Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Cambridge University Press eBooks

Multiplicative weights update method

Alexander M. Dalzell, Sam McArdle, Mario Berta, Przemyslaw Bienias, Chi-Fang Chen, András Gilyén, Connor T. Hann, Michael J. Kastoryano, Emil T. Khabiboulline, Aleksander Kubica, Grant Salton, Samson Wang, Fernando G. S. L. Brandão

Published Apr 24, 2025• 1 citation
Publisher

Abstract

This chapter covers the multiplicative weights update method, a quantum algorithmic primitive for certain continuous optimization problems. This method is a framework for classical algorithms, but it can be made quantum by incorporating the quantum algorithmic primitive of Gibbs sampling and amplitude amplification. The framework can be applied to solve linear programs and related convex problems, or generalized to handle matrix-valued weights and used to solve semidefinite programs.

DOI
10.1017/9781009639651.026

Related papers

Quantum and classical algorithms for SOCP based on the multiplicative weights update method

Garrido, M. Isabel Franco, Alexander M. Dalzell, Sam McArdle

We give classical and quantum algorithms for approximately solving second-order cone programs (SOCPs) based on the multiplicative weights (MW) update method. Our approach follows the MW framework previously applied to semidefinite programs (SDPs), of which SOCP is a special case. We show that the additional structure of SOCPs can be exploited to give better runtime with SOCP-specific algorithms. For an SOCP with $m$ linear constraints over $n$ variables partitioned into $r \leq n$ second-order cones, our quantum algorithm requires $\widetilde{O}(\sqrt{r}γ^5 + \sqrt{m}γ^4)$ (coherent) queries to the underlying data defining the instance, where $γ$ is a scale-invariant parameter proportional to the inverse precision. This nearly matches the complexity of solving linear programs (LPs), which are a less expressive subset of SOCP. It also outperforms (especially if $n \gg r$) the naive approach that applies existing SDP algorithms onto SOCPs, which has complexity $\widetilde{O}(γ^{4}(n + γ\sqrt{n} + \sqrt{m}))$. Our classical algorithm for SOCP has complexity $\widetilde{O}(nγ^4 + m γ^6)$ in the sample-and-query model.

arXiv:2507.14127•Jul 18, 2025•1 citation

Continuous optimization

Alexander M. Dalzell, Sam McArdle, Mario Berta +10 more

This chapter covers applications of quantum computing in the area of continuous optimization, including both convex and nonconvex optimization. We discuss quantum algorithms for computing Nash equilibria for zero-sum games and for solving linear, second-order, and semidefinite programs. These algorithms are based on quantum implementations of the multiplicative weights update method or interior point methods. We also discuss general quantum algorithms for convex optimization which can provide a speedup in cases where the objective function is much easier to evaluate than the gradient of the objective function. Finally, we cover quantum algorithms for escaping saddle points and finding local minima in nonconvex optimization problems.

Cambridge University Press eBooks•Apr 24, 2025•0 citations

Quantum Speed-Ups for Solving Semidefinite Programs

Fernando G. S. L. Brandão, Krysta M. Svore

We give a quantum algorithm for solving semidefinite programs (SDPs). It has worst-case running time n1/2m1/2s2poly(log(n), log(m), R, r, 1/δ), with n and s the dimension and row-sparsity of the input matrices, respectively, m the number of constraints, δ the accuracy of the solution, and R, r upper bounds on the size of the optimal primal and dual solutions, respectively. This gives a square-root unconditional speed-up over any classical method for solving SDPs both in n and m. We prove the algorithm cannot be substantially improved (in terms of n and m) giving a Ω(n1/2+ m2) quantum lower bound for solving semidefinite programs with constant s, R, r and δ. The quantum algorithm is constructed by a combination of quantum Gibbs sampling and the multiplicative weight method. In particular it is based on a classical algorithm of Arora and Kale for approximately solving SDPs. We present a modification of their algorithm to eliminate the need for solving an inner linear program which may be of independent interest.

Annual Symposium on Foundations of Computer Science•Oct 1, 2017•159 citations

Quantum Speed-ups for Semidefinite Programming

Fernando G. S. L. Brandão, Krysta M. Svore

We give a quantum algorithm for solving semidefinite programs (SDPs). It has worst-case running time $n^{\frac{1}{2}} m^{\frac{1}{2}} s^2 \text{poly}(\log(n), \log(m), R, r, 1/δ)$, with $n$ and $s$ the dimension and row-sparsity of the input matrices, respectively, $m$ the number of constraints, $δ$ the accuracy of the solution, and $R, r$ a upper bounds on the size of the optimal primal and dual solutions. This gives a square-root unconditional speed-up over any classical method for solving SDPs both in $n$ and $m$. We prove the algorithm cannot be substantially improved (in terms of $n$ and $m$) giving a $Ω(n^{\frac{1}{2}}+m^{\frac{1}{2}})$ quantum lower bound for solving semidefinite programs with constant $s, R, r$ and $δ$. The quantum algorithm is constructed by a combination of quantum Gibbs sampling and the multiplicative weight method. In particular it is based on a classical algorithm of Arora and Kale for approximately solving SDPs. We present a modification of their algorithm to eliminate the need for solving an inner linear program which may be of independent interest.

arXiv:1609.05537•Sep 18, 2016•94 citations

Amplitude amplification and estimation

Alexander M. Dalzell, Sam McArdle, Mario Berta +10 more

This chapter covers the quantum algorithmic primitives of amplitude amplification and amplitude estimation. Amplitude amplification is a generalization of Grover’s quantum algorithm for the unstructured search problem. Amplitude estimation can be understood in a similar framework, where it utilizes quantum phase estimation to estimate the value of the amplitude or probability associated with a quantum state. Both amplitude amplification and amplitude estimation provide a quadratic speedup over their classical counterparts, and feature prominently as an ingredient in many end-to-end algorithms.

Cambridge University Press eBooks•Apr 24, 2025•0 citations

Quantum interior point methods

Alexander M. Dalzell, Sam McArdle, Mario Berta +10 more

This chapter covers quantum interior point methods, which are quantum algorithmic primitives for application to convex optimization problems, particularly linear, second-order, and semidefinite programs. Interior point methods are a successful classical iterative technique that solve a linear system of equations at each iteration. Quantum interior point methods replace this step with quantum a quantum linear system solver combined with quantum tomography, potentially offering a polynomial speedup.

Cambridge University Press eBooks•Apr 24, 2025•0 citations
This chapter covers the multiplicative weights update method, a quantum algorithmic primitive for certain continuous optimization problems. This method is a framework for classical algorithms, but it can be made quantum by incorporating the quantum algorithmic primitive of Gibbs sampling and amplitude amplification. The framework can be applied to solve linear programs and related convex problems, or generalized to handle matrix-valued weights and used to solve semidefinite programs.
We give classical and quantum algorithms for approximately solving second-order cone programs (SOCPs) based on the multiplicative weights (MW) update method. Our approach follows the MW framework previously applied to semidefinite programs (SDPs), of which SOCP is a special case. We show that the additional structure of SOCPs can be exploited to give better runtime with SOCP-specific algorithms. For an SOCP with mmm linear constraints over nnn variables partitioned into r≤nr \leq nr≤n second-order cones, our quantum algorithm requires O~(rγ5+mγ4)\widetilde{O}(\sqrt{r}γ^5 + \sqrt{m}γ^4)O(r​γ5+m​γ4) (coherent) queries to the underlying data defining the instance, where γγγ is a scale-invariant parameter proportional to the inverse precision. This nearly matches the complexity of solving linear programs (LPs), which are a less expressive subset of SOCP. It also outperforms (especially if n≫rn \gg rn≫r) the naive approach that applies existing SDP algorithms onto SOCPs, which has complexity O~(γ4(n+γn+m))\widetilde{O}(γ^{4}(n + γ\sqrt{n} + \sqrt{m}))O(γ4(n+γn​+m​)). Our classical algorithm for SOCP has complexity O~(nγ4+mγ6)\widetilde{O}(nγ^4 + m γ^6)O(nγ4+mγ6) in the sample-and-query model.
This chapter covers applications of quantum computing in the area of continuous optimization, including both convex and nonconvex optimization. We discuss quantum algorithms for computing Nash equilibria for zero-sum games and for solving linear, second-order, and semidefinite programs. These algorithms are based on quantum implementations of the multiplicative weights update method or interior point methods. We also discuss general quantum algorithms for convex optimization which can provide a speedup in cases where the objective function is much easier to evaluate than the gradient of the objective function. Finally, we cover quantum algorithms for escaping saddle points and finding local minima in nonconvex optimization problems.
We give a quantum algorithm for solving semidefinite programs (SDPs). It has worst-case running time n1/2m1/2s2poly(log(n), log(m), R, r, 1/δ), with n and s the dimension and row-sparsity of the input matrices, respectively, m the number of constraints, δ the accuracy of the solution, and R, r upper bounds on the size of the optimal primal and dual solutions, respectively. This gives a square-root unconditional speed-up over any classical method for solving SDPs both in n and m. We prove the algorithm cannot be substantially improved (in terms of n and m) giving a Ω(n1/2+ m2) quantum lower bound for solving semidefinite programs with constant s, R, r and δ. The quantum algorithm is constructed by a combination of quantum Gibbs sampling and the multiplicative weight method. In particular it is based on a classical algorithm of Arora and Kale for approximately solving SDPs. We present a modification of their algorithm to eliminate the need for solving an inner linear program which may be of independent interest.
We give a quantum algorithm for solving semidefinite programs (SDPs). It has worst-case running time n12m12s2poly(log⁡(n),log⁡(m),R,r,1/δ)n^{\frac{1}{2}} m^{\frac{1}{2}} s^2 \text{poly}(\log(n), \log(m), R, r, 1/δ)n21​m21​s2poly(log(n),log(m),R,r,1/δ), with nnn and sss the dimension and row-sparsity of the input matrices, respectively, mmm the number of constraints, δδδ the accuracy of the solution, and R,rR, rR,r a upper bounds on the size of the optimal primal and dual solutions. This gives a square-root unconditional speed-up over any classical method for solving SDPs both in nnn and mmm. We prove the algorithm cannot be substantially improved (in terms of nnn and mmm) giving a Ω(n12+m12)Ω(n^{\frac{1}{2}}+m^{\frac{1}{2}})Ω(n21​+m21​) quantum lower bound for solving semidefinite programs with constant s,R,rs, R, rs,R,r and δδδ. The quantum algorithm is constructed by a combination of quantum Gibbs sampling and the multiplicative weight method. In particular it is based on a classical algorithm of Arora and Kale for approximately solving SDPs. We present a modification of their algorithm to eliminate the need for solving an inner linear program which may be of independent interest.
This chapter covers the quantum algorithmic primitives of amplitude amplification and amplitude estimation. Amplitude amplification is a generalization of Grover’s quantum algorithm for the unstructured search problem. Amplitude estimation can be understood in a similar framework, where it utilizes quantum phase estimation to estimate the value of the amplitude or probability associated with a quantum state. Both amplitude amplification and amplitude estimation provide a quadratic speedup over their classical counterparts, and feature prominently as an ingredient in many end-to-end algorithms.
This chapter covers quantum interior point methods, which are quantum algorithmic primitives for application to convex optimization problems, particularly linear, second-order, and semidefinite programs. Interior point methods are a successful classical iterative technique that solve a linear system of equations at each iteration. Quantum interior point methods replace this step with quantum a quantum linear system solver combined with quantum tomography, potentially offering a polynomial speedup.