Galapagos Questions

I am new to Galapagos and would appreciate if anyone can direct me to answers for the following questions (I am trying to optimize 5-variables of a light shelf in terms of daylight availability):

1. As I 'reinstate' a a Galapagos run is there a way to update the slider positions AND the Grasshopper view? In the picture below I could not get the light shelf image to reflect the slider rotation values.

2. Any resource that explains A, B, C, D and E.

Replies to This Discussion

Permalink Reply by David Rutten on November 18, 2013 at 3:00pm

Hi Shamim,

A means that the solver found a better answer. Galapagos runs iteratively and it keeps on trying to find specific genomes that yield a higher fitness, whenever it manages to find such a genome there will be one of those little [+] symbols. Also, when it finds a new best-answer-yet the I'm-giving-up counter is reset to zero.

B is the average fitness of the entire population over time. It is not a particularly interesting statistic.

C represents the portion of the population that is fitter than a single standard deviation away from the average, and E represents the portion that is unfitter than one standard deviation. In a similar fashion, D represents that part of the population that is within one standard deviation of the average. None of these are particularly interesting from the user's point of view, but it does give you a sense about the general fitness variability within a population. I.e. "all genomes are quite fit but there are one or two slackers" vs. "all genomes are absolutely terrible save for a rare few" vs. "genomes are pretty well distributed along the fitness spectrum"

The vertical blue bar indicates that you currently have generation 17 selected. A 'population' of genomes evolves over time and every time-step is called a 'generation'. If all goes well, the fittest individuals in any specific generation are fitter than the fittest individuals from the previous generation. If this doesn't happen -say- 20 generations in a row, the solver will abort the search.

A single generation contains a fixed number of genomes or individuals. When you select a generation, those individuals will be displayed in the bottom three graphs. On the left you see a 'similarity representation' of this generation. The closer two dots are the more similar their genetic make-up. Black dots represent genomes with offspring, red crosses represent genomes that did not contribute to the next generation.

In the middle you see a multi-dimensional-point-graph. Each slider that is being manipulated by Galapagos is represented by a vertical line. Each genome is then drawn as a polyline connecting these vertical lines at the percentage of the slider value they all have. This representation shows not just clusters of similar genomes, it also shows you which slider layout they roughly have. You can select genomes in this graph.

On the right is a list of genomes (sorted from fittest to least fit) with the fitness value written next to it. The green bands are once again indicative of the slider layout of each genome, so if two capsules look alike, they have a similar slider layout.

David Rutten

david@mcneel.com

Tirol, Austria

Permalink Reply by Shamim Javed on November 19, 2013 at 7:25am

Hi David,

I really appreciate your taking the time for a comprehensive reply to my second question. As for the first question, the sliders and the Rhino model should both be updating on 'reinstating' a generation. I will keep on trying to achieve this.

Shamim.

Permalink Reply by Nick Tyrer on November 19, 2013 at 7:40am

I've never really considered what that bottom centre graph was doing. That description is actually really helpful. Thanks

Permalink Reply by David Rutten on November 19, 2013 at 8:11am

If the rotation sliders are the last two genes, then they seem to be maximized. It's difficult to see whether or not there's a bug here though without the file.

David Rutten

david@mcneel.com

Tirol, Austria

RSS