Some limits can be placed on the number of possible paths by considering some general properties of an optimal solution:
1. Assuming the start point is A, and the path leaves the start side at point B, the optimal path will take a straight line from A to B. Proof: if it were not straight, it could be replaced by a straight line which would shorten the path, contradicting the assumption the path was optimum.
2. Once the optimal path leaves a side, it will not return to that side. Proof. Suppose the path leaves the side at point B, then later enters at C, and then leaves again at D. The route from B to C could be replaced with a straight line from C to D, which would shorten the path, again contradicting the assumption that the path was optimum.
This cuts down the number of possible path templates greatly, since a given surface can only appear once in the optimal path, and only contain a single straight line segment.
I expect that a little more reasoning can deduce some more limits on possible optimal paths, to get it down to just two or three possible equations to maximize.
sqrt(1+x^2) + sqrt((6-x)^2+y^2)+sqrt(12^2+z^2)+
sqrt((30-z-y)^2+u^2)+sqrt((6-u)^2+1)
40 at {x -> 0.75, y -> 7, z -> 16, u -> 5.25}
http://www.wolframalpha.com/input/?i=minimize%5B%7Bsqrt%281%...your equation: http://www.wolframalpha.com/input/?i=minimize+2%28sqrt%281%2...
I ended up with the right answer by finding the minimum of the equation
which is the length of the path taken if the ant goes up to the ceiling at a point x to the East of straight up, and from there goes to the side wall at a point y to the south, and then goes from there to the center of that wall. (And then it's mirrored).I should have thought to unravel the room! Errr.