The Modified HZ Conjugate Gradient Algorithm for Large-Scale Nonsmooth Optimization.

Area of Science:

• Optimization Theory
• Numerical Analysis
• Computational Mathematics

Background:

• Large-scale nonsmooth convex minimization presents significant computational challenges.
• Existing conjugate gradient (CG) methods may lack efficiency for these complex problems.

Purpose of the Study:

• To introduce and analyze the Hager and Zhang (HZ) and modified HZ (MHZ) conjugate gradient methods.
• To establish convergence properties for these novel optimization algorithms.
• To evaluate the practical efficiency of the proposed methods on large-scale nonsmooth problems.

Main Methods:

• Development of the Hager and Zhang (HZ) conjugate gradient method.
• Formulation of the modified Hager and Zhang (MHZ) conjugate gradient method.
• Theoretical analysis to establish convergence under mild conditions.
• Numerical experimentation on large-scale nonsmooth optimization problems up to 100,000 variables.

Main Results:

• Convergence of the HZ and MHZ conjugate gradient methods is proven under specific assumptions.
• Numerical results indicate superior efficiency of the proposed methods compared to existing approaches.
• The methods are validated on problems with up to 100,000 variables, demonstrating scalability.

Conclusions:

• The HZ and MHZ conjugate gradient methods offer an efficient approach for large-scale nonsmooth convex minimization.
• The established convergence properties and demonstrated numerical performance support their utility in computational mathematics.
• These methods represent a valuable advancement for tackling complex optimization tasks.