A pescador paper?

[bmcfee]

What do folks think about putting together a short paper about pescador, once the next major revision is done? I'm thinking the journal of open source software would be a good venue: see http://joss.theoj.org/about .

Discuss?

[cjacoby]

👍

On Wed, Aug 10, 2016, 07:14 Brian McFee [email protected] wrote:

What do folks think about putting together a short paper about pescador,
once the next major revision is done? I'm thinking the journal of open
source software would be a good venue: see http://joss.theoj.org/about .

Discuss?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#32, or mute the thread
https://github.com/notifications/unsubscribe-auth/AA4t8yw67gU9gFY6anVIwICij5u-eC6Mks5qebKLgaJpZM4JhCPV
.

[cjacoby]

What about also a medium post, or something like?

[bmcfee]

I don't want to do medium, for lack of peer review / credibility. But we can certainly publicize it there aftwerward.

I think JORS might be a better venue for this than JOSS. JOSS is extremely light-weight in terms of paper contents, and the focus is mainly on the code quality. JORS is more about software as research artifacts, and the papers can be a bit longer and include experiments and motivation.

In my head, this paper could look as follows:

1. Intro - why something like pescador is necessary
2. Tour of components, roughly following the API docs
   a. streamers
   b. muxen
   c. zmq
   d. maps
3. Example use cases and patterns

• roughly following the "advanced examples", but not in so much detail
• training streamers
• validation epochs
• stratified sampling

4. Experiments
   a. Measuring poissonmux behavior (deviation from uniform distribution as a function of rate, n_active, n_streams)
   b. Benchmarking zmq / mux interplay

What do y'all think? I'm happy to take the lead on this, and think we could jam it out pretty quickly after the 2.0 release.