An Efficient Dimer Method With Preconditioning And Linesearch
arXiv:1407.28177 citationsh-index: 37
Originality Synthesis-oriented
AI Analysis
For researchers using Hessian-free saddle point algorithms, this work provides an improved variant with automatic step size selection and preconditioning, though the improvements are incremental.
The authors augment the dimer method for saddle point computation with linesearch and preconditioning, proving local linear convergence and demonstrating significant performance gains in numerical tests.
The dimer method is a Hessian-free algorithm for computing saddle points. We augment the method with a linesearch mechanism for automatic step size selection as well as preconditioning capabilities. We prove local linear convergence. A series of numerical tests demonstrate significant performance gains.