search

wujian16 / Cornell-MOE

A Python library for the state-of-the-art Bayesian optimization algorithms, with the core implemented in C++.
Other
261 stars 64 forks source link

[Question] Appropriate way to define search domain parametric curves/surface meshes as input domain #66

Closed jerdra closed 5 years ago

jerdra commented 5 years ago

Apologies in advance if this is the wrong place to post questions.

I'm interested in applying this package on some data where the input data lies on a simple curved discrete surface (2D triangulated mesh embedded in 3D space), i have two questions with respect to this application:

  1. Does the package handle discrete search domains, if so, how?
  2. If that software does not handle discrete domains. Suppose I were to approximate my discrete surface with some bounded polynomial surface such that it is continuous, are there ways to specify this constraint (data must lie on polynomial plane)?

I'm a little wary of using a bounding box search space because the true theoretical optimal solution is outside the parametric surface I've defined but it is unfeasible physically.

jerdra commented 5 years ago

closing since I found a workaround for my case