Variable Metric Method for Unconstrained Multiobjective Optimization Problems

Jian Chen,Gaoxi Li,Xinmin Yang

DOI: https://doi.org/10.48550/arXiv.2207.07286

2022-07-15

Abstract:In this paper, we propose a variable metric method for unconstrained multiobjective optimization problems (MOPs). First, a sequence of points is generated using different positive definite matrices in the generic framework. It is proved that accumulation points of the sequence are Pareto critical points. Then, without convexity assumption, strong convergence is established for the proposed method. Moreover, we use a common matrix to approximate the Hessian matrices of all objective functions, along which, a new nonmonotone line search technique is proposed to achieve a local superlinear convergence rate. Finally, several numerical results demonstrate the effectiveness of the proposed method.

Optimization and Control

What problem does this paper attempt to address?

This paper attempts to address some key challenges in multi - objective optimization problems (MOPs). Specifically, the main contribution of the paper lies in proposing a variable metric method for unconstrained multi - objective optimization problems. The following are the specific problems that the paper attempts to solve:

1. **Generating Pareto Critical Points**: The paper first generates a series of points by using different positive definite matrices and proves that the accumulation points of these points are Pareto critical points. This solves the problem of how to find a set of reasonable solutions in multi - objective optimization.

2. **Strong Convergence**: Without assuming convexity, the paper establishes the strong convergence of the proposed method. This means that even if the objective function is not convex, the algorithm can converge to Pareto critical points.

3. **Local Super - Linear Convergence Rate**: In order to improve the convergence speed of the algorithm, the paper proposes a new non - monotonic line - search technique, using a general matrix to approximate the Hessian matrices of all objective functions. This method can achieve a local super - linear convergence rate, thus finding the optimal solution more quickly in practical applications.

4. **Numerical Results Verification**: Finally, the paper verifies the effectiveness of the proposed method through a series of numerical experiments, demonstrating its application potential in practical problems.

### Key Technologies and Methods in the Paper

- **Variable Metric Method**: By using different positive definite matrices to generate descent directions, it is ensured that the algorithm can effectively find Pareto critical points.
- **Proof of Strong Convergence**: Without assuming convexity, the proof of the strong convergence of the algorithm is provided, which is an important theoretical contribution in multi - objective optimization.
- **Non - Monotonic Line - Search Technique**: A new non - monotonic line - search technique is proposed, enabling the algorithm to maintain a high convergence speed at each iteration.
- **Approximation of Hessian Matrix**: A general matrix is used to approximate the Hessian matrices of all objective functions, simplifying the computational complexity while maintaining the efficiency of the algorithm.

### Application Background

Multi - objective optimization problems are widely applied in many fields such as engineering, economics, bioinformatics, and water resource management. The variable metric method proposed in the paper provides a new and effective tool for solving complex optimization problems in these fields.