Talk-Polywell.org

Has anyone else seen this? It's called an "Advanced Inertial Confinement Nuclear Fusion Reactor", but the students elsewhere describe it as "electrostatic", and it's clearly some kind of IEC scheme.

Their claim is that they've been able to use "genetic algorithms" to discover novel and optimal reactor designs with much fewer losses than conventional IEC designs. Their web page is here. Not a whole lot of detail to be found. They're asking for money to fund an improved reactor. There's a breakdown of what they need on their Kickstarter page.

Genetic algorithms need a good numeric model of whatever system you're trying to optimize. Absence of any detail on the model or results has me wondering.

Okay, I found these threads on the fusor.net forum, which appear to be relevant:

Simion and Fusor Simulation
Building and Testing a Grid Idea.
Not quite getting full plasma Ignition.
Assuming Extreme Grid Transparency, what's the next issue?
New Kickstarter Project, working with genetic algorithms.

hanelyp wrote:Genetic algorithms need a good numeric model of whatever system you're trying to optimize. Absence of any detail on the model or results has me wondering.

Right, specifically the fitness evaluation function. Must have a way to score how well each solution in a population of solutions does as compared to others in each generation. IMO, this would require something like a very good polywell simulation of some sort. Unless the genetic algorithm is being used on some sub-component.