Parameter space for positional arguments, with indexing symbol (for readability only, useless in practice).

Exp(; name = nothing, iterations = 1, workflow = [], datadeps = [])

Parameter space for keywords arguments.

Representation of a set of function parameters (positional arguments args and keyword arguments kwargs).

Short representation of a parameter space.

A workflow with contains only one PrmSet for each function.

run(ue; remaining_iterations)

Runs an UnitExp for remaining_iterations, and returns .

addvar(e, init; save_output=true)

Creates an experiment variable, which will be initialized at init at each run.

Example: [:a => [a1,a2], :b => [b1,b2,b3]] will produce: [[:a=>a1,:b=>b1],[:a=>a1,:b=>b2],[:a=>a1,:b=>b3], [:a=>a2,:b=>b1],[:a=>a2,:b=>b2],[:a=>a2,:b=>b3]]

prmSpace = Dict(Sympols=>Arrays) => expanded prm space = [Dict(…) …]

Expands a pair :fun=>prmspace to an array of elements FunCall(fun, prmsᵢ).

Expands a PrmSpace to an array of PrmSet.

expand_value(k) returns a function f such that:

• f(v=>p::KwArgs) returns the union of k=>v and p (expanded)
• f(v) = Dict(k=>v) otherwise

Expands (Symbol, Array{FunCall}) to Array{Tuple{Symbol, FunCall}}

Returns an array of FunCalls, or [ds] if ds does not refer to a function (e.g. is already an array).

getnumthreads() Returns the number of threads used by BLAS.

Merge two expanded parameter spaces, i.e. arrays of dicts, and throws an error if they have at least one key in common.

Merge two dictionaries, and throws an error if they have a key in common that would be overwritten otherwise.

newpath(resultsdir, basename)

Creates file name of the form basename_X.csv that does not exists in resultsdir.

prun(e, backend_kwargs; use_threads)

WIP Iterates over the parameter space similarly to run, but using several processes (or 1 process and multiple threads if use_threads is true). This function does not returns anything, and thus results should be stored to disk.

update_outputs!(outputs, out)

Adds the new output out to outputs. More precisely, as both out and outputs are dictionaries with keys corresponding to variable names, each pair :var => data in out yields data being added to outputs[:var].

vcat_addmissing(tables)

Concatenates vertically all DataFrames in tables, adding missing values if necessary.