BakerStreet

This package facilitates one of the many ways DrWatson.jl enables users to set up and run simulations, and to store and retrieve the results of these simulations.

The workflow targeted by BakerStreet.jl has the following structure:

using BakerStreet
using DataFrames

function adder(;a, b)
    return (;c = a+b)
end

a = [1,2,3]
@runsims adder a b=[10,20]

df = @collect_results

The simulation is started on by macro @runsims. A configuration is created and the payload method adder is called over the Cartesian product of parameter ranges (a,b). The dictionary that is written to disk is the merge of the inputs and outputs of adder (so in this case the keys are (a,b,c)). The results are written in a subdirectory of DrWatson.datadir() with name splitext(@__FILE__)[1]. Likewise, @collect_results looks for output data in this directory.

The DataFrame created by @collect_results is:

 Row │ c       b       a       path
     │ Int64?  Int64?  Int64?  String?
─────┼───────────────────────────────────────────────────────────
   1 │     11      10       1  C:\\Users\\.julia\\dev\…
   2 │     21      20       1  C:\\Users\\.julia\\dev\…
   3 │     12      10       2  C:\\Users\\.julia\\dev\…
   4 │     22      20       2  C:\\Users\\.julia\\dev\…
   5 │     13      10       3  C:\\Users\\.julia\\dev\…
   6 │     23      20       3  C:\\Users\\.julia\\dev\…

Running this script results in the following directory contents:

│   helloworld.jl
│   Manifest.toml
│   Project.toml
│
└───data
    └───helloworld
            a=1_b=10.jld2
            a=1_b=20.jld2
            a=2_b=10.jld2
            a=2_b=20.jld2
            a=3_b=10.jld2
            a=3_b=20.jld2

This package can be seen as a template or recipe that encodes a very specific choice for the options and conventions that one can use when running simulations with DrWatson.jl.