BenchmarkProfiles.jl
BenchmarkProfiles.jl copied to clipboard
Performance and data profiles
They advertise fastest time to first plot in Julia. https://github.com/JuliaComputing/PlotlyLight.jl
https://github.com/JuliaSmoothOptimizers/BenchmarkProfiles.jl/blame/ccf20e9027f6d67053a4ae3bb65b2c434d96cc41/src/requires.jl#L149 and https://github.com/JuliaSmoothOptimizers/BenchmarkProfiles.jl/blame/ccf20e9027f6d67053a4ae3bb65b2c434d96cc41/src/requires.jl#L174 @geoffroyleconte
GRUtils is a user-friendly Julian interface to GR. A `stairs` function is available: https://heliosdrm.github.io/GRUtils.jl/stable/api/plotting/#GRUtils.stairs
GR has a `stairs` function but it does not appear in the documentation: https://github.com/jheinen/GR.jl/blob/e787dadb261206d4d93b7f6478fa55bb0de6606e/src/jlgr.jl#L1816 GR comes with much fewer dependencies than Plots.jl and is much faster.
https://jso.dev/tutorials/introduction-to-benchmarkprofiles/ Connected to https://github.com/JuliaSmoothOptimizers/JuliaSmoothOptimizers.github.io/issues/172
Implemented according to #48 and with [SolverBenchmark.jl](https://github.com/JuliaSmoothOptimizers/SolverBenchmark.jl/blob/master/src/profiles.jl) as metioned by @tmigot
Using data frames from [DataFrames.jl](https://github.com/JuliaData/DataFrames.jl) as input for `performance_profile()` would be usefull. Perhaps, the syntax could be to have a DataFrame as ``` julia df = DataFrames(ProbName=String[],Solver1=Int[],Solver2=Int[],Solver3=Int[]) ``` then push...
Currently, we implement Dolan and Moré performance profiles and Moré and Wild data profiles. For certain applications, including DFO, it would be useful to implement Moré and Wild performance profiles.