Does eatch iteration recomputing whole definition?

[Fariultra]

Hi, I was testing a few days ago tunny and experiencing significantly slower iterations compared to the Galapagos. My definition is quite demanding due to geometrical operations. To make it faster I preprocessed a couple of operations before being affected by gene pool variables. I got the impression that Tunny recomputing a whole definition every iteration, is it the case?

[hrntsm]

Hi @Fariultra,

Thank you for your interest.
I am aware that the problem is that the calculations are slower than Galapagos because everything is recalculated as you mentioned.
Do you know of a better solution?

[Fariultra]

Hi @hrntsm,

I forgot to thank you for your work in the first place. I assume that the recalculation of only components after variables is not simply doable? Somehow I am unable to explain so tried to draw:

The current state is the RED lines?

[hrntsm]

Hi @Fariultra, Thanks your drawing.

as you mentioned current is red line.
I will try to change tunny operation to yellow and blue line.

[hrntsm]

Hi @Fariultra
I fixed tunny to your blue line.

Could you test below version?

https://github.com/hrntsm/Tunny/actions/runs/8323613670/artifacts/1334399744

[hrntsm]

Fix this in v0.12