I tried the suggestion at http://www.mapleprimes.com/forum/optimizing and added the following code AreaGrad:=proc(V,G) global CoverArea13: local x,y,z: G[1] := fdiff('CoverArea13'(x,y,z),x,[x=V[1],y=V[2],z=V[3]]); G[2] := fdiff('CoverArea13'(x,y,z),y,[x=V[1],y=V[2],z=V[3]]); G[3] := fdiff('CoverArea13'(x,y,z),z,[x=V[1],y=V[2],z=V[3]]); return NULL: end proc: NLPSolve(CoverArea13, 0..Pi,-2..2, -2..2,objectivegradient=AreaGrad); But I got: Error, (in Optimization:-NLPSolve) no improved point could be found Also when I try NLPSolve(CoverArea13, objectivegradient=AreaGrad,method=nonlinearsimplex); I get the error: Error, (in Optimization:-NLPSolve) The three given points are AreCollinear Which I not sure what that means. Can anyone lend some insight?