A modified scaling BFGS method for nonconvex minimization

Key Points

• The aim is to develop a modified BFGS method for effective unconstrained nonconvex minimization.
• Modification of the traditional BFGS algorithm.
• Focus on global convergence without convexity assumptions.
• Establishment of superlinear convergence under specific conditions.
• Numerical experiments to demonstrate effectiveness.
• The modified method shows improved performance over the traditional BFGS method.
• Results indicate global convergence for nonconvex optimization problems.
• Superlinear convergence is achieved under given assumptions.
• Numerical results validate theoretical claims.

Abstract