CoChat

Efficient quantum state tomography with convolutional neural networks

Open access
Schmale, Tobias, Reh, Moritz, Gärttner, Martin · 2022 · npj Quantum Information

The DOI matches its Crossref record; 50 of the 51 references it cites were found in a registry or an index.

Abstract

Modern day quantum simulators can prepare a wide variety of quantum states but the accurate estimation of observables from tomographic measurement data often poses a challenge. We tackle this problem by developing a quantum state tomography scheme which relies on approximating the probability distribution over the outcomes of an informationally complete measurement in a variational manifold represented by a convolutional neural network. We show an excellent representability of prototypical ground- and steady states with this ansatz using a number of variational parameters that scales polynomially in system size. This compressed representation allows us to reconstruct states with high classical fidelities outperforming standard methods such as maximum likelihood estimation. Furthermore, it achieves a reduction of the estimation error of observables by up to an order of magnitude compared to their direct estimation from experimental data.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41534-022-00621-4

Cites

50 of 50 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 CNN architectures and their expressivity. a , b Schematics of the CNN architectures. Entries of cells should be understood as a dependency, not as literal values, e.g. bottom right cell of standard CNN depends on a 2 , a 3 and a 4. Lighter cells are padding cells. The green arrows indicate the dot-product of the originating cells with a kernel. See the methods section for details. a Standard CNN, here with circular boundary (CCNN), b Autoregressive CNN (ARCNN). c , d h σ z i σ z j i correlator as a function of distance d = ∣ i -j ∣ , for varying network depth L and kernel size K , shown for 1D transverse- fi eld Ising model (TFIM) ground states ( N = 16, J / B = 1). Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.
Fig. 1 CNN architectures and their expressivity. a , b Schematics of the CNN architectures. Entries of cells should be understood as a dependency, not as literal values, e.g. bottom right cell of standard CNN depends on a 2 , a 3 and a 4. Lighter cells are padding cells. The green arrows indicate the dot-product of the originating cells with a kernel. See the methods section for details. a Standard CNN, here with circular boundary (CCNN), b Autoregressive CNN (ARCNN). c , d h σ z i σ z j i correlator as a function of distance d = ∣ i -j ∣ , for varying network depth L and kernel size K , shown for 1D transverse- fi eld Ising model (TFIM) ground states ( N = 16, J / B = 1). Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.CC-BY · From the PDF
Image 2
CC-BY · From the PDF
Fig. 2 Tomography of 1D Ising groundstates: NN vs MLE. a Residual classical in fi delity of CCNN and MLE on 1D Ising ground state with periodic boundary conditions, J / B = 1. Shown for different dataset sizes Ns . b Residual network in fi delity normalized to residual MLE in fi delity. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.
Fig. 2 Tomography of 1D Ising groundstates: NN vs MLE. a Residual classical in fi delity of CCNN and MLE on 1D Ising ground state with periodic boundary conditions, J / B = 1. Shown for different dataset sizes Ns . b Residual network in fi delity normalized to residual MLE in fi delity. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.CC-BY · From the PDF
Fig. 3 Tomography of 2D Ising groundstates: NN vs plain data. Comparing the performance of the CCNN against the plain dataset in estimating local observables on a 4 × 4 Ising lattice with periodic boundaries. a Histogram of observable estimates from independent datasets and network initializations, with J / B = 0.3 and Ns = 10 k . b RMS errors of observables for varying coupling strengths and dataset sizes, as well as comparisons of NN and plain dataset in terms of the quotient of their respective RMS errors. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters. 2
Fig. 3 Tomography of 2D Ising groundstates: NN vs plain data. Comparing the performance of the CCNN against the plain dataset in estimating local observables on a 4 × 4 Ising lattice with periodic boundaries. a Histogram of observable estimates from independent datasets and network initializations, with J / B = 0.3 and Ns = 10 k . b RMS errors of observables for varying coupling strengths and dataset sizes, as well as comparisons of NN and plain dataset in terms of the quotient of their respective RMS errors. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters. 2CC-BY · From the PDF
Fig. 4 Tomography of noisy long-range interacting groundstates: NN vs plain data. Evaluating the ARCNN by evaluating observables from Eq. (9) and comparing to those obtained from the plain dataset, on a length 16 ion chain with 3% dephasing noise, J / B = 0.6 and open boundaries, with a dataset size of Ns = 10 k. Dashed lines show results of applying local MLE directly to dataset (orange) and to a neural network enhanced dataset (blue). a Observables according to Eq. (9), b residual RMS error, c residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.
Fig. 4 Tomography of noisy long-range interacting groundstates: NN vs plain data. Evaluating the ARCNN by evaluating observables from Eq. (9) and comparing to those obtained from the plain dataset, on a length 16 ion chain with 3% dephasing noise, J / B = 0.6 and open boundaries, with a dataset size of Ns = 10 k. Dashed lines show results of applying local MLE directly to dataset (orange) and to a neural network enhanced dataset (blue). a Observables according to Eq. (9), b residual RMS error, c residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.CC-BY · From the PDF
Fig. 5 Tomography of steady states: NN vs plain data. Dissipative phase transition of a 4 × 4 TFIM with spontaneous emission at J = -1.25 γ with open boundaries. Ns = 1 k (insets: 100 k). Gray vertical line: Switch ↑ and ↓ in POVM for CNN. a Observables according to Eq. (12) but summed only over system diagonal, b residual RMS error, c residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters. Fi5
Fig. 5 Tomography of steady states: NN vs plain data. Dissipative phase transition of a 4 × 4 TFIM with spontaneous emission at J = -1.25 γ with open boundaries. Ns = 1 k (insets: 100 k). Gray vertical line: Switch ↑ and ↓ in POVM for CNN. a Observables according to Eq. (12) but summed only over system diagonal, b residual RMS error, c residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters. Fi5CC-BY · From the PDF
Fig. 6 Tomography of steady states: demonstration of a selfaveraging effect. Dissipative phase transition of a 4 x 4 TFIM with spontaneous emission at J = -1.25 γ with open boundaries. Ns = 1 k (insets: 100 k ). Gray vertical line: Point where ↑ and ↓ have been exchanged in POVM. a Observables according to Eq. (12) summed over entire system. b Residual RMS error. c Residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.
Fig. 6 Tomography of steady states: demonstration of a selfaveraging effect. Dissipative phase transition of a 4 x 4 TFIM with spontaneous emission at J = -1.25 γ with open boundaries. Ns = 1 k (insets: 100 k ). Gray vertical line: Point where ↑ and ↓ have been exchanged in POVM. a Observables according to Eq. (12) summed over entire system. b Residual RMS error. c Residual network RMS normalized to dataset RMS. Error bars are standard deviations across multiple independent tomography datasets and neural network training runs with randomly initialized parameters.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 50 references

Research works printed but not linked: 4

  1. Deng, D.-L., Li, X. & Das Sarma, S. Quantum entanglement in neural network states. Phys. Rev. X 7, 021021 (2017).
  2. Huang, H. & Situ, H. Investigating reconstruction of quantum state distributions with neural networks. Eur. Phys. J. 136, 204 (2021).
  3. Lohani, S., Kirby, B. T., Brodsky, M., Danaci, O. & Glasser, R. T. Machine learning assisted quantum state estimation. Mach. Learn.: Sci. Technol. 1, 035007 (2020).
  4. Kingma, D. P. & Ba, J. Adam: A method for stochastic optimization. arXiv:1412.6980. http://arxiv.org/abs/1412.6980 (2017). · arXiv

Cited by

See all 113

Metadata

Authors
Schmale, T., Reh, M., & Gärttner, M.
Affiliations
Heidelberg University (DE)
Corresponding author
  • Schmale, Tobias
  • Reh, Moritz
  • Gärttner, Martin
Venue
npj Quantum Information
Published
2022-09-23
Volume
8
Issue
1
ISSN
2056-6387 (ISSN-L)
OpenAlex
W4296836560
Cited by
76
References
51
Citations by year
2026: 13 · 2025: 31 · 2024: 20 · 2023: 11 · 2022: 1
FWCI
5.18
Type
journal article
Language
en
Open access
Licence
Locations
  • npj Quantum Information (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
Funders
  • Deutsche Forschungsgemeinschaft (EXC2181/1-390900948, SFB 1225, INST 40/575-1 FUGG)
  • Gauss Centre for Supercomputing e.V.
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum many-body systems
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum Information and Cryptography
Keywords
quantum state tomography, convolutional neural networks, variational ansatz, informationally complete measurements, density estimation, quantum simulators, ground state, steady state, maximum likelihood estimation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Efficient quantum state tomography with convolutional neural networks. https://app.cochat.ai/papers/10.1038/s41534-022-00621-4
The paper
Schmale, T., Reh, M., & Gärttner, M. (2022). Efficient quantum state tomography with convolutional neural networks. npj Quantum Information. https://doi.org/10.1038/s41534-022-00621-4

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.