--- title: LinearSystemData (latest version) description: API reference for qiskit_noise_learning.analysis.LinearSystemData in the latest version of qiskit-noise-learning source: https://quantum.cloud.ibm.com/docs/en/api/qiskit-noise-learning/generated/analysis-linear-system-data --- # qiskit\_noise\_learning.analysis.LinearSystemData *class* `qiskit_noise_learning.analysis.LinearSystemData(A: ndarray, b: ndarray, sigma_b: ndarray, row_diagnostics: Mapping[str, ndarray], row_index_map: Mapping[RowIndex, int], column_index_map: Mapping[ColumnIndex, int], time_lb: datetime64, time_ub: datetime64)` [GitHub](https://github.com/Qiskit/qiskit-noise-learning/tree/stable/0.1/qiskit_noise_learning/analysis/model_solve.py) Bases: [`Generic`](https://docs.python.org/3/library/typing.html#typing.Generic)\[`RowIndex`, `ColumnIndex`] The linear system to solve and metadata in raw format. A linear system `A @ x = b` with axis labels and metadata. The row and column labels are of arbitrary hashable types: this class carries no assumptions about what a row or column denotes. In the systems built by [`from_fit()`](#qiskit_noise_learning.analysis.LinearSystemData.from_fit "qiskit_noise_learning.analysis.LinearSystemData.from_fit") the rows are [`Path`](/docs/api/qiskit-noise-learning/generated/sequences-path "qiskit_noise_learning.sequences.Path") objects and the columns are the fidelity model’s parameter labels. The index maps are the authoritative record of how labels correspond to positions in `A`. Anything that needs to align a label-keyed quantity with the arrays should index through `row_index_map` or `column_index_map` rather than rebuilding the correspondence from [`row_labels`](#qiskit_noise_learning.analysis.LinearSystemData.row_labels "qiskit_noise_learning.analysis.LinearSystemData.row_labels") or [`column_labels`](#qiskit_noise_learning.analysis.LinearSystemData.column_labels "qiskit_noise_learning.analysis.LinearSystemData.column_labels"). **Parameters** - **A** – The matrix with shape `(m, n)`. - **b** – The target vector length `m`. - **sigma\_b** – Statistical `1`-sigma uncertainty on `b` per row, with length `m`. - **row\_diagnostics** – Named per-row quantities recorded by the stages that produced `b`, each an array of length `m` holding `nan` for rows the quantity is undefined for. Keys are the metadata names used upstream; see [`from_fit()`](#qiskit_noise_learning.analysis.LinearSystemData.from_fit "qiskit_noise_learning.analysis.LinearSystemData.from_fit"). - **row\_index\_map** – A mapping from row labels to their integer row position in `A`. - **column\_index\_map** – A mapping from column labels to their integer column position in `A`. - **time\_lb** – Earliest time bound across the rows. - **time\_ub** – Latest time bound across the rows. ### \_\_init\_\_ `__init__(A: ndarray, b: ndarray, sigma_b: ndarray, row_diagnostics: Mapping[str, ndarray], row_index_map: Mapping[RowIndex, int], column_index_map: Mapping[ColumnIndex, int], time_lb: datetime64, time_ub: datetime64) → None` ## Methods | | | | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------- | | [`__init__`](#qiskit_noise_learning.analysis.LinearSystemData.__init__ "qiskit_noise_learning.analysis.LinearSystemData.__init__")(A, b, sigma\_b, row\_diagnostics, ...) | | | [`from_fit`](#qiskit_noise_learning.analysis.LinearSystemData.from_fit "qiskit_noise_learning.analysis.LinearSystemData.from_fit")(fit) | Build the linear system arrays from a [`Fit`](/docs/api/qiskit-noise-learning/generated/analysis-fit "qiskit_noise_learning.analysis.Fit"). | ## Attributes | | | | ------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------- | | [`column_labels`](#qiskit_noise_learning.analysis.LinearSystemData.column_labels "qiskit_noise_learning.analysis.LinearSystemData.column_labels") | Column labels, ordered by their column position in `A`. | | [`row_labels`](#qiskit_noise_learning.analysis.LinearSystemData.row_labels "qiskit_noise_learning.analysis.LinearSystemData.row_labels") | Row labels, ordered by their row position in `A`. | | `A` | | | `b` | | | `sigma_b` | | | `row_diagnostics` | | | `row_index_map` | | | `column_index_map` | | | `time_lb` | | | `time_ub` | | ### row\_labels Type: [`list`](https://docs.python.org/3/library/stdtypes.html#list)\[`RowIndex`] Row labels, ordered by their row position in `A`. ### column\_labels Type: [`list`](https://docs.python.org/3/library/stdtypes.html#list)\[`ColumnIndex`] Column labels, ordered by their column position in `A`. ### from\_fit *classmethod* `from_fit(fit: Fit) → LinearSystemData[Path, Hashable]` Build the linear system arrays from a [`Fit`](/docs/api/qiskit-noise-learning/generated/analysis-fit "qiskit_noise_learning.analysis.Fit"). Rows are the [`Path`](/docs/api/qiskit-noise-learning/generated/sequences-path "qiskit_noise_learning.sequences.Path") objects of the [`AggregatedObservableData`](/docs/api/qiskit-noise-learning/generated/data-aggregated-observable-data "qiskit_noise_learning.data.AggregatedObservableData"), columns are the fidelity model’s parameter labels, and `row_diagnostics` holds every real-valued per-observable metadata entry under the name the producing stage used — for instance `"reduced_chi_squared"` from [`CurveFitObservables`](/docs/api/qiskit-noise-learning/generated/analysis-curve-fit-observables "qiskit_noise_learning.analysis.CurveFitObservables"). **Warns** **UserWarning** – If any row’s uncertainty is non-positive or non-finite, giving those rows’ positions in [`row_labels`](#qiskit_noise_learning.analysis.LinearSystemData.row_labels "qiskit_noise_learning.analysis.LinearSystemData.row_labels"). Such a row carries no usable statistical weight, and how it is treated is up to the solver.