Questions about running a workshop

General Usage
1

Hello,

I am a co-chair of an international conference in mathematics & technology to be held in December. In an unguided moment, I breezily offered to run a one hour workshop on Julia, especially emphasizing its speed over, say, Python.

Now I can do this, but my besetting sin is always trying to cram too much in - my enthusiasm gets the better of me. But I do have a few questions:

  • What is the easiest environment to use? I’d rather something online, as any downloads of software are always fraught with difficulties.

  • Any ideas on topics which are particularly good for Julia and mathematics? I was thinking of exploring a few brute force computations of exponential or factorial size, but I don’t actually know if these show Julia at its best.

In case you’re wondering - why is this fool intending to run a workshop when he seems to know nothing about the language? - I offer the excuse that I used to use Julia extensively, but haven’t really done so since my retirement over 2 years ago, and I am quite rusty.

Anyway, thanks for any advice!

Alasdair

2 Likes
2

I would recommend maybe even doing the whole talk / Workshop just in Pluto.

As resources I would recommend

as I did for example when I designed this talk

where the package details in the end are very likely too technical - and instead of LaTeX, Id probably use Pluto and its presentation mode next time.

3

Google Colab has support for Julia now, so that is likely the easiest online environment for this. This will also allow you to share a folder of notebooks with the audience beforehand and structure the workshop around them.

JuliaHub also offers 20 hours of free usage per month per account, but that will involve everyone setting up a JuliaHub account - and in my experience, there’s always at least a few that don’t do it beforehand even if requested to do so, so there’s waiting and hassles and distracted listeners. On the plus side, I believe this will land them in a VS Code environment with the editor and (I think) LSP and a terminal, so it’s something closer to the working environment of most regular Julia users.

JuliaHub would also give them the shiny new releases, whereas with Colab:

We’ve chosen to distribute the 1.10 LTS version of Julia for longer-term stability. In addition to the required IJulia kernel package, we’ve pre-installed a few of the most common packages for data science, such as CSV, DataFrames, Makie, and Plots, and we’re evaluating pre-installing CUDA on our GPU runtimes, but for now we’ve left it off.

IMHO 1.10 is plenty for new users, and the pre-installed packages should also give you some useful amount of things to work with and show off, without having to straightaway jump into precompilation and TTFX-related topics (though I’d recommend covering those too, just a bit later after they’ve been able to run some useful code and get a taste for the language).