Abstract:This paper consider solving a class of nonconvex-strongly-convex distributed stochastic bilevel optimization (DSBO) problems with personalized inner-level objectives. Most existing algorithms require computational loops for hypergradient estimation, leading to computational inefficiency. Moreover, the impact of data heterogeneity on convergence in bilevel problems is not explicitly characterized yet. To address these issues, we propose LoPA, a loopless personalized distributed algorithm that leverages a tracking mechanism for iterative approximation of inner-level solutions and Hessian-inverse matrices without relying on extra computation loops. Our theoretical analysis explicitly characterizes the heterogeneity across nodes denoted by $b$, and establishes a sublinear rate of ${\mathcal{O}}( {\frac{1}{(1-\rho)^2K} + \frac{{b^{\frac{2}{3}} }}{{\left( {1 - \rho } \right)^{\frac{2}{3}}K^{\frac{2}{3}} }} + \frac{1}{\sqrt{ K }} ( \sigma_{\rm{p}} + \frac{1}{\sqrt{m}}\sigma_{\rm{c}}) } )$ without the boundedness of local hypergradients, where $\sigma_{\rm p}$ and $\sigma_{\rm c}$ represent the gradient sampling variances associated with the inner- and outer-level variables, respectively. We also develop a variant of LoPA based on gradient tracking to eliminate the impact of data heterogeneity, yielding an improved rate of ${\mathcal{O}}(\frac{1}{ (1-\rho)^4K } + \frac{1}{\sqrt{K}}( \sigma_{\rm{p}} + \frac{1}{\sqrt{m}}\sigma_{\rm{c}} ) )$. The computational complexity of LoPA is of ${\mathcal{O}}({\epsilon^{-2}})$ to an $\epsilon$-stationary point, matching the communication complexity due to the loopless structure, which outperforms existing counterparts for DSBO. Numerical experiments validate the effectiveness of the proposed algorithm.

Partially distributed outer approximation

Distributed Primal Outer Approximation Algorithm for Sparse Convex Programming with Separable Structures

An outer approximation algorithm for multi-objective mixed-integer linear and non-linear programming

The decomposition-based outer approximation algorithm for convex mixed-integer nonlinear programming

Resilient Approximation-Based Distributed Nonconvex Optimization

Distributed Primal-Dual Proximal Algorithms for Convex Optimization Involving Three Composite Functions

On Solving Nonsmooth Mixed-Integer Nonlinear Programming Problems by Outer Approximation and Generalized Benders Decomposition

Continuous-Time Algorithm for Approximate Distributed Optimization with Affine Equality and Convex Inequality Constraints.

An approximation algorithm for convex multi-objective programming problems

Distributed synchronous and asynchronous algorithms for semi-definite programming with diagonal constraints

A distributed primal-dual interior-point method for loosely coupled problems using ADMM

Outer Approximation Scheme for Weakly Convex Constrained Optimization Problems

Local to Global: A Distributed Quantum Approximate Optimization Algorithm for Pseudo-Boolean Optimization Problems

A New Randomized Block-Coordinate Primal-Dual Proximal Algorithm for Distributed Optimization

Byzantine-Resilient Distributed Optimization of Multi-Dimensional Functions

A Convexification-based Outer-Approximation Method for Convex and Nonconvex MINLP

Distributed Stochastic Bilevel Optimization: Improved Complexity and Heterogeneity Analysis

Distributed Discrete-time Dynamic Outer Approximation of the Intersection of Ellipsoids

Distributed Synchronous and Asynchronous Algorithms for Semidefinite Programming With Diagonal Constraints

Distributed outer approximation of the intersection of ellipsoids

The Complexity of Distributed Approximation of Packing and Covering Integer Linear Programs