Tim DuBois

Faster looks like we could pull out a few more vectorisations without going to intrinsics. Take a look.

When looking into the best way of visualising volumetric data, I've come across a few more ways of doing that. So consider these as demo output options: - [Vaex](http://vaex.astro.rug.nl/)

I think it'd be a good idea to run a fuzzer over everything in addition to unit testing. Let's do that soon.

Output type enum

4

Now that I've implemented serde 1.0 in ndarray, we can serialise data. For the moment we have the ability to write csv, message pack, yaml and json. Config input is...

Consider optional installs for file formats

2

At the moment we have yaml, json, ron, messagepack and csv formats - each require a library to pull in and compile. None of them really need too much overhead,...

We have an explicit time step, but our problem is also quite stiff. We may be able to a) relax the `dn^2/3` condition slightly or better yet b) relax the...

Turning any type of symmetry on at the moment causes a good deal of crazy. It's probably due to the indexing of things since we're doing this manually.

Push data write to thread

3

Since we have `w_store` persistent, we can send of the file writes to a future or rayon thread.

fatal runtime error: out of memory

2

Not too sure who is calling this, but if we look at pushing boundaries and calculate a grid of 1500x1500x1500, we hit the above error sometime during the run. RAM...

Complex numbers as option

6

Currently we are using just reals for potential and wavefunction values. Need to extend this to complex values. However, I don't want to be sending round two `f64` values for...