Open HenrikMettler opened 2 years ago
good catch that we can not just remove the configuration option for reorder while we support multiple rows.
how do you feel about removing the possibility of setting n_rows
and always use n_rows=1
? my impression is that modern general purpose CGP anyways uses exclusively only one row but may columns, while multiple rows were historically used for evolving circuits.
yeah, i think I agree. Should we also remove levels_back
? Atm I can't see any use case where anything but levels_back=n_columns
is useful.
yes, i think you're right, levels_back
can also be removed. :+1:
could you please create a separate PR, removing n_rows
and levels_back
? after that has been merged, we can come back to this one.
Moves the boolean
reorder_genome
into population to check whether genome conditions (n_rows=1, levels_back=n_columns) forreorder=True
are met. I am not removing the possibility to configure this, sincereorder
is limited to certain genome parameter configurations. Closes #347.