|Thomas v. Girsewald
Architectural Association School of Architecture
Emergent Technologies + Design 2004-2005
06. February 2005
Genetic mapping algorithms of the Backus-Naur Form specification in GENR8
|The exercise takes an accelerated look at the genotype, probing the direct behavioural
effects on an implemented phenotype. It concentrates specifically on the contributions
of Martin Hemberg to the organic growth model language. The ambition was to manipulate
and apply Hemberg Extended Map Lsystems (HEMLS) within GENR8, a sophisticated plug-in
for Alias Maya; developed and hard-coded by Mr Hemberg in 2001 as a collaborator with
the Artificial Intelligence Lab and Emergent Design Group at the Massachusetts Institute
of Technology. Martin Hemberg is currently a tutor for the Emergent Design and Technologies
postgraduate programme at the Architectural Association, London as well as a PhD student in
the Biomolecular Science programme at the Department of Chemistry and Biochemistry, Imperial
First identified are the key advances of HEMLS in evolutionary computation. Variables, production rules and operator values in re-write systems are defined. Next, I classified predefined and user-defined grammars graphically within a one step fixed constant scaffold, empty environment growth model.
In a series of separate procedures isolated within a five step, fixed parameter empty environment growth model, interpreted genome mapped multi-variable time dependant and stochastic production grammars are executed.
These optimized fixed-constant user-defined bnf specifications are then re-mapped to a surface in a formally manipulated environment. The results of these bottom-up processed reactive growth models are analyzed and evaluated; to be further developed though genr8’s grammatical evolution algorithm.