Git Product home page Git Product logo

vic2d's Introduction

vic2d

A command-line two-dimensional fluid simulator using a novel semi-Lagrangian advection technique

Building vic2d

We are now using CMake to build the code, though you can still use the unmaintained unix Makefile.

git clone https://github.com/markstock/vic2d.git
cd vic2d
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

Or, Windows users can run the cross-compiled binary included in this distribution (vic2d.exe).

Running vic2d

To get a list of options, run

vic2d -h

Note that all diffusivities are coefficients, so for Reynolds number 10000, you should set momentum diffusivity with -vd 0.0001

Also note that the actual resolution of the output images will be one higher than that given on the command-line. So -x 1024 -y 1024 will generate square PNG images with 1025 pixels on a side. And any input images for such a run should have 1025 pixels on a side. I know, this is not intuitive. I created it this way because for the multigrid solver to function efficiently, you should try to run at resolutions that are small multiples of large factors of 2, so 2560 (5 * 128) is much better than 2544 (159 * 16), and those numbers are easier to remember.

For some sample command lines and output, including links to YouTube videos, check out the old vic2d page.

Theory

This code uses a vortex method solver, which means that pressure doesn't enter in the equations: the formulation is in velocity-vorticity coordinates. The velocity-vorticity inversion is accomplished with MUDPACK, a multigrid solution method. Interpolation to and from the grid is done with a 4th order kernel (M4'). Advection uses a semi-Lagrangian method, in which the vorticity at each grid point is drawn from a 4th order (Runge-Kutta) backward-looking projection. All of these methods contribute to the code's remarkable lack of numerical diffusion - you can run problems with Reynolds numbers in the hundreds of millions - though the drawback is that the simulation is not unconditionally stable.

The idea for a semi-Lagrangian vortex methods comes from a 1969 paper on atmospheric physics that I, unfortunately, cannot find any more.

Thanks

This code uses libpng and mudpack, a multigrid solver. All the rest of this ugly, ugly code is mine.

vic2d's People

Contributors

markstock avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.