numbbo / coco

Numerical Black-Box Optimization Benchmarking Framework
https://numbbo.github.io/coco
Other
262 stars 87 forks source link

Distinguish names (here for DD-CMA-ES-... data sets) #2134

Open nikohansen opened 2 years ago

nikohansen commented 2 years ago
import cocopp

cocopp.main('bbob/20.*/.*Ghar')

Gives currently for example this figure: pprldmany_default

with two line with indistinguishable names DD-CMA-E, because the font size is adapted to be large when the number of lines is small. In this case, the problem is also not solved by prepending the year, because the data are from the same year. We could prepend a number, but this would not quite solve the problem to identify which is which (accept that we know in principle the order in which we used the data sets as input to cocopp).

ttusar commented 2 years ago

In addition, using a dark background, this is what I see: image

The issue is that the background is transparent instead of white. A possible fix is to do set fig.patch.set_alpha(1.0) right before the fig.savefig (from https://stackoverflow.com/questions/71010292/cannot-change-background-color-of-matplotlib-figure-when-saving-as-image).

nikohansen commented 2 years ago

Possible solutions:

ttusar commented 2 years ago

I vote for the first one :-)