scientific article; zbMATH DE number 1193520

MaRDI QIDQ3841983zbMATH OpenFDO

Publication date 21 January 1999

Cauchy Debye estimation of integrals Laplace Nekrasov Riemann steepest descent method

History of mathematics in the 19th century (01A55) History of mathematics in the 20th century (01A60)

The saddle point method or the method of the fastest (steepest) descent is a natural development of methods Laplace used to estimate integrals of the form \[ I_n=\int^b_af^n(x)g(x)dx=\int^b_a e^{nu(x)}g(x)dx,\quad f(x)>0, u(x)=\ln f(x) \] for \(n\to\infty\) with the conditions that \(u\) has a unique maximum at \(c\in (a, b)\) and \(u'(c)=0, u(c)<0\) and \(g\) is continuous and \(g(c)\neq 0\). The methods have to be changed in case \(u\) is complex-valued, by deforming the contour of integration inside the domain where functions \(u\) and \(g\) are analytic in such a way that certain conditions are satisfied. The authors discuss Cauchy's anticipation of this method in his 1827 paper; they show how Cauchy made a crude error in applying methods of Laplace. The saddle point method was more finely used (or introduced) in a Riemann's manuscript published by Schwartz in Italian, in 1863; Schwartz made substantial additions to this paper. The authors further argue that it was P. A. Nekrasov who has precedence in the saddle-point method (published in 1886) over usually quoted P. Debye (published in 1909). Nekrasov published later a number of references in the area, except Riemann, hence it appears he was not familiar with Riemann's work. Nekrasov uses the saddle-point method to estimate the general integral, and he allows for several saddle points with arbitrary multiplicity while Debye considers a special case in each of these settings.

Recommendations

Cited in

(8)

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3841983)