Convergence guarantee for linearly-constrained combinatorial optimization with a quantum alternating operator ansatz

Brayden Goldstein-Gelb,Phillip C. Lotshaw

2024-09-27

Abstract:We present a quantum alternating operator ansatz (QAOA$^+$) that solves a class of linearly constrained optimization problems by evolving a quantum state within a Hilbert subspace of feasible problem solutions. Our main focus is on a class of problems with a linear constraint containing sequential integer coefficients. For problems in this class, we devise QAOA$^+$ circuits that provably converge to the optimal solution as the number of circuit layers increases, generalizing previous guarantees for solving unconstrained problems or problems with symmetric constraints. Our approach includes asymmetric ``mixing" Hamiltonians that drive transitions between feasible states, as well as a method to incorporate an arbitrary known feasible solution as the initial state, each of which can be applied beyond the specific linear constraints considered here. This analysis extends QAOA$^+$ performance guarantees to a more general set of linearly-constrained problems and provides tools for future generalizations.

Quantum Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to develop a Quantum Alternating Operator Ansatz Variational Algorithm (QAOA+), in order to solve a class of combinatorial optimization problems with linear constraints. Specifically, the paper focuses on linear - constraint problems with sequential integer coefficients, and designs QAOA+ circuits that can ensure convergence to the optimal solution as the number of circuit layers increases. This extends the previous performance guarantees for unconstrained or symmetric - constraint problems, providing a theoretical basis for a wider type of linear - constraint problems.

### Background and Problem Description of the Paper

#### 1. Combinatorial Optimization Problems
Combinatorial optimization problems aim to minimize an objective function \(C(z)\), where \(z=(z_1, z_2,\ldots, z_N)\) is a string consisting of \(N\) bits, and each \(z_i\in\{0, 1\}\). The objective function \(C(z)\) can be expressed in quadratic form:
\[C(z)=\sum_{i < j}J_{ij}z_iz_j+\sum_ih_iz_i\]
where \(J_{ij}\) and \(h_i\) are real - number coefficients, which define the specific instances of the problem. The solution \(z\) also needs to satisfy certain constraint conditions, which limit the set of potential solutions from all possible \(2^N\) bit strings to a smaller feasible subset \(S\).

#### 2. Linear - Constraint Problems
This paper specifically focuses on optimization problems with linear constraints. This type of problem can be expressed as:
\[z_{\text{opt}}=\arg\min_zC(z)\]
\[\text{s.t.}\quad\sum_is_{\alpha,i}z_i = b_\alpha\quad\forall\alpha\in\{1, 2,\ldots, t\}\]
\[z_i\in\{0, 1\}\quad\forall i\]
where the elements \(s_{\alpha,i}\in\mathbb{N}\) of matrix \(S\) and the components \(b_\alpha\in\mathbb{N}\) of vector \(b\) encode \(t\) total constraints.

#### 3. QAOA+ Algorithm
QAOA+ is a quantum algorithm that directly optimizes the constrained objective function \(C(z)\) by evolving the quantum state within the Hilbert subspace of feasible solutions. To achieve this, QAOA+ uses a suitable mixing Hamiltonian \(B^+\), which maintains the feasibility of the initial feasible state. The evolution process of QAOA+ is defined as:
\[|\psi_p^+(\beta,\gamma)\rangle=\left(\prod_{l = 1}^pU_{B^+}(\beta_l)U_C(\gamma_l)\right)|\psi_0^+\rangle\]
where \(U_{B^+}(\beta_l)=e^{-i\beta_lB^+}\) and \(U_C(\gamma_l)=e^{-i\gamma_lC}\).

### Main Contributions

#### 1. Construction of the Mixing Family
The paper proposes a new class of mixing operators \(M^{(m)}_{I, i^*}\) for driving the transition between different feasible solutions. These operators are in the form of:
\[M^{(m)}_{I, i^*}=|0_{i_1}, 0_{i_2},\ldots, 0_{i_m}, 1_{i^*}\rangle\langle 1_{i_1}, 1_{i_2},\ldots, 1_{i_m}, 0_{i^*}|+|1_{i_1}, 1_{i_2},\ldots, 1_{i_m}, 0_{i^*}\rangle\langle 0_{i_1}, 0_{i_2}\]