Stuck in local minimum & defining prior solutions with delay
• Hey @all,

I simulated some data and let Eureqa find the solutions. But often Eureqa gets stuck in a local mimimum and thus does not find the correct solution even after letting it run for a long time. After stopping and restarting the search Eureqa finds the correct solution. The problem is that I sometimes need to stop and restart the search a few times before Eureqa is on the right track.

Defining some Prior solutions seems to be a way out, but Eureqa does not accept prior solutions in the form of say: delay(a,1)+delay(b,1)+delay(c,1) if my target expression is in the for of say d=f(delay(a,1),delay(b,1),delay(c,1),delay(e,1)). It does accept them if my target expression is d=f(a,b,c,e) though.

Anyhow, since I don't know what the correct solution will be with non-simulated data I was wondering if anyone has some hints or ideas how to handle such an issue?

thanks in advance :)

 Tweet
• Hi holistic,

Have you seen this previous comment by Alison? It may help clear up some of your questions around defining prior solutions...
http://formulize.nutonian.com/forum/discussion/comment/740#Comment_740

- Jess
• Thank you, these explanations helped me out, works now :).

I my need I tried kind of a work around which also seems to find the correct solutions by requiring delay(a,1)+delay(b,1)+delay(c,1) to be in the solution. It it not the same as defining a prior solutions but maybe this is helpful for someone:

d = f(delay(a, 1), delay(b, 1), delay(c, 1), delay(e, 1)) + 0*require(contains(f(delay(a, 1), delay(b, 1), delay(c, 1), delay(e, 1)), delay(a, 1))) + 0*require(contains(f(delay(a, 1), delay(b, 1), delay(c, 1), delay(e, 1)), delay(b, 1))) + 0*require(contains(f(delay(a, 1), delay(b, 1), delay(c, 1), delay(e, 1)), delay(c, 1)))

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!