add log safe Makie #116
add log safe Makie #116
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #116 +/- ##
==========================================
+ Coverage 80.37% 80.74% +0.36%
==========================================
Files 11 12 +1
Lines 790 805 +15
==========================================
+ Hits 635 650 +15
Misses 155 155 ☔ View full report in Codecov by Sentry. |
|
🤩, thanks a bunch, will take a look after some meetings. |
ext/FHistMakieExt.jl
Outdated
| const SAFE_LOG = Ref(true) | ||
|
|
||
| """ | ||
| set_safe_log(x::Bool) | ||
|
|
||
| Turn on/off SAFE_LOG, to plot histograms without/with negative values. By default is `true`. | ||
| """ | ||
| function set_safe_log(x::Bool) |
There was a problem hiding this comment.
I don't think we can add new things into the namespace via Pkg Extension mechanism.
julia> using Revise, CairoMakie, FHist
^LPrecompiling FHistMakieExt
1 dependency successfully precompiled in 4 seconds. 222 already precompiled.
julia> FHist.SAFE_LOG
ERROR: UndefVarError: `SAFE_LOG` not defined
Stacktrace:
[1] getproperty(x::Module, f::Symbol)
@ Base ./Base.jl:31
[2] top-level scope
@ REPL[2]:1
julia> FHist.set_safe_log
ERROR: UndefVarError: `set_safe_log` not defined
Stacktrace:
[1] getproperty(x::Module, f::Symbol)
@ Base ./Base.jl:31
[2] top-level scope
@ REPL[3]:1I will see if we can resolve this using plotting arguments
There was a problem hiding this comment.
My bad, I just tested that the clipping was doing the right thing, but haven't check that the switch was actually working...
There was a problem hiding this comment.
yeah no worries, this is more than enough! I will try to make it nicer to change for users!
|
I did try to make this a bit more clean. Now everything is in a submodule This is roughly how it should be working: julia> h = Hist1D(bincounts=[0.5,-0.5],binedges=[1,2,3],sumw2=[1.01,1.01])
julia> f, ax, p = stairs(h)
julia> errorbars!(ax,h)gives: While setting log safe to true, clips to:
And plotting in log scale does not throw errors! 🥳 |
|
Personally not a fan of global options in packages if we can avoid them. |
I can see that, I tried to put everything in a submodule to make "safer" and make the global value a reference. I'm not sure how to do this otherwise. But having such a feature would be very useful, but indeed it should be not the default behaviour. |
| el_vec[mask] = c_vec[mask] .- min_positive | ||
|
|
||
| # clip higher errors | ||
| @. eh_vec = max(eh_vec - (c_vec - c_vec_def), min_positive) |
There was a problem hiding this comment.
The logic here is that if you camp bin counts, you need to adjust also the upper error accordingly. E.g. your data point is y = - 0.1 σ = 0.2. If you clip the central value to y' = 0, the upper error should be set to σ' = σ - (y' - y) = 0.1. Now if central value is very low or error is very small, you could get σ' < 0, and therefore you take the max between that and min_positive value. Does that make sense?
There was a problem hiding this comment.
All this is in the spirit that even if you're clamping stuff to look at log scales, it could still be interesting information that there's a data point right below the 0, but with an error big enough that makes it compatible with a positive value, and up to which positive value the error extends. E. g. the second plot here.
|
done in #118 |
This is an attempt at making the MakieExt safe for plotting in log scale. But keeping the option to disable such a feature, and have histograms with negative values. Tagging @Moelf 😍