The difference between Galapagos and ParaGen

Replies to This Discussion

Permalink Reply by David Rutten on November 8, 2014 at 5:23am

Perhaps you can furnish us with a link to whatever ParaGen is supposed to be?

Permalink Reply by M.Elsayed on November 8, 2014 at 5:30am

ParaGon software developed by Peter Von Buelow (Michigan Univ.).

ParaGen is presented as a tool to support the exploration of the parametric design alternatives. ParaGen combines parametric modeling, performance simulation software and genetic algorithms, together with a database to store and retrieve the solutions for subsequent exploration. below is a journal article in details.

http://www.sciencedirect.com/science/article/pii/S1474034611000577

Permalink Reply by David Rutten on November 8, 2014 at 7:15am

Interesting, Rudi Stouffs was one of my thesis supervisors.

I don't think there's a fundamental difference between these two apps. They both seem to use GA (genetic algorithms/evolutionary solvers) to search potentially very large solution spaces.

ParaGen seems to have a more geometric focus, Galapagos is just a number cruncher, but it could be that it's just the examples that make it look that way.

Permalink Reply by M.Elsayed on November 8, 2014 at 9:09am

Good, so Galapagos will be enough for the moment.

Actually, I have a serious problem. In my thesis, I made a research plan to develop a prototype plan designed for housing projects dedicated to low income citizens. The idea was to make that prototype plan a parametric model so, if the user wants to change any item in the provided plan (by the state) for example enlarge his bed room or move the dining to view the garden, he can run the genetic algorithm to review the possible design solutions with respect to regulations without pay to a professional architect.

And now, I have a problem in the fitness function how the genetic algorithm will evaluate results to select generation over generation.

I have site constraints like set back and floor area ratio (FAR)...etc., and this is the easy part. The hard one is to find fitness function for design development.

All genetic algorithms based on evaluating engineering criteria (shortest distance, the largest area,…etc.). I need help in how to evaluate function or from where I can start my work. Any advice.

Permalink Reply by David Rutten on November 8, 2014 at 9:33am

Galapagos tries to minimize, maximize or optimize a single fitness value. It doesn't matter whether that value represents an area, a length, an amount of money or something else.

You're right in that the biggest challenge is often to formulate a proper fitness function that balances all your (potentially conflicting) partial goals. The easiest approach is usually to rewrite everything so that they are measured in the same unit. Money is a popular one because almost everything can be reduced to money. But on occasion you have to come up with another solution.