Decoders

class panqec.decoders.BaseDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float)

Base class for decoders

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float)

abstract property allowed_codes: List[str] | None

List of codes concerned by the decoders (by id). Takes the value None if all codes are allowed.

Example: [‘Toric2DCode’, ‘Planar2DCode’, ‘RotatedPlanar2DCode’]

abstract decode(syndrome: ndarray, **kwargs) → ndarray

Given a code and a syndrome, returns a correction to apply to the qubits

Parameters:

• syndrome (np.ndarray) – Syndrome as an array of size m, where m is the number of stabilizers. Each element contains 1 if the stabilizer is activated and 0 otherwise

• kwargs (dict) – Decoder-specific parameters (implemented by subclasses)

Returns:

correction – Correction as an array of size 2n (with n the number of qubits) in the binary symplectic format.

Return type:

np.ndarray

property id: str

Class name

abstract property label: str

Label used in plots and result files E.g. ‘Toric 2D Matching’

abstract property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

class panqec.decoders.BeliefPropagationOSDDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_bp_iter: int = 1000, channel_update: bool = False, osd_order: int = 10, bp_method: str = 'msl')

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_bp_iter: int = 1000, channel_update: bool = False, osd_order: int = 10, bp_method: str = 'msl')

allowed_codes = None

decode(syndrome: ndarray, **kwargs) → ndarray

Get X and Z corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

update_probabilities(correction: ndarray, px: ndarray, py: ndarray, pz: ndarray, direction: str = 'x->z') → ndarray

Update X probabilities once a Z correction has been applied

class panqec.decoders.MatchingDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, error_type: str | None = None, weights: Tuple[ndarray, ndarray] | None = None)

Matching decoder for 2D Toric Code, based on PyMatching

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, error_type: str | None = None, weights: Tuple[ndarray, ndarray] | None = None)

Constructor for the MatchingDecoder class

Parameters:

• code (StabilizerCode) – Code used by the decoder

• error_model (BaseErrorModel) – Error model used by the decoder (to find the weights)

• error_rate (int, optional) – Error rate used by the decoder (to find the weights)

• error_type (str, optional) – Determines which type of errors (X or Z) to decode. Can take the values “X”, “Z”, or None if we want to decode all errors

• weights (Tuple[np.ndarray, np.ndarray], optional) – Matching weights for X and Z errors

decode(syndrome: ndarray, **kwargs) → ndarray

Get X corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

class panqec.decoders.MemoryBeliefPropagationDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_bp_iter: int = 100, alpha: float = 0.4, beta: float = 0)

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_bp_iter: int = 100, alpha: float = 0.4, beta: float = 0)

allowed_codes = None

decode(syndrome: ndarray, **kwargs) → ndarray

Get X and Z corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

class panqec.decoders.RotatedSweepDecoder3D(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, seed: int = 0, max_rounds: int = 32)

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, seed: int = 0, max_rounds: int = 32)

decode(syndrome: ndarray, **kwargs) → ndarray

Get Z corrections given measured syndrome.

flip_edge(edge: Tuple, signs: ndarray)

Flip signs at index and update correction.

get_default_direction()

The default direction when all faces are excited.

get_face_syndromes(full_syndrome: ndarray) → ndarray

Get only the syndromes for the vertex Z stabilizers. Z vertex stabilizers syndromes are discarded for this decoder.

get_initial_state(syndrome: ndarray) → ndarray

Get initial cellular automaton state from syndrome.

get_sweep_edges(vertex, sweep_direction)

Get coordinates of neighbouring edges in sweep direction.

get_sweep_faces(vertex, sweep_direction)

Get the coordinates of neighboring faces in sweep direction.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

sweep_move(signs: ndarray, correction: Dict[Tuple, str], sweep_direction: Tuple[int, int, int]) → ndarray

Apply the sweep move once along a particular direciton.

class panqec.decoders.RotatedSweepMatchDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_rounds=32)

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, max_rounds=32)

decode(syndrome: ndarray, **kwargs) → ndarray

Get X and Z corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

class panqec.decoders.SweepDecoder3D(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, seed: int = 0, max_sweep_factor: int = 32)

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float, seed: int = 0, max_sweep_factor: int = 32)

decode(syndrome: ndarray, **kwargs) → ndarray

Get Z corrections given measured syndrome.

flip_edge(location: Tuple, signs: ndarray)

Flip signs at index and update correction.

get_default_direction()

The default direction when all faces are excited.

get_face_syndromes(full_syndrome: ndarray) → ndarray

Get only the syndromes for the vertex Z stabilizers. Z vertex stabilizers syndromes are discarded for this decoder.

get_initial_state(syndrome: ndarray) → ndarray

Get initial cellular automaton state from syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

sweep_move(signs: ndarray, correction: Dict[Tuple, str]) → ndarray

Apply the sweep move once.

class panqec.decoders.SweepMatchDecoder(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float)

__init__(code: StabilizerCode, error_model: BaseErrorModel, error_rate: float)

decode(syndrome: ndarray, **kwargs) → ndarray

Get X and Z corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}

class panqec.decoders.XCubeMatchingDecoder(code: StabilizerCode, error_model: PauliErrorModel, error_rate: float)

Matching decoder for 2D Toric Code, based on PyMatching

__init__(code: StabilizerCode, error_model: PauliErrorModel, error_rate: float)

Constructor for the MatchingDecoder class

Parameters:

• code (StabilizerCode) – Code used by the decoder

• error_model (PauliErrorModel) – Error model used by the decoder (to find the weights)

• error_rate (int, optional) – Error rate used by the decoder (to find the weights)

• error_type (str, optional) – Determines which type of errors (X or Z) to decode. Can take the values “X”, “Z”, or None if we want to decode all errors

decode(syndrome: ndarray, **kwargs) → ndarray

Get X corrections given code and measured syndrome.

property params: dict

List of class arguments (as a dictionary), that can be saved and reused to instantiate the same decoder. It should not include code, error_model and error_rate

Example: {‘num_iterations’: 10}