Refining qubit calibration models using supervised learning
A computer-implemented method for refining a qubit calibration model is described. The method comprises receiving, at a learning module, training data, wherein the training data comprises a plurality of calibration data sets, wherein each calibration data set is derived from a system comprising one or more qubits, and a plurality of parameter sets, each parameter set comprising extracted parameters obtained using a corresponding calibration data set, wherein extracting the parameters includes fitting a qubit calibration model to the corresponding calibration data set using a fitter algorithm. The method further comprises executing, at the learning module, a supervised machine learning algorithm which processes the training data to learn a perturbation to the qubit calibration model that captures one or more features in the plurality of calibration data sets that are not captured by the qubit calibration model, thereby to provide a refined qubit calibration model.
1. A computer-implemented method for refining a qubit calibration model, comprising:
receiving, at a learning module, training data, wherein the training data comprises:
a plurality of calibration data sets, wherein each calibration data set is derived from a system comprising one or more qubits;
a plurality of parameter sets, each parameter set comprising extracted parameters obtained using a corresponding calibration data set, wherein extracting the parameters includes fitting a model to the corresponding calibration data set using a fitter algorithm;
executing, at the learning module, a supervised machine learning algorithm which processes the training data to learn a perturbation to a qubit calibration model that captures one or more features in the plurality of calibration data sets that are not captured by the qubit calibration model, thereby to provide a refined qubit calibration model.
2. The computer-implemented method according to claim 1 , wherein the refined qubit calibration model is given by f (x, p, p′)=f(x, p)+g(x, p′), wherein f(x, p) is the model and g(x, p′) is a perturbation function, wherein x represents one or more hardware parameters, p represents one or more model parameters, and p′ represents parameters of the perturbation function.
3. The computer-implemented method according to claim 2 , wherein a basis expansion method is used to determine the perturbation.
4. The computer-implemented method according to claim 1 , wherein the supervised machine learning algorithm comprises a linear or nonlinear regression algorithm based on linear or nonlinear least squares regression, or a neural network.
5. The computer-implemented method according to claim 1 , wherein the supervised machine learning algorithm comprises a machine learning classifier.
6. The computer-implemented method according to claim 5 , wherein each parameter set of the training data defines at least one decision boundary defining a boundary between one or more groups of calibration data points.
7. The computer-implemented method according to claim 5 , wherein the machine learning classifier comprises a linear or nonlinear classifier based on least squares, linear or quadratic discriminant analysis, logistic regression, support vector machine or a neural network.
8. The computer-implemented method according to claim 1 , wherein the learning module employs one or more of shrinkage, regularization, restriction, or selection methods.
9. The computer-implemented method according to claim 1 , comprising comparing the refined qubit calibration model to the qubit calibration model based on at least one of: test and/or generalization error, and model complexity.
10. The computer-implemented method according to claim 1 , wherein the qubit is a system based on at least one of: a superconducting qubit, a system based on one or more ion traps, quantum dots, neutral atoms, Rydberg states, solid-state defects, molecules, or photons.
11. The computer-implemented method according to claim 1 , wherein the refined qubit calibration model permits extraction of one or more qubit parameters which could not be extracted using the qubit calibration model.
12. An apparatus configured to carry out the method according to claim 1 .
13. A computer-readable medium having computer-readable code stored thereon, the computer-readable code, when executed by the processor, causing the method of according to claim 1 to be performed.
14. An automated qubit calibration method, comprising:
performing calibrations on one or more qubits to obtain calibration data sets and corresponding extracted parameter sets, wherein performing each calibration comprises:
obtaining a calibration data set from a system comprising one or more qubits;
determining one or more extracted parameters, comprising fitting one or more parameters of a model to the calibration data set, using a fitter algorithm;
storing the one or more extracted parameters, and
refining one or more qubit calibration models used for one or more respective calibrations in accordance with the method according to claim 1 .
15. The method according to claim 14 , further comprising:
performing a plurality of calibrations using a plurality of respective refined qubit calibration models; and
further refining one or more of said refined qubit calibration models in accordance with the method of according to claim 1 .
16. The method according to claim 14 , comprising selecting a qubit and performing a sequence of calibrations on the selected qubit, wherein one or more of said calibrations are performed dependent on whether one or more earlier calibrations are performed successfully, comprising determining whether a calibration data set should be included in the training data dependent on whether a later calibration in the sequence is performed.
17. The method according to claim 14 , comprising determining whether a calibration data set should be used as training data using a computer-implemented classification algorithm.
18. The method according to claim 14 , wherein the training data further comprises an indicator as to whether the parameter set correctly describes or is capable of correctly describing the behaviour of a qubit.
19. The method according to claim 18 , wherein the indicator is determined based on whether one or more parameters in the parameter set fall within one or more hardware defined limits.
20. A computer-implemented method according to claim 1 , further comprising performing a calibration using a refined model given by f (x, p, p′)=f(x, p)+g(x, p′), wherein f(x, p) is the model and g(x, p′) is a perturbation function, wherein x represents one or more hardware parameters, p represents a set of model parameters, and p′ represents parameters of the perturbation, wherein performing the calibration using the refined model comprises:
obtaining a calibration data set from a system comprising a qubit;
fitting one or more of the parameters p, p′ of the refined model to the calibration data set, using a fitter algorithm to obtain one or more fitted parameters p fit , p′ fit corresponding to the calibration data set;
obtaining an extracted parameter set corresponding to the calibration data set using the fitted parameters p fit ; and
storing at least a part of the extracted parameter set.