A direct method is described for locating saddle pomts on multidimensional surfaces. It combines aspects of the COIIJUgate gradient method and the quasi-Newton mmimization method\_
A method for locating a limit saddle point
โ Scribed by S.L. Ivanov
- Publisher
- Elsevier Science
- Year
- 1986
- Weight
- 245 KB
- Volume
- 26
- Category
- Article
- ISSN
- 0041-5553
No coin nor oath required. For personal study only.
๐ SIMILAR VOLUMES
A gradient method is presented which can be viewed as a continuous-time version of the 'distinguished (reaction) coordinate' method, one of the oldest techniques for finding transition states (saddles of first order). The method offers a simple uniform algorithm for finding saddles of any order, is
We present an algorithm which can be used to find saddle points of non-linear functionals or functions with a large number of variables. The saddle points that can be found have only one unstable direction. The algorithm is tested on a "~" functional and yields the well known circular "critical drop