Use standard data sets for benchmarking and examples.

[danielballan]

Low priority, but a note for later...

It would be nice to have examples at different densities to answer questions like, "Can our BTree mode ever beat KDTree, even though it's in pure Python?"

We could use a subset of the data from this 2012 particle tracking competition. They systematically cover a range of densities and signal-to-noise ratios. The results were published in Nature. You can download the videos and so-called "ground truth" trajectories against which entries were judged.

h/t to Ben Schuster for pointing me to the paper.

[danielballan]

I had hoped to incorporate subsections of the tracking challenge videos in the trackpy-examples repo. Acquiring them from the "tracking challenge" requires user registration, which only especially committed contributors would bother to do.

For the present, Erik declined to give permission to redistribute the video, but he supports my intention to use it and direct others to use it too:

Thanks for your interest. We are still in the process of revamping the website and writing down the precise conditions of further usage of the challenge material. As indicated on the website, originally the material was available only to registered participants and only for the purpose of the challenge. Now that the challenge paper is published we will open up the usage for anyone interested in benchmarking their algorithms. However, at least for this moment, we ask people not to redistribute the material but to provide a link to the website, so that the website remains the centralized repository for the challenge material, where all complete info can be found.

[tacaswell]

The solution may be to make sure they provide a stable url and we just need to learn enough of urllib to fetch files. In the end, that is better anyway so we don't have to put all that binary data in our repositories.

If we are super clever, we could get our wrappers to open urls just as easily as local files.