Use float with less precision for variables

[M1chV]

Hi!
I have a large amount of n_var and I want to have float type of this variables and by default it's np.float64. I was trying to setup vtype=np.float16 or vtype=np.float32, but I still have np.float64.
So, do we have solution for it, or I need to create own type like https://pymoo.org/customization/custom.html?
Thanks :)

[blankjul]

Currently, this is not really supported. Most operators use float which I think for most hardware maps to float64. However, I should also mention that no algorithm in pymoo is suitable for truly large-scale problems (n_var > 10,000 or so). There are EA variants that have been proposed to work more efficient though.