Abstract:This paper focuses on the cardinality constrained mean-variance portfolio optimization, in which only a small number of assets are invested. We first treat the covariance matrix of asset returns as a diagonal matrix with a special matrix processing technique. Using the dual theory, we formulate the lower bound problem of the original problem as a max-min optimization. For the inner minimization problem with the cardinality constraint, we obtain its analytical solution for the portfolio weights. Then, the lower bound problem turns out to be a simple concave optimization with respect to the Lagrangian multipliers. Thus, the interval split method and the supergradient method are developed to solve it. Based on the precise lower bound, the depth-first branch and bound method are designed to find the global optimal investment selection strategy. Compared with other lower bounds and the current popular mixed integer programming solvers, such as CPLEX and SCIP, the numerical experiments show that our method has a high searching efficiency. History: Accepted by Pascal Van Hentenryck, Area Editor for Computational Modeling: Methods & Analysis. Funding: This work was supported by the National Natural Science Foundation of China [Grants 12101317, 12271071, and 11991024], the Natural Science Foundation of Jiangsu Province [Grant BK20200819], the Team Project of Innovation Leading Talent in Chongqing [Grant CQYC20210309536], the Contract System Project of Chongqing Talent Plan [Grant cstc2022ycjh-bgzxm0147], and the Philosophy and Social Science Fund of Education Department of Jiangsu Province [Grant 2020SJA0168]. K.F.C. Yiu is supported in part by the Research Grants Council of Hong Hong [Grant PolyU 15223419], the Hong Kong Polytechnic University [Grants 4-ZZPT and 1-WZ0E], and the Research Centre for Quantitative Finance [Grant 1-CE03]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2022.0344 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2022.0344 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

Efficient Projected Gradient Methods for Cardinality Constrained Optimization

Efficient Projected Gradient Methods for L 0 Constrained Optimization

Convergence of Projected Subgradient Method with Sparse or Low-Rank Constraints

Convergence Analysis of Projected Gradient Descent for Schatten-p Nonconvex Matrix Recovery

An Efficient Optimization Approach for a Cardinality-Constrained Index Tracking Problem

O(log T) Projections for Stochastic Optimization of Smooth and Strongly Convex Functions

Functional Constrained Optimization for Risk Aversion and Sparsity Control

Random Multi-Constraint Projection: Stochastic Gradient Methods for Convex Optimization with Many Constraints

Computing the Projection on the Intersection of Linear Equality and Cardinality Constraints

A Smoothing Proximal Gradient Algorithm for Nonsmooth Convex Regression with Cardinality Penalty.

A New Nonmonotone Spectral Projected Gradient Algorithm for Box-Constrained Optimization Problems in M×n Real Matrix Space with Application in Image Clustering.

Scaled Proximal Gradient Methods for Sparse Optimization Problems

Averaging Random Projection: A Fast Online Solution for Large-Scale Constrained Stochastic Optimization.

Monotone Projected Gradient Methods for Large-Scale Box-Constrained Quadratic Programming

Cardinality Constrained Portfolio Optimization via Alternating Direction Method of Multipliers

Efficient Euclidean projections via Piecewise Root Finding and its application in gradient projection

Delayed Projection Techniques for Linearly Constrained Problems: Convergence Rates, Acceleration, and Applications

An Efficient Global Optimal Method for Cardinality Constrained Portfolio Optimization

Efficient Projection-free Algorithms for Saddle Point Problems

Projected Neural Network for a Class of Sparse Regression with Cardinality Penalty

Optimal Stochastic Strongly Convex Optimization with a Logarithmic Number of Projections