The Steepest-Descent Method ECE 6382 The Steepest-Descent Method D. R. Wilton ECE Dept. Notes are adapted from ECE 6341

Steepest-Descent Method Complex Integral: The method was published by Peter Debye in 1909. Debye noted in his work that the method was developed in a unpublished note by Bernhard Riemann (1863). Peter Joseph William Debye (March 24, 1884 – November 2, 1966) was a Dutch physicist and physical chemist, and Nobel laureate in Chemistry. Georg Friedrich Bernhard Riemann (September 17, 1826 – July 20, 1866) was an influential German mathematician who made lasting contributions to analysis and differential geometry, some of them enabling the later development of general relativity. http://en.wikipedia.org/wiki/Peter_Debye http://en.wikipedia.org/wiki/Bernhard_Riemann

Steepest-Descent Method (cont.) Complex Integral: We want to obtain an approximate evaluation of the integral when the real parameter Ω is very large. The functions f (z) and g(z) are analytic (except for poles or branch points), so that the path C may be deformed if necessary (possibly adding residue contributions or branch-cut integrals). Saddle Point (SP):

Steepest-Descent Method (cont.) Path deformation: If the path does not go through a saddle point, we assume that it can be deformed to do so. If any singularities are encountered during the path deformation, they must be accounted for (e.g., residue of captured poles). X

Steepest-Descent Method (cont.) Denote Cauchy Reimann eqs.: Hence

Steepest-Descent Method (cont.) or If then Near the saddle point: Ignore (rotate coordinates to eliminate)

Steepest-Descent Method (cont.) In the rotated coordinate system: Note: The angle of rotation necessary to do this will be clear later (it is the departure angle SDP of the steepest-descent path). Assume that the coordinate system is rotated so that

Steepest-Descent Method (cont.) The u (x, y ) function has a “saddle” shape near the SP:

Steepest-Descent Method (cont.) Note: The saddle does not necessarily open along one of the principal axes (only when uxy (x0, y0) = 0).

Steepest-Descent Method (cont.) Along any descending path (where the u = Re(g) function decreases) we will have convergence: Exponentially decreasing function

Steepest-Descent Method (cont.) Note: The parameter s is a measure of distance along the path from the saddle point. (Negative s means before we get to the saddle point, positive s means after we leave the saddle point.) When  is large most of contribution is from near z0.

Steepest-Descent Method (cont.) Along any descending path (where the u = Re(g) function decreases): Both the phase and amplitude change along an arbitrary descending path C. If we can find a path along which the phase does not change (v is constant), the integrand will have a purely exponentially decaying behavior (no phase term), making the integral easy to evaluate.

Steepest-Descent Method (cont.) Choose path of constant phase:

Steepest-Descent Method (cont.) Gradient Property (proof on next slide): Hence C0 is either a “path of steepest descent” (SDP) or a “path of steepest ascent” (SAP). SDP: u(x,y) decreases as fast as possible along the path away from the saddle point. SAP: u(x,y) increases as fast as possible along the path away from the saddle point.

Steepest-Descent Method (cont.) Proof C.R. Equations x y C0 Hence, Also, Hence

Steepest-Descent Method (cont.) Because the v function is constant along the SDP, we have or Real

Steepest-Descent Method (cont.) Local behavior near SP so y z0+dz r  Denote z0 x

Steepest-Descent Method (cont.) SAP: SDP: Note: The two paths are 90o apart at the saddle point.

Steepest-Descent Method (cont.) SAP SDP SP x y u decreases u increases 90o Note: v(x,y) is constant along both paths.

Steepest-Descent Method (cont.) The landscape of the function u(z) near z0 Valley Hill

Steepest-Descent Method (cont.) SAP SDP

Steepest-Descent Method (cont.) Set This defines SDP

Steepest-Descent Method (cont.) We then have or (leading term of the asymptotic expansion) Hence

Steepest-Descent Method (cont.) To evaluate the derivative: (Recall: v is constant along SDP.) At the SP this gives 0 = 0. Take one more derivative:

Steepest-Descent Method (cont.) At so Hence, we have

Steepest-Descent Method (cont.) Note: There is an ambiguity in sign for the square root: To avoid this ambiguity, define

Steepest-Descent Method (cont.) The derivative term is therefore Hence

Steepest-Descent Method (cont.) To find SDP : Denote

Steepest-Descent Method (cont.) SDP SAP Note: The direction of integration determines The sign. The “user” must determine this.

Steepest-Descent Method (cont.) Summary

Example where Hence, we identify:

Example (cont.) x y

Example (cont.) Identify the SDP and SAP: SDP and SAP:

Example (cont.) SDP and SAP: SAP Examination of the u function reveals which of the two paths is the SDP.

Example (cont.) SDP Vertical paths are added so that the path now has limits at infinity. SDP = C + Cv1 + Cv2 It is now clear which choice is correct for the departure angle:

Example (cont.) (If we ignore the contributions of the vertical paths.) Hence, so

Example (cont.) Hence

Example (cont.) Examine the path Cv1 (the path Cv2 is similar). Let

Example (cont.) since Since  is becoming very large, we can write: Hence, we have

Example (cont.) Hence, Hence, Iv1 is a much smaller term than what we obtain from the method of steepest descents, when  gets large, and we can ignore it.

Example The Gamma (generalized factorial) function: Let (This is discussed more in Notes 13.) Note: There is no saddle point! Let

Example (cont.) Hence or

Example (cont.) Recall: Hence or This is Sterling’s formula. Note: The departure angle is zero, not , since we are integrating from 0 to . The SDP is the real axis. or This is Sterling’s formula.

Complete Asymptotic Expansion We can even obtain the complete asymptotic expansion of the integral with the steepest-descent method (including as many higher-order terms as we wish). Then we have

Complete Asymptotic Expansion (cont.) (Extending the limits to infinity gives an exponentially small error: This does not affect the asymptotic expansion.) Assume (only even powers contribute to the integral below) Then This result is called “Watson’s Lemma”: We can obtain the complete asymptotic expansion of I by substituting in the complete asymptotic expansion for h and integrating term-by-term.

Complete Asymptotic Expansion (cont.) Performing the integration, where We then have

Complete Asymptotic Expansion (cont.) Example: We have SDP: y = 0 (real axis) SAP: x = 0 (imaginary axis)

Complete Asymptotic Expansion (cont.) so Hence

Complete Asymptotic Expansion (cont.) Hence Recall:

Complete Asymptotic Expansion (cont.) Recall: Hence we have: so where (Please see Notes 13.)

Complete Asymptotic Expansion (cont.) Hence we have: as Note: In a similar manner, we could obtain higher-order terms in the asymptotic expansion of the Bessel function or the Gamma function.