# Numerically solving a 2D oscillating integral

I'm having trouble solving this integral numerically:

Integrand2 :=  I/λ Holo[ξ, η] E^(I (2 π)/λ r)/r
ResultsTable2 = Table[NIntegrate[Integrand2, {ξ, -20, 20}, {η, -20, 20},
Method -> "MultidimensionalRule"], {x, -15, 15, 5}, {y, -15, 15, 5}]


The function called Holo is modelling an aperture that looks like this:

To put it simply: I'm trying to do Fresnel diffraction with this as the aperture.

My goal is to get a 2D inverse Fourier transform by summing over the entire Holo term (a function of ξ and η). As you can see from the image, it oscillates rapidly, so unless I integrate over a very small area of ξ and η, I get the following error:

NIntegrate::slwcon: Numerical integration converging too slowly; suspect one of the following: singularity, value of the integration is 0, highly oscillatory integrand, or WorkingPrecision too small.

To get a good 2D inverse Fourier transform, I need to have a larger area. Ideally, integrating over 100 by 100 would be good. Is there some kind of method or work around that can make this happen?

I can provide more details if needed. I'm new to Mathematica and to this site, so any answer is appreciated. I can learn from any comments or suggestions you have :)

EDIT:

I guess to make my question a bit clearer: how can I numerically integrate a 2D oscillating function?

-
I'm afraid I can't help with this but providing code for Holo may help you get answers. –  Mr.Wizard Jul 22 '14 at 14:26
You can try to increase the WorkingPrecision by adding the option WorkingPrecision->50 (with 50 just as an example). Some general information about numerical integration, their warning messages, and potential solutions can be found here‌​. Sufficent code to reproduce your problem seems to be necessary to give a more specific answer. –  Karsten 7. Jul 22 '14 at 15:02