Skip to content

peter-nekrasov/variable-coefficients

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

75 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Peter's repository for solving variable coefficient scattering problems in 2D. The exact form of these scattering problems is below. This code uses the fact that the (adjoint) Lippman-Schwinger equation for each problem looks like an identity term plus the sum of diagonal times translationally invariant operators:

$$ \mu(x) + \sum_{i=1}^N V_i(x) \int_{\Omega} K_i (x-y) \mu(y) \ d y = f(x) $$

where $\Omega \subset \mathbb{R}^2$. The code stores both the coefficients $V_i$ and the kernels $K_i$ as rank 3 tensors and sums them up in a systematic way. The quadrature used here is Zeta-corrected trapezoidal rule (see Wu & Martinson, 2021).

There is currently support for:

  • Helmholtz:

$$ \Delta u + k^2(1+V(x)) u = 0 $$

  • Simplified Flexural (sf):

$$ \Delta^2 u - k^4(1+V(x)) u = 0 $$

  • Simplified Flexural-Gravity (sfg):

$$ \frac12 ( \alpha \Delta^2 - \beta (1+V) ) u + \gamma \int_{\Omega} \frac{1}{4\pi | x - y \ |} u(y) dy = 0 $$

  • Flexural (flex):

$$ \left( \Delta( \alpha \Delta) + (1-\nu) \left( 2 \partial_{x_1 x_2} \alpha \ \partial_{x_1 x_2} - \partial_{x_1}^2 \alpha \ \partial^2_{x_2} - \partial_{x_2}^2 \alpha \ \partial_{x_1}^2 \right) - \beta \right) u = 0 $$

  • Flexural-Gravity (flex-gravity):

$$ \frac12 \left( \Delta( \alpha \Delta) + (1-\nu) \left( 2 \partial_{x_1 x_2} \alpha \ \partial_{x_1 x_2} - \partial_{x_1}^2 \alpha \ \partial^2_{x_2} - \partial_{x_2}^2 \alpha \ \partial_{x_1}^2 \right) - \beta \right) u + \gamma \int_{\Omega} \frac{1}{4\pi | x - y \ |} u(y) dy = 0 $$

where $\alpha$ is the rigidity, $\beta$ is the difference between the plate's inertia and gravitational force on the fluid (when it exists) and $\gamma$ is related to the hydrodynamic pressure.

These scattering problems are solved using either FFT + GMRES or using recursive skeletonization with proxy surfaces. The former tends to be faster, but the latter may be better for geometries with resonances where GMRES tends to suffer.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •