algorithmic modeling for Rhino
Many thanks for joining the group! Hope you've had a chance to have a quick play with the tool. As always, it would be great to hear from you so please drop a comment below or anywhere on the group page to let us know your experiences - it has definitely helped development direction so far.
Pleased to say we have another release, 0.3.0 available on github here.
The major update is the incorporation of performance based selection for multiple criteria. This can occur during an evolution run interchangeably with artificial selection. Simply hit the radio buttons listed against each performance criteria to overrule the artificial selection (yes, having both at the same time in future would be nice!). A tooltip will show you which button does which, because you may wish to maximise and minimise certain criteria at the same time:
This 'classic' genetic algorithm utilises the whole population (say 100 designs) during selection, so after you hit the evolve button it will probably take longer as all phenotypes will be calculated. Essentially, this is similar to Galapagos except you can use artificial and natural selection interchangeably, and the history canvas of course.
In addition to this, the history canvas now displays performance values, population averages, performance-based generation results and designs may now be double-clicked to display in Rhino:
Finally, a new component output gives a datatree of parameter states. This is not terribly useful at present, so we are next going to work on storing the instance data in a better way, so when the window is closed your data is not lost.
We hope this is useful and many thanks for your support,
John & Cecilie.