If I just need a chain of cartesian coordinates (as parameters/dimensions) that photon travels one after another until it reaches to a stop (hitting a wall or something), can GA minimize its path to get true ray tracing, if objective function returns total time of travel and whole 3D space is uniformly divided into nodes that a photon can travel?
Genetic algorithm is an optimization technique, which can be used to find minima of an N-dimensional function. However, this function needs to be more or less smooth and “nice”.
I assume the paths can diverge very fast (after a few reflections or refractions) even if their initial conditions (i.e. position and direction) are almost the same, hence the fitness function would rather be like a random function at least in case of certain inputs. For other inputs, it could work, but it seems to be a quite inefficient way of ray path calculations.