I have an idea! But how do I realize it with GH?

Hey all,

I just had a look at CPPN's after looking at one of Robert Vier's videos. Actually I didn't know about CPPN's and whoever does not know about it, too, they are "Compositional pattern-producing networks".

But why should you get to know them and what benefit could they give to your projects?

I asked myself these questions several times and postponed some research on i.e. physic systems (Kangaroo), graph networks (Spider Web) or evolutinary solver or other stuff in the past. Because most of the time the answer was the oposite:

What kind of knowledge about algorithms, plugins etc. do I need to realize my project?

I often had to search for the answers quite long. And if I would have known about some of the above mentioned a bit more detailed, I could have found a solution much faster.

Technical terms often make it even more complicated. Surface Relaxation is a common term used for the production of tensile structures? Yeah most Kangaroo users know it, but think of even more difficult things like the one mentioned in the first paragraph!

Wouldn't it be nice and useful to have something like an abstract database about what kind of algorithms to use for what kind of projects? I haven't thought so much about how.

Maybe it is just a library of technical terms (yeah...the wrong way around), you click on them and see who is doing research on them and with it also an application. These technical terms can then be connected to other terms. A bit like academic work, referring to other papers and also giving advice on where research is missing.

If we want to bring algorithmic design forward, it is probably easier having a nice sharing culture. We could maybe find concepts for a use of neural networks or the generation of floorplans or maybe even combine them?!

Probably I am a bit naive now and it will be a problem to find a good concept, realize a database and also gather the information on projects, but if you like it, just start thinking and share it. I would like to hear some feedback :-)

Greetings,
Martin

Replies to This Discussion

Permalink Reply by David Rutten on February 1, 2015 at 3:51pm

Algorithms range from the positively atomic to the absolutely leviathan. Typically algorithms are made up of smaller algorithms, strung together in the right order. It is a very big topic, and although I can definitely see how it would be nice to have a 'dictionary of algorithms', it's sort of hard to see how it would help those who don't know any names. After all, the database would have to be rather large to be of any kind of use, but then what do you search for? "Relaxation" and "Force Density" are words you already have to know. If you're just going to search for "tensile form finding and algorithm"... google does a pretty good job of that already.

I suppose what would be nice is to have a cross referenced collection of algorithm descriptions, all written in the same style and aimed at the same audience (novice and beginner programmers, without a maths doctorate).

It's actually not a bad idea to include a list of a couple dozen/hundred descriptions in the help files for GH2. We may get around to that.

Permalink Reply by Martin D. on February 2, 2015 at 5:53am

The idea just came to my mind yesterday and I directly shared it. I have thought about it a bit longer afterwards and actually quite often the forum does a good job. Not always, but most of the times you get the right ideas.

However I like the idea for GH2. If you find the time for that, it would be great! Maybe it is nice if users can add their own descriptions in the same format or something like a link/help box in their components, referring to their blogposts, like you do in the Galapagos component.

RSS