Where do I find sourcecode of a 2D vortex method implementation (not "Blobflow")? - I'm interested in how a few things are done best... Type of flow: Mainly (inviscid) free flow (interaction of vortices, ...), i.e. as basic as possible.

Thank you Michael Bergdorf

If you wish to study the basics of vortex interactions in an inviscid flow environment (using vortex methods) you can write the code yourself in less than a week. It is really straight-forward. Check out any decent paper or the book by Chorin & Marsden. All you really need to do is write a subroutine that evaluates the velocities for N particles (that's just two nested do loops), and another subroutine to move the particles based on your computed velocities.

It appears that blobflow is free, so you can take the diffusion part out of the computation and you have what you need, really. Rossi (the author of blobflow) has developed a higher order method based on eliptical core functions (vs. circular), so you'll have the opportunity to study the effect of core function on solution accuracy and convergence rate as well (I'm sure you can convert the ellipse to a circle in the code, though I haven't checked the code myself)

The difficulty/complexity with vortex methods is in the implementation of boundaries, fast solvers and diffusion schemes, especially in 3-D - non of which is what you're looking for.

Adrin Gharakhani

I'm primarily interested in different means of implementation (of the fast summation), e.g: - poisson integral method - of course fast multipole - normal barnes-hut And I'm encountering difficulties in writing those from scratch.

Thanks for your response

Michael Bergdorf

If it is comparison between fast and direct solvers you are looking for (for debugging purposes), then coding the direct part is (less than) a one day job. It's just a couple of nested do loops summing the Biot-Savart law over all the particles. I don't know if you can get free fast solvers. (If you do, I'd be interested to know the source)

Adrin Gharakhani

(1). Could you write down the equations and the procedures first, then find a programmer to code it for you? (2). If the problem is not in coding, then I guess, you will have to do some more journal paper searches to locate the method first. Have you try the forum/resources and books sections?

A 2D fast multipole method is available by Petros Koumoutsakos, petros@ifd.mavt.ethz.ch, as said in his book "Vortex method: theory and practice" (Cambridge University Press).It is an O(N) vectorized code, wich can be used to practice LES.

For other fast solvers, you can try with astrophysics codes, which are N-problem solvers. Indeed the code from Barnes is downloadable at: ftp://hubble.ifa.hawaii.edu/pub/barnes/treecode/

Best,

Jerome