FHist.jl
FHist.jl copied to clipboard
Moelf
A pure Julia 1/2/3D histogram package that focus on speed and is thread-safe.
I worked on plotting the recipe and dispatch in `Plots.jl` using `Hist1D`. The recipe might below on this package - model with data: `WithData` object, and ```julia data = randn(1000);...
Closes #126 New `Plots.jl` recipe makes it easy to plot a curve over a histogram. There are two new functionality: - `plot(model::Function, WithData(h))`: to adjust normalization of the model taking...
Breaks with trying to plot the histogram with negative bin content. ```julia Random.seed!(123) h0 = Hist1D(rand(100); binedges=range(0,1,10)) h0.bincounts .-= mean(h0.bincounts) plot(h0) ``` throws, ```julia DomainError with -1.4545795461603218e29: sqrt was called...
```julia using FHist, CairoMakie begin h1 = Hist1D(randn(10000); binedges=-3:0.1:3) h2 = Hist1D(randn(10000), binedges=-3:0.1:3) fig = Figure() ax = Axis(fig[1,1], yscale=log10) stackedhist!(ax, [h1, h2]) current_figure() end ```  cc. @mfarrington1
I love this package, both the interface and the performances. I'd really love to have GPU support, at least for 1d histogram. Is this planned or can it be easily...
## The timing is distribution dependent ```julia julia> @benchmark Hist1D(x; binedges=range(-1,1;length=11)) setup=x=rand(10^6) BenchmarkTools.Trial: 1621 samples with 1 evaluation. Range (min … max): 1.418 ms … 2.619 ms ┊ GC (min...
I've noticed that `integral` is exported since `NumericalDistributions` [also exports `integral` ](https://github.com/mmikhasenko/NumericalDistributions.jl/blob/main/src/NumericalDistributions.jl#L34C1-L34C16)and they are in conflict. It would be good to find Julia standard `integral` interface and stick to it....