dprada
commented
2 years ago The binary file as the first attempt
Why don't we store the trajectory as a binary file? If we know that each frame entry comes with four floats: time, x, y, and z. And we agree on the units: time in ns and lengths in angstroms. What would be the reduction factor in the size of the trajectory file (compared with the UTF-8 text file)?
We are currently saving the trajectory in a UTF-8 encoded text file with four columns: time, $x$, $y$, and $z$. Not a very wise choice.
The size of trajectory with 15 microseconds saved every 10 picoseconds is ~118M. Not big if we work locally with this file on our machine. But big enough to be uploaded to this repository (There is a 100M limit per file). So... we need to work with a better file format.