cirq.GreedyQubitManager

Greedy allocator that maximizes/minimizes qubit reuse based on a configurable parameter.

Inherits From: QubitManager

cirq.GreedyQubitManager(
    prefix: str, *, size: int = 0, maximize_reuse: bool = False
)

GreedyQubitManager can be configured, using maximize_reuse flag, to work in one of two modes:

  • Minimize qubit reuse (maximize_reuse=False): For a fixed width, this mode uses a FIFO (First in First out) strategy s.t. next allocated qubit is one which was freed the earliest.
  • Maximize qubit reuse (maximize_reuse=True): For a fixed width, this mode uses a LIFO (Last in First out) strategy s.t. the next allocated qubit is one which was freed the latest.

If the requested qubits are more than the set of free qubits, the qubit manager automatically resizes the size of the managed qubit pool and adds new free qubits, that have their last freed time to be -infinity.

For borrowing qubits, the qubit manager simply delegates borrow requests to self.qalloc, thus always allocating new clean qubits.

prefix The prefix to use for naming new clean ancillas allocated by the qubit manager. The i'th allocated qubit is of the type cirq.NamedQubit(f'{prefix}_{i}').
size The initial size of the pool of ancilla qubits managed by the qubit manager. The qubit manager can automatically resize itself when the allocation request exceeds the number of available qubits.
maximize_reuse Flag to control a FIFO vs LIFO strategy, defaults to False (FIFO).

Methods

qalloc

View source

qalloc(
    n: int, dim: int = 2
) -> List['cirq.Qid']

Allocate n clean qubits, i.e. qubits guaranteed to be in state |0>.

qborrow

View source

qborrow(
    n: int, dim: int = 2
) -> List['cirq.Qid']

Allocate n dirty qubits, i.e. the returned qubits can be in any state.

qfree

View source

qfree(
    qubits: Iterable['cirq.Qid']
) -> None

Free pre-allocated clean or dirty qubits managed by this qubit manager.

resize

View source

resize(
    new_size: int, dim: int = 2
) -> None