Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 331 of the 332 references it cites were found in a registry or an index; 179 of the 264 sources its claims cite were checked for support.

Variational quantum algorithms

Open access
Cerezo, M., Arrasmith, Andrew, Babbush, Ryan, Benjamin, Simon C., Endo, Suguru, Fujii, Keisuke, McClean, Jarrod R., Mitarai, Kosuke, Yuan, Xiao, Cincio, Lukasz, Coles, Patrick J. · 2021 · Nature Reviews Physics

Abstract

Applications such as simulating complicated quantum systems or solving large-scale linear algebra problems are very challenging for classical computers, owing to the extremely high computational cost. Quantum computers promise a solution, although fault-tolerant quantum computers will probably not be available in the near future. Current quantum devices have serious constraints, including limited numbers of qubits and noise processes that limit circuit depth. Variational quantum algorithms (VQAs), which use a classical optimizer to train a parameterized quantum circuit, have emerged as a leading strategy to address these constraints. VQAs have now been proposed for essentially all applications that researchers have envisaged for quantum computers, and they appear to be the best hope for obtaining quantum advantage. Nevertheless, challenges remain, including the trainability, accuracy and efficiency of VQAs. Here we overview the field of VQAs, discuss strategies to overcome their challenges and highlight the exciting prospects for using them to obtain quantum advantage. The advent of commercial quantum devices has ushered in the era of near-term quantum computing. Variational quantum algorithms are promising candidates to make use of these devices for achieving a practical quantum advantage over classical computers. Variational quantum algorithms (VQAs) are the leading proposal for achieving quantum advantage using near-term quantum computers. VQAs have been developed for a wide range of applications, including finding ground states of molecules, simulating dynamics of quantum systems and solving linear systems of equations. VQAs share a common structure, where a task is encoded into a parameterized cost function that is evaluated using a quantum computer, and a classical optimizer trains the parameters in the VQA. The adaptive nature of VQAs is well suited to handle the constraints of near-term quantum computers. Trainability, accuracy and efficiency are three challenges that arise when applying VQAs to large-scale applications, and strategies are currently being developed to address these challenges. Variational quantum algorithms (VQAs) are the leading proposal for achieving quantum advantage using near-term quantum computers. VQAs have been developed for a wide range of applications, including finding ground states of molecules, simulating dynamics of quantum systems and solving linear systems of equations. VQAs share a common structure, where a task is encoded into a parameterized cost function that is evaluated using a quantum computer, and a classical optimizer trains the parameters in the VQA. The adaptive nature of VQAs is well suited to handle the constraints of near-term quantum computers. Trainability, accuracy and efficiency are three challenges that arise when applying VQAs to large-scale applications, and strategies are currently being developed to address these challenges.

Overview

CoChat Summary

TL;DR

Variational quantum algorithms (VQAs) train parametrized quantum circuits with a classical optimizer and are presented as the leading strategy for obtaining quantum advantage on noisy intermediate-scale quantum (NISQ) devices. The review surveys VQA components, applications, challenges, and prospects, emphasizing that trainability, accuracy, and efficiency remain central obstacles.

Key findings

  • VQAs have been proposed for essentially all applications envisioned for quantum computers, including finding ground and excited states, dynamical quantum simulation, combinatorial optimization, linear systems, compiling, error correction, machine learning, and metrology.
  • Current state-of-the-art device size ranges from 50 to 100 qubits, and VQAs are argued to be the best hope for near-term quantum advantage because they keep circuit depth shallow and outsource parameter optimization to classical optimizers.
  • Trainability is threatened by barren plateaus, where cost-function partial derivatives vanish exponentially with system size on average; noise can also induce barren plateaus regardless of the ansatz.
  • Resource and scaling examples reported include: VQE requires \(O(1/\epsilon^2)\) measurements for precision \(\epsilon\), whereas quantum phase estimation requires \(O(\log(1/\epsilon))\); chemical Hamiltonians have Pauli-string counts scaling as \(n^4\), with commuting groupings reducing measurement scaling to \(n^3\); classical shadows scale logarithmically with \(n\); and a variational quantum linear solver study used \(n=10,\ldots,30\) qubits for Ising-inspired systems and \(n=2,\ldots,7\) qubits for random sparse linear systems.

Methods

  • Design: narrative review/overview of the VQA field, not a primary experimental or theoretical study.
  • Data or sample: no primary data or sample; the paper synthesizes prior literature and cites numerical and experimental results from other works.
  • Main analysis: conceptual description of VQA architecture (cost function, ansatz, gradients, optimizers), applications, challenges, and opportunities for near-term quantum advantage.

Limitations

  • The paper states that challenges remain including the trainability, accuracy, and efficiency of VQAs, and that barren plateaus could erase a potential quantum advantage.
  • The paper states that the degree to which error mitigation can correct noise effects is still an open question, and that measurement efficiency is not guaranteed even without barren plateaus.
  • Evident from the text: as a review, it presents no new experimental results, and many performance claims rely on cited heuristic or numerical studies rather than a systematic meta-analysis.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s42254-021-00348-9

Cites

331 of 332 references verified0 citation errors · 1 not checkable

Statistics

Nothing to recompute.

Figures

FIG. 1 Schematic diagram of a Variational Quantum Algorithm (VQA) . The inputs to a VQA are: a cost function C ( θ ) , with θ a set of parameters that encodes the solution to the problem, an ansatz whose parameters are trained to minimize the cost, and (possibly) a set of training data { ρ k } used during the optimization. Here, the cost can often be expressed in the form in Eq. (3), for some set of functions { f k } . Also, the ansatz is shown as a parameterized quantum circuit (on the left), which is analogous to a neural network (also shown schematically on the right). At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients). This information is fed into a classical computer that leverages the power of optimizers to navigate the cost landscape C ( θ ) and solve the optimization problem in Eq. (1). Once a termination condition is met, the VQA outputs an estimate of the solution to the problem. The form of the output depends on the precise task at hand. The red box indicates some of the most common types of outputs.
FIG. 1 Schematic diagram of a Variational Quantum Algorithm (VQA) . The inputs to a VQA are: a cost function C ( θ ) , with θ a set of parameters that encodes the solution to the problem, an ansatz whose parameters are trained to minimize the cost, and (possibly) a set of training data { ρ k } used during the optimization. Here, the cost can often be expressed in the form in Eq. (3), for some set of functions { f k } . Also, the ansatz is shown as a parameterized quantum circuit (on the left), which is analogous to a neural network (also shown schematically on the right). At each iteration of the loop one uses a quantum computer to efficiently estimate the cost (or its gradients). This information is fed into a classical computer that leverages the power of optimizers to navigate the cost landscape C ( θ ) and solve the optimization problem in Eq. (1). Once a termination condition is met, the VQA outputs an estimate of the solution to the problem. The form of the output depends on the precise task at hand. The red box indicates some of the most common types of outputs.From the PDF
FIG. 2 Schematic diagram of an ansatz . The unitary U ( θ ) , with θ a set of parameters, can be expressed as a product of L unitaries U l ( θ l ) sequentially acting on an input state. As indicated, each unitary U l ( θ l ) can in turn be decomposed into a sequence of parametrized and unparametrized gates.
FIG. 2 Schematic diagram of an ansatz . The unitary U ( θ ) , with θ a set of parameters, can be expressed as a product of L unitaries U l ( θ l ) sequentially acting on an input state. As indicated, each unitary U l ( θ l ) can in turn be decomposed into a sequence of parametrized and unparametrized gates.From the PDF
FIG. 3 Applications of Variational Quantum Algorithms (VQAs) . Many applications have been envisioned for VQAs. Here we show some of the key applications that are discussed in this Review.
FIG. 3 Applications of Variational Quantum Algorithms (VQAs) . Many applications have been envisioned for VQAs. Here we show some of the key applications that are discussed in this Review.From the PDF
FIG. 4 Variational Quantum Eigensolver (VQE) implementation . The VQE algorithm can be used to estimate the ground state energy E G of a molecule. The interactions of the system are encoded in a Hamiltonian H , usually expressed as a linear combination of simple operators h k with coefficients c k . Taking H as input, VQE outputs an estimate ˜ E G of the ground-state energy. The lower part of the figure shows the results of a VQE implementation for the electronic structure problem of an H 2 molecule, whose exact energy is shown as a dashed line. The experimental results were obtained using two of the five qubits in one of IBM's superconducting quantum processors (the inset illustrates qubit connectivity with Q 0 . . . Q 4 denoting the qubits ). Due to the presence of hardware noise the estimated energy ˜ E G has a gap with the true energy. In fact, amplifying the noise strength (that is increasing the quantity s ), deteriorates the solution quality. However, as discussed below, one can use error mitigation techniques to improve the solution quality. Figure adapted from Ref. [106], Springer Nature Limited.
FIG. 4 Variational Quantum Eigensolver (VQE) implementation . The VQE algorithm can be used to estimate the ground state energy E G of a molecule. The interactions of the system are encoded in a Hamiltonian H , usually expressed as a linear combination of simple operators h k with coefficients c k . Taking H as input, VQE outputs an estimate ˜ E G of the ground-state energy. The lower part of the figure shows the results of a VQE implementation for the electronic structure problem of an H 2 molecule, whose exact energy is shown as a dashed line. The experimental results were obtained using two of the five qubits in one of IBM's superconducting quantum processors (the inset illustrates qubit connectivity with Q 0 . . . Q 4 denoting the qubits ). Due to the presence of hardware noise the estimated energy ˜ E G has a gap with the true energy. In fact, amplifying the noise strength (that is increasing the quantity s ), deteriorates the solution quality. However, as discussed below, one can use error mitigation techniques to improve the solution quality. Figure adapted from Ref. [106], Springer Nature Limited.From the PDF
FIG. 5 Quantum Approximate Optimization Algorithm (QAOA) . a. Schematic representation of the Trotterized adiabatic transformation in the ansatz. The algorithm only loosely follows the evolution of the ground state of H ( t ) = (1 -t ) H M + tH P for every t ∈ [0 , 1] , as one is interested in making the final state close to the ground state of the problem Hamiltonian H P , with H M being a mixer Hamiltonian. The free parameters { β l } p l =1 and { γ l } p l =1 are trained, with p being the number of QAOA rounds. b. Problem Hamiltonian H P and graph 〈 jk 〉 for a Max-Cut task. Each node in the graph (circle) represents a spin. Vertices connecting two nodes indicate an interaction σ z j σ z k in H P , with σ z k the Pauli z operator on spin k . The solution is encoded in the ground state of H P where some spins are pointing up (green) whereas others point down (blue).
FIG. 5 Quantum Approximate Optimization Algorithm (QAOA) . a. Schematic representation of the Trotterized adiabatic transformation in the ansatz. The algorithm only loosely follows the evolution of the ground state of H ( t ) = (1 -t ) H M + tH P for every t ∈ [0 , 1] , as one is interested in making the final state close to the ground state of the problem Hamiltonian H P , with H M being a mixer Hamiltonian. The free parameters { β l } p l =1 and { γ l } p l =1 are trained, with p being the number of QAOA rounds. b. Problem Hamiltonian H P and graph 〈 jk 〉 for a Max-Cut task. Each node in the graph (circle) represents a spin. Vertices connecting two nodes indicate an interaction σ z j σ z k in H P , with σ z k the Pauli z operator on spin k . The solution is encoded in the ground state of H P where some spins are pointing up (green) whereas others point down (blue).From the PDF
FIG. 6 Barren plateau (BP) phenomenon. a. Variance of the cost function partial derivative, Var ( ∂θ 1 , 1 E ) , for a particular parameter θ 1 , 1 in the ansatz versus number of qubits ( n ). Results were obtained from a Variational Quantum Eigensolver implementation with a deep unstructured ansatz. The y -axis is on a log scale. As the number of qubits increases the variance vanish exponentially with the system size. b. Visualization of the landscape of a global cost function which exhibits a BP for the quantum compilation implementation, . The orange (blue) landscape was obtained for n = 24 ( n = 4 ) qubits. As the number of qubits increases, the landscape becomes flatter. Moreover, this cost also exhibits the narrow gorge phenomenon [166], where the volume of parameters leading to small cost values shrinks exponentially with n . Panel a is adapted from Ref. [194], CC BY 4.0; Panel b is adapted from Ref. [166], CC BY 4.0.
FIG. 6 Barren plateau (BP) phenomenon. a. Variance of the cost function partial derivative, Var ( ∂θ 1 , 1 E ) , for a particular parameter θ 1 , 1 in the ansatz versus number of qubits ( n ). Results were obtained from a Variational Quantum Eigensolver implementation with a deep unstructured ansatz. The y -axis is on a log scale. As the number of qubits increases the variance vanish exponentially with the system size. b. Visualization of the landscape of a global cost function which exhibits a BP for the quantum compilation implementation, . The orange (blue) landscape was obtained for n = 24 ( n = 4 ) qubits. As the number of qubits increases, the landscape becomes flatter. Moreover, this cost also exhibits the narrow gorge phenomenon [166], where the volume of parameters leading to small cost values shrinks exponentially with n . Panel a is adapted from Ref. [194], CC BY 4.0; Panel b is adapted from Ref. [166], CC BY 4.0.From the PDF
FIG. 7 Qubit trajectories on the Bloch sphere with the Zero-Noise Extrapolation (ZNE) technique. The accuracy of a noisy quantum computer can be improved with the ZNE error mitigation method. a. Here, one repeats a given calculation with different levels of noise. The green curve corresponds to a rotation on the Bloch sphere with a higher noise level than that leading to the red curve. b. Taking data from the red and green curves, ZNE can be used to estimate what the trajectory (blue) would be like in the absence of noise. Adapted from Ref. [106], Springer Nature Limited.
FIG. 7 Qubit trajectories on the Bloch sphere with the Zero-Noise Extrapolation (ZNE) technique. The accuracy of a noisy quantum computer can be improved with the ZNE error mitigation method. a. Here, one repeats a given calculation with different levels of noise. The green curve corresponds to a rotation on the Bloch sphere with a higher noise level than that leading to the red curve. b. Taking data from the red and green curves, ZNE can be used to estimate what the trajectory (blue) would be like in the absence of noise. Adapted from Ref. [106], Springer Nature Limited.From the PDF

Verification

Claims the paper makes

53 Valid · 75 Not valid · 50 Not checked · 84 Own results and interpretations

125 more claims · 57 with their cited sources checked

Create a free account to see all 262 claims and to verify the 13 references still open.

See all claims — free account
  • Probably not a substantial claim. 94
  • It may state two separate claims as one. 57

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

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

Each contradiction between two passages listed here was confirmed by an independent check.

2 of 2

Citations

Cites

Show all 200 references

Research works printed but not linked: 5

  1. IBM Makes Quantum Computing Available on IBM Cloud to Accelerate Innovation. Press release at https://www-03.ibm.com/press/us/en/pressrelease/49661.wss (2016).
  2. Rattew, A. G., Hu, S., Pistoia, M., Chen, R. & Wood, S. A domain-agnostic, noise-resistant, hardware-efficient evolutionary variational quantum eigensolver. Preprint at https://arxiv.org/abs/1910.09694 (2019). · arXiv
  3. Wang, G., Koh, D. E., Johnson, P. D. & Cao, Y. Minimizing estimation runtime on noisy quantum computers. Phys. Rev. X Quantum 2, 010346 (2021).
  4. Stoudenmire, E. & Schwab, D. J. Supervised learning with tensor networks. Adv. Neural Inf. Proc. Syst. 29, 4799–4807 (2016).
  5. Franken, L. & Georgiev, B. Explorations in quantum neural networks with intermediate measurements. In Proc. ESANN (2020); https://www.esann.org/sites/default/files/proceedings/2020/ES2020-197.pdf

Cited by

See all 3,385

Metadata

Authors
Cerezo, M., Arrasmith, A., Babbush, R., Benjamin, S. C., Endo, S., Fujii, K., McClean, J. R., Mitarai, K., Yuan, X., Cincio, L., & Coles, P. J.
Affiliations
  • Los Alamos National Laboratory (US)
  • Google (United States) (US)
  • University of Oxford (GB)
  • NTT (Japan) (JP)
  • RIKEN (JP)
  • Japan Science and Technology Agency (JP)
  • Peking University (CN)
Corresponding author
  • Cerezo, M.
  • Coles, Patrick J.
Venue
Nature Reviews Physics
Published
2021-08-12
Volume
3
Issue
9
Pages
625-644
ISSN
2522-5820 (ISSN-L)
OpenAlex
W3111162498
Cited by
3,385
References
200
Citations by year
2026: 874 · 2025: 896 · 2024: 699 · 2023: 521 · 2022: 317 · 2021: 78
FWCI
29.21
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Reviews Physics (publishedVersion)
  • arXiv (Cornell University) (publishedVersion)
  • OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion, cc-by)
  • arXiv (Cornell University)
Funders
  • U.S. Department of Energy (#89233218CNA000001, KA2401032, DE-AC52-06NA2-5396, AC52-06NA25396, 20190065DR)
  • Ministry of Education, Culture, Sports, Science and Technology (JPMXS0118068682, JPMXS0118067394, JP-MXS0120319794, CREST)
  • Office of Science (89233218CNA000001, KA2401032, DE-AC52-06NA25396)
  • National Nuclear Security Administration (DEAC52- 06NA25396, 89233218C-NA000001)
  • Horizon 2020 Framework Programme (820495)
  • Engineering and Physical Sciences Research Council (820495, M013243/1, EP/T001062/1, EP/M013243/1, T001062, H2020)
  • Japan Society for the Promotion of Science (JPMXS0118067394, 20K22330, JPMJER1601, JPMXS0120319794, CREST, 16H02211)
  • Advanced Scientific Computing Research (DE-AC52-06NA25396, 89233218CNA000001)
  • Laboratory Directed Research and Development (20200056DR, DE-AC52-06NA2-5396, 89233218CNA000001, 20190065DR, 20180628ECR)
  • Core Research for Evolutional Science and Technology (JPMJCR1673)
  • Precursory Research for Embryonic Science and Technology (JPMJPR2019)
  • Exploratory Research for Advanced Technology (JPMJER1601)
  • Los Alamos National Laboratory (20200056DR, 89233218CNA 000001, KA2401032, AC52-06NA25396, 20180628ECR, 20190065DR, DE-AC52-06NA25396)
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
variational quantum algorithms, variational quantum circuits, quantum advantage, near-term quantum computing, quantum chemistry simulation, quantum linear algebra, trainability, noisy quantum hardware, classical optimization

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.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

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: Variational quantum algorithms. https://app.cochat.ai/papers/10.1038/s42254-021-00348-9
The paper
Cerezo, M., Arrasmith, A., Babbush, R., Benjamin, S. C., Endo, S., Fujii, K., McClean, J. R., Mitarai, K., Yuan, X., Cincio, L., & Coles, P. J. (2021). Variational quantum algorithms. Nature Reviews Physics. https://doi.org/10.1038/s42254-021-00348-9

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.