FiQCIEstimator

FiQCIEstimator#

FiQCIEstimator computes expectation values of Pauli observables from quantum circuits with built-in error mitigation. It supports both readout error mitigation (M3) and zero-noise extrapolation (ZNE).

Basic Configuration#

Initialize the estimator with an IQM backend, mitigation level, and optional parameters:

from fiqci.ems import FiQCIEstimator

# Initialize estimator with mitigation level 1
estimator = FiQCIEstimator(backend, mitigation_level=1, calibration_shots=2000, calibration_file="cals.json")

For more details see the API reference documentation for FiQCIEstimator.

Mitigation Levels#

Mitigation levels apply predefined sets of error mitigation techniques.

Level	Mitigation Applied	Technique
0	None	Raw results
1	Readout Error Mitigation	M3 (matrix-free measurement mitigation)
2	Level 1 + additional	TBD
3	Level 2 + Zero Noise Extrapolation	Exponential Extrapolation, Local Folding

Mitigation Options#

Mitigators can also be configured manually using the provided methods.

Readout Error Mitigation (REM)
Zero Noise Extrapolation (ZNE)

REM (Readout Error Mitigation)#

Readout error mitigation uses M3 (matrix-free measurement mitigation) to correct measurement errors. It is enabled by default at mitigation level 1.

Configure REM using the rem() method:

estimator.rem(enabled=True, calibration_shots=2000, calibration_file="cals.json")

Parameter	Default	Description
`enabled`	`True`	Enable or disable readout error mitigation
`calibration_shots`	`1000`	Number of shots used for M3 calibration circuits
`calibration_file`	`None`	Path to save/load calibration data (JSON). Reuses cached calibrations when available.

ZNE (Zero-Noise Extrapolation)#

ZNE artificially scales circuit noise by folding gates, then extrapolates to the zero-noise limit. It is enabled at mitigation level 3.

Configure ZNE using the zne() method:

estimator.zne(
    enabled=True,
    fold_gates=["cx", "cz"],
    scale_factors=[1, 3, 5],
    folding_method="local",
    extrapolation_method="exponential",
)

Parameter	Default	Description
`enabled`	`True`	Enable or disable ZNE
`fold_gates`	`None`	Gate names to fold (e.g. `["cx", "cz"]`). If `None`, folds all gates.
`scale_factors`	`[1, 3, 5]`	List of positive odd integers for noise scaling. At least 2 required.
`folding_method`	`"local"`	`"local"` (per-gate folding) or `"global"` (whole-circuit folding). When `"global"`, `fold_gates` is ignored.
`extrapolation_method`	`"exponential"`	Extrapolation fit method. One of: `"exponential"`, `"richardson"`, `"polynomial"`, `"linear"`.
`extrapolation_degree`	`None`	Polynomial degree (only for `"polynomial"` method). Defaults to `min(n_scales - 1, 2)`.

Inspecting Options#

Use the mitigator_options property to view currently applied mitigation settings:

estimator.mitigator_options

Results#

run() returns a FiQCIEstimatorJobCollection with the following methods:

Method	Description
`expectation_values(index=None)`	Mitigated expectation values
`raw_expectation_values(index=None)`	Raw (pre-extrapolation) expectation values
`jobs()`	All jobs executed by the estimator
`results()`	Results for each job
`observables(index=None)`	Observables used in the computation