The stationary phase method estimates an oscillatory integral for large positive . Where the real phase has nonzero derivative, oscillations largely cancel and integration by parts gives decay. Near an isolated nondegenerate stationary point , the quadratic expansion leaves a contributing width of order .[ref]
After localizing a smooth amplitude near , the one-dimensional stationary-phase formula gives the leading termThe Fresnel integral supplies the phase shift. In several dimensions, the Hessian matrix determinant determines the scale and its signature determines the phase shift. Contributions from different stationary points add, while boundaries or degenerate stationary points require their own analysis. A stationary point whose amplitude vanishes can have a smaller leading order; it need not dominate an ordinary endpoint.
The stationary endpoint contributes by method of steepest descent along . Its ordinary term vanishes because . A nonstationary upper endpoint with contributes and therefore dominates the stationary contribution. The stationary phase method must retain both this boundary term and the first nonzero local-amplitude term.
For with Lorentzian displacement and zero initial velocity, the Fourier transform is . Along a fixed ray , the two phases have saddles and curvatures . Applying stationary phase and adding the conjugate contributions gives the displayed leading term. The finite pulse width supplies the attenuation factor; it may be dropped only if . The expansion is an additive statement, which remains meaningful at zeros of the leading cosine.
If the first two phase derivatives vanish at an endpoint but the third does not, an oscillatory contribution is concentrated on a width proportional to . For a smooth nonzero endpoint oscillatory integral amplitude, rescaling reduces the leading contribution to a cubic oscillatory integral. Its constant follows from the Gamma function Fourier integral. This is a degenerate form of the stationary phase method, and a quadratic formula is insufficient.
At equal argument and large integer order, the Bessel function of the first kind has phase near the endpoint. The cubic stationary endpoint yields the displayed positive constant. It is also the zero-argument value of the Airy function in the uniform Bessel turning-point approximation.
In an integral , the real-valued function is the phase. Its stationary points satisfy ; its Hessian matrix determines the local quadratic oscillation and the signature factor in stationary phase method. Degenerate stationary points and boundaries require separate analysis.
For a smooth real phase with an isolated nondegenerate interior critical point , smooth amplitude , and no competing boundary contribution, the leading contribution is , where is the Hessian matrix. Orthogonal diagonalization gives two one-dimensional Fresnel integrals. A degenerate critical point or a stationary boundary needs its own local analysis.
For fixed , the vertical sides of the half-strip integral of contribute . Integration by parts on the nonstationary horizontal pieces contributes the opposite endpoint term. After cancellation, the two stationary points give .
A stationary point of a smooth phase is nondegenerate when its Hessian matrix at is an invertible matrix. In one dimension, this condition is .
Let and let be a smooth real phase on an interval . Assume is the only stationary point of on the amplitude support and . As ,The compact support avoids endpoint terms. A local quadratic coordinate change reduces the central contribution to a Fresnel integral; repeated integration by parts controls the nonstationary remainder.
The additive remainder is deliberate. If , an asymptotic equivalence to the zero expression would be meaningless, whereas the displayed estimate still says that the integral is . For several separated nondegenerate stationary points, localize and add their contributions. A cubic stationary endpoint or a nonzero boundary amplitude is outside these hypotheses.
If a phase at an endpoint satisfies with , the leading region has width . Rescaling reduces the leading term to a cubic oscillatory integral and produces decay .
Forfixed and , the stationary point near givesWhen , the stationary point coalesces cubically with the endpoint at zero and the leading order changes to .
If the phase derivative has no zero on the support of a smooth amplitude, repeated integration by parts makes the oscillatory integral decay faster than any prescribed inverse power, subject to corresponding decay and regularity of the amplitude.
Articles by others on the same topic
There are currently no matching articles.