Milan Klöwer

What about adding a `Distributions.isprobvec(p)` check in `entropy` to avoid these cases?

Haha, I even get this  with ``` # Title ## Heading 1 ### Heading 2 Text ### Heading 3 Also text ``` What?? 😆

Okay now it gets spooky 👻  I've deleted the `### Heading 2`, written by hand another `### Heading 4` underneath and copied the working `### Heading 3` below that....

Yeah, current working around is: I take a `### HeadingX` that renders fine, copy it into the place where the non-working heading is and change `HeadingX` to `HeadingY`, how weird??

@rsignell-usgs The number of significant bits `nsb` is obtained from the number of significant bits `nsd` as `nsb = ceil(log2(10)*nsd)` whereby the `ceil` is just because rounding down can mean...

> You can see that NSD=3 keeps 11 mantissa bits for two values, 10 for three values, 9 for four values and 8 for one value. This is why it...

> For a given NSD we're only talking about a small range (up to 4) in the number of keep bits, so four patterns to substitute for instead of one...

While I don't know how many lossless algorithms work in detail, I usually think of them as "removing redudancies". If all bits `m` to `n` of every float in an...

Thanks for your input, indeed your example also works for me. In my slightly more extensive version of that code (see [output.jl](https://github.com/milankl/juls/blob/master/src/output.jl)) the variable t gets correctly written in some...

Okay, I think I found the issue. Let's extend your code so that two variables u,v with same dimensions get written to two different files, both with the same dimensions....