Scribd
Upload
35 views1 page

Saddle Point Integrals

This document discusses the saddle point approximation method for evaluating integrals of the form I= ∫ab g(t)e^f(t) dt, where f(t) has a sharp maximum at t0 inside the integration interval. The key steps are: 1) Taylor expand f(t) around t0, keeping terms up to order (t-t0)^2. 2) Approximate the integral as a Gaussian integral, yielding I ≈ 2πg(t0)e^f(t0)|f'(t0)|. 3) Estimate the next order correction term, showing the approximation is valid when |f'(t0)| >> 1/t0^2, meaning

Uploaded by

strumnalong27
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
35 views1 page

Saddle Point Integrals

This document discusses the saddle point approximation method for evaluating integrals of the form I= ∫ab g(t)e^f(t) dt, where f(t) has a sharp maximum at t0 inside the integration interval. The key steps are: 1) Taylor expand f(t) around t0, keeping terms up to order (t-t0)^2. 2) Approximate the integral as a Gaussian integral, yielding I ≈ 2πg(t0)e^f(t0)|f'(t0)|. 3) Estimate the next order correction term, showing the approximation is valid when |f'(t0)| >> 1/t0^2, meaning

Uploaded by

strumnalong27
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

PHY410, Notes #1, Saddle Point Approximation

Consider integral of the type


b

I=
a

g (t)ef (t) dt,

(1)

where function f (t) has sharp maximum at some point t0 inside of the integration interval a < t0 < b while g (t) is smooth near t0 . It is also important that ef (t) is rapidly decaying function away from t0 . Under such initial assumptions the integrand g (t)ef (t) can be greatly simplied near t0 . Indeed, let us expand f (t) into the Taylor series near the maximum f (t) f (t0 ) + f (t0 ) (t t0 )2 + . . . 2 (2)

The rst order term is absent by assumption, since at the maximum f (t0 ) = 0, furthermore f (t0 ) < 0. The sharpness of the maximum is guaranteed by the condition |f (t0 )| 1/t2 0 . We then have approximately
+

I g (t0 )

ef (t0 )

|f (t0 )| (tt0 )2 2

dt =

2 g (t0 )ef (t0 ) |f (t0 )|

(3)

where integration limits can be extended to innite range [, +] since integrand is rapidly decreasing function for any deviation t > 1/ |f (t0 )| t0 . Let us estimate now the next leading order term to the integral. If we take cubic term in the Taylor series of f (t) and then expand exponent this term drops out upon integration since function is odd over t t0 . We thus need forth order term in the Taylor series which we write (iv ) (iv ) 4 ef (tt0 ) /24 1 + f 24 (t t0 )4 and nd for the correction
+

I = g (t0 )ef (t0 )

|f | f (iv) 2 (t t0 )4 e 2 (tt0 ) dt = 24

2 f (iv) g (t0 )ef (t0 ) . |f (t0 )| 8(f )2

(4)

To have I/I 1 we must guarantee that f (iv) /(f )2 1. If f (t) is the single parameter function ( iv ) 2 then f /(f ) 1/f (t0 ). As a result, saddle point approximation is applicable if f (t0 ) 1 which 2 is equivalent to the earlier made assumption |f (t0 )| 1/t0 . The name of the method saddle point approximation comes from the complex analysis. If one makes substitution t t0 = iz then exponent under integrand will become a growing function. In other words, in the complex plane of t point t0 is a saddle point. Then integration must be determined by the steepest descent away from the saddle point for convergence.

You might also like