computed. Must be an instance of Operator.

• which – Specifies which eigenvalues to compute. Accepted values are: - "SA": Smallest Algebraic - eigenvalues with smallest real parts, ordered ascending - "LA": Largest Algebraic - eigenvalues with largest real parts, ordered descending

• hermitian – Whether the operator is Hermitian. This affects the choice of eigenvalue algorithm and can improve performance for Hermitian operators. Currently, only Hermitian operators are supported.

• config – Specify options for the eigensolver as a OperatorSpectrumConfig object. Alternatively, a dict containing the parameters for the OperatorSpectrumConfig constructor can also be provided. If not specified, the defaults documented in OperatorSpectrumConfig are used.

Methods

__init__(operator: Operator, which: str, hermitian: bool, config: Optional[Union[OperatorSpectrumConfig, dict]] = None) → None

compute(t: float, params: Optional[Union[numpy.ndarray, cupy.ndarray, Sequence[float]]], states: Sequence[DensePureState], tol: Optional[Union[float, numpy.ndarray]] = None) → OperatorSpectrumResult

Compute the spectrum of the provided Operator using block Krylov iteration. This function is blocking, i.e. it will only return after the computation is complete.

Parameters

• t – Time argument to be passed to all callback functions.

• params – Additional arguments to be passed to all callback functions. The element type is required to be float (i.e "float64" for arrays). If batched operators or coefficients are used, they need to be passed as 2-dimensional the last dimension of which is the batch size. To avoid copy of the argument array, it can be passed as a fortran-contiguous GPU array.

• states – Block of linearly-independent initial states for Krylov iteration, which will be in-place updated with the final eigenstates. The number of requested eigenstates is equal to the number of provided initial states.

• tol – Tolerance for residuals ||A*x - lambda*x||. Can be either scalar or array of shape [num_eigvals] or [num_eigvals, batch_size]. If None, square root of machine precision is used.

Returns

Dataclass holding the requested eigenvalues and eigenstates of the Operator.

Return type

OperatorSpectrumResult

configure(config: Union[OperatorSpectrumConfig, dict]) → None

Configure the eigensolver.

Parameters

config – Configuration for the eigensolver. Can be instance of OperatorSpectrumConfig or dict.

prepare(ctx: WorkStream, repr_state: DensePureState, max_num_eigvals: int = 1) → None

Prepare the eigensolver for computation.

Parameters

• ctx – WorkStream to use for computation. This includes library handle, workspace buffer allocation as well stream on which the computation will be performed.

• repr_state – Representative state to use for computation.

• max_num_eigvals – Maximum number of eigenvalues to compute.

Attributes

config

Configuration options for this solver. Direct modification of the return argument by the user leads to undefined behaviour. Call OperatorSpectrumSolver.configure() to modify the configuration.

is_hermitian

Whether the operator is Hermitian.

which

Which eigenvalues this solver computes.