## Highlights

- •Quantum noise leaves a fingerprint that can be identified.
- •We develop Machine Learning models to classify different quantum noise fingerprints.
- •We adopt Support Vector Machines as Machine Learning method.
- •We propose a way to discern if quantum noise is machine dependent.
- •We can evaluate if quantum noise exhibits a dependency on the execution time.

## Abstract

## Keywords

Current code version | v1 |

Permanent link to code/repository used for this code version | https://github.com/SoftwareImpacts/SIMPAC-2022-9 |

Permanent link to Reproducible Capsule | https://codeocean.com/capsule/8363708/tree/v1 |

Legal Code License | GNU General Public License v3.0 |

Code versioning system used | GIT |

Software code languages, tools, and services used | python, IBM quantum services |

Compilation requirements, operating environments & dependencies | numpy, qiskit, qiskit_terra, scikit_learn |

If available Link to developer documentation/manual | |

Support email for questions | [email protected] |

## 1. Introduction

*quantum supremacy*[

*Noisy Intermediate-Scale Quantum*(NISQ) technology has been recently introduced [

2022, https://quantum-computing.ibm.com/. Visited on 2022.

2022, https://www.rigetti.com/. Visited on 2022.

*testbed quantum circuit*– composed by a fixed number of qubits – is designed, then made run for a sufficient number of times and finally locally measured in correspondence of each qubit. From the measurements of the qubits (the measurement observable was the Pauli matrix ${\sigma}_{z}$), a set of measurement outcomes is recorded, collected, and then used to train a

*machine learning*(ML) algorithm [

*support vector machine*(SVM) [

*quantum data*) from the testbed quantum circuit, and subsequently train ML (classical) algorithms. In fact, no quantum noise modelling is required nor, in principle, the testbed circuit has to be controlled by time-dependent pulses [

## 2. Testbed quantum circuit

*simulate*quantum dynamics and to

*program*a given set of operations on a real quantum computer. Currently, one has at disposal up to $11$ superconducting quantum computers, ranging from a single qubit up to $15$ qubits, with different topology and calibration routines. For all the available devices and their specifications, we direct the reader to the IBM documentation [

2022, https://quantum-computing.ibm.com/. Visited on 2022.

### 2.1 Data acquisition

## 3. Machine learning models

### 3.1 Support vector machine

*scikit-learn*python library [

### 3.2 Data interpretation

*(i)*identifies what model has to be used, and

*(ii)*set optional arguments to tune the number of hyperparameters (mask) and to control if the method is verbose (verbose) and if the results have to be written in an output file (writeToFile). Practically, the function runSVM first calls extractData whose purpose is to load the dataset file, extract the data in the desired time steps and split them in

*training*,

*validation*and

*test*sets. After that, runSVM proceeds to train a set of possible SVM models on the training set, by then evaluating them on the validation set and computing on the test set the resulting accuracy of the model that performed better on the validation set. The possible models that can be employed are:

*(i)*Standard linear SVM (using two different libraries),

*(ii)*SVM with

*polynomial*kernel with degree from 2 to 4, and

*(iii)*SVM with RBF kernel.

## 4. Impacts

*(i)*distinguish the noise fingerprints in different quantum devices;

*(ii)*classify the noise fingerprint on the same quantum devices but in different times;

*(iii)*learn if and how a given noise fingerprint changes over time.

*accurate*(more than 99% of effectiveness) in classifying a clear machine-related noise fingerprint in each of the analysed IBM quantum computers, and even

*robust*since any noise fingerprint is highly predictable over time in windows of consecutive runs. Also an evident time-dependence of the noise fingerprints has been classified, by observing changes over time after few hours from the first execution of the testbed quantum circuit.

*inaccessible*quantum machines.

### 4.1 Applications

*to identify*from which specific quantum device certain data (a collection of measurement outcomes) are generated, just looking at the noise fingerprint of the device. Moreover, the proposed solution might be employed

*to certify*the time-scheduling in which a given quantum computation is executed. Both these applications are expected to play a key role for diagnostics purposes – especially in all those contexts where quantum computations cannot be error-corrected [

### 4.2 Outlook

## Declaration of Competing Interest

## Acknowledgements

## References

- Quantum technology: the second quantum revolution.
*Philos. Trans. R. Soc. A.*2003; 361: 1655-1674 - Photonic quantum technologies.
*Nat. Photon.*2009; 3: 687-695 - Integrated photonic quantum technologies.
*Nat. Photon.*2020; 14: 273-284 - Rydberg atom quantum technologies.
*J. Phys. B At. Mol. Opt.*2020; 53012002 - Superconducting qubits: A short review.2004 (arXiv preprint Cond-Mat/0411174)
- Superconducting quantum bits.
*Nature.*2008; 453: 1031-1042 - Strong quantum computational advantage using a superconducting quantum processor.
*Phys. Rev. Lett.*2021; 127180501 - Quantum information processing with trapped ions.
*Phil. Trans. R. Soc. A.*2003; 361: 1349-1361 - Compact ion-trap quantum computing demonstrator.
*PRX Quantum.*2021; 2020343 - Boson sampling on a photonic chip.
*Science.*2013; 339: 798-801 - Quantum teleportation on a photonic chip.
*Nat. Photonics.*2014; 8: 770-774 - Topological quantum computation.
*Bull. Amer. Math. Soc.*2003; 40: 31-38 - Quantum supremacy using a programmable superconducting processor.
*Nature.*2019; 574: 505-510 - Quantum computing in the NISQ era and beyond.
*Quantum.*2018; 2: 79 2022, https://quantum-computing.ibm.com/. Visited on 2022.

2022, https://www.rigetti.com/. Visited on 2022.

- Learning the noise fingerprint of quantum devices.2021 (arXiv preprint arXiv:2109.11405)
- Pattern Recognition and Machine Learning.first ed. Springer-Verlag, New York2006
- The Elements of Statistical Learning: Data Mining, Inference, and Prediction.second ed. Springer series in statistics, New York2009
- Information theoretical limits for quantum optimal control solutions: Error scaling of noisy channels.2020 (arXiv preprint arXiv:2006.16113)
- Beyond quantum noise spectroscopy: modelling and mitigating noise with quantum feature engineering.
*Npj Quantum Inf.*2020; 6: 95 - Machine learning non-Markovian quantum dynamics.
*Phys. Rev. Lett.*2020; 124140502 - Estimating the degree of non-Markovianity using machine learning.
*Phys. Rev. A.*2021; 103022425 - Learning non-Markovian quantum noise from moiré-enhanced swap spectroscopy with deep evolutionary algorithm.2019 (arXiv preprint arXiv:1912.04368)
- Efficient learning of quantum noise.
*Nat. Phys.*2020; : 1-5 - Machine learning approach for quantum non-Markovian noise classification.2021 (arXiv preprint arXiv:2101.03221)
- Using deep learning to understand and mitigate the qubit noise environment.
*PRX Quantum.*2021; 2010316 - Qiskit: An open-source framework for quantum computing.2019
- Quantum Computation and Quantum Information.American Association of Physics Teachers, 2002
- Validating quantum computers using randomized model circuits.
*Phys. Rev. A.*2019; 100032328 - Scikit-learn: Machine learning in python.
*J. Mach. Learn. Res.*2011; 12: 2825-2830 - Harnessing the power of the second quantum revolution.
*PRX Quantum.*2020; 1020101 - Benchmarking an 11-qubit quantum computer.
*Nature Commun.*2019; 10: 5464 - Quantum certification and benchmarking.
*Nat. Rev. Phys.*2020; 2: 382-390 - Deep Learning.MIT press Cambridge, 2016 (http://www.deeplearningbook.org)
- Deep learning in neural networks: An overview.
*Neural Netw.*2015; 61: 85-117

## Article info

### Publication history

### Footnotes

The code (and data) in this article has been certified as Reproducible by Code Ocean: (https://codeocean.com/). More information on the Reproducibility Badge Initiative is available at https://www.elsevier.com/physical-sciences-and-engineering/computer-science/journals.

### Identification

### Copyright

### User license

Creative Commons Attribution (CC BY 4.0) |## Permitted

- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes

Elsevier's open access license policy