← Back to context

Comment by lupsasca

4 days ago

In a nutshell, the app computes a map from texture coordinate to texture coordinate. This map is itself stored as a texture --- to obtain the value of the map on texture coordinates (x,y), one samples the texture at (x,y) and the resulting float4 contains the outputs (x',y') as well as a status code.

When the user selects the "Static black hole" mode, this texture is computed on the GPU and cached. The "Kerr black hole" textures, however, have been precomputed in Mathematica, due to the need for double precision floating point math, which is not natively available in Apple's Metal shading language.

The source code, including the Mathematica notebook, can be found here https://github.com/graveltr/BlackHoleVision.

5 comments

lupsasca

Reply
lupsasca  4 days ago

We hope you enjoy watching the world with Black Hole Vision and welcome any questions or feedback. If you like the app, please share it with your friends!

The code was written at Vanderbilt University by Trevor Gravely with input from Dr. Roman Berens and Prof. Alex Lupsasca. This project was supported by CAREER award PHY-2340457 and grant AST-2307888 from the National Science Foundation.

License: This app includes a port of the GNU Scientific Library's (GSL) implementation of Jacobi elliptic functions and the elliptic integrals to Metal. It is licensed under the GNU General Public License v3.0 (GPL-3.0). You can view the full license and obtain a copy of the source code at: https://github.com/graveltr/BlackHoleVision.

  • timthorn  4 days ago

    By any chance, was Andrew Strominger involved in this at all? He gave the Andrew Chamblin Memorial Lecture in Cambridge last month and demoed something that looked similar.

    • lupsasca  4 days ago

      I think what he showed you was likely a version of this that was coded up by Harvard graduate student Dominic Chang: https://dominic-chang.com/bhi-filter/

      It works very well (and in a browser!) but is limited to a non-rotating (Schwarzschild) black hole---we really wanted to include black hole spin (the Kerr case). As we write on the github, talking with Dominic about his implementation was very useful and we are hoping to get a paper explaining both codes out before the end of the year.

      1 reply →