[FEATURE] use previous computation

[arita37]

Just wondering, if there is a way to use previous formulaes
as starting points for evolution ?

Mostly, load as initial population (vs starting from scratch every time...),
Thanks

[darioizzo]

You can feed them into the pagmo population as chromosomes?

[arita37]

Thx. Goal is to find a way to inject either previous computation as initial seed or some human formulae as initial seed to prevent from starting from scratch again every time. Thz

…

On Oct 7, 2022, at 20:15, Dario Izzo ***@***.***> wrote:  You can feed them into the pagmo population as chromosomes? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.

[darioizzo]

Yes, understood. At the end of an evolution you can save the best chromosome (i.e. a formula) of the population.

Then you can create at any point an empty population and push-back that chromosome in togheter with any other random or evolved formula.

At that point you can continue evolving.

[arita37]

Sure, thanks Is there any simple example available ? (where seed is custom injected). thanks

…

On Oct 7, 2022, at 23:52, Dario Izzo ***@***.***> wrote:  Yes, understood. At the end of an evolution you can save the best chromosome (i.e. a formula) of the population. Then you can create at any point an empty population and push-back that chromosome in togheter with any other random or evolved formula. At that point you can continue evolving. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.

[darioizzo]

Unfortunately not. Its going to be something like this ...

x = previosly computed seed
pop = pg.population(sr_prob,3) # creates a population with 3 random individuals
pop.push_back(x) # now the population has 4 individuals, one is the seed
pop = algo,evolve(pop)

but sr_prob must have the same dimensions and kernels as the one that created x.

As per your question on the possibility to use a human formula as seed ... there is no utility atm that creates a chromosome from a formula. This is not difficult per se, but there are infinite ways of doing so (i.e. infinite chromosomes that correspond to the same formula) so one would have to take some decisions there, which will also affect the evolution afterwards.