Homework #2: Problem 5 Here is sample output from the newt3 program, first an error from incorrect input, and then the results of the suggested starting points for (x,y,z): einstein 7> newt3 usage: newt3 [] einstein 8> newt3 1.0 -1.0 -0.25 Iter x y z 1 1.4278066245790060E+00 -9.5197155296796876E-01 -1.9232255593430930E-01 2 1.3977664333923043E+00 -9.8543399384905372E-01 -1.7416588474639941E-01 3 1.3977994333164945E+00 -9.8469310727442372E-01 -1.7481060709965163E-01 4 1.3977994362977424E+00 -9.8469258308486018E-01 -1.7481118217176370E-01 5 1.3977994362977610E+00 -9.8469258308462249E-01 -1.7481118217209335E-01 1.397799436297761 -0.9846925830846225 -0.1748111821720933 Thus, the guess is a good one, and the method finds a convergence after 5 iterations. I then altered each input (keeping the other two fixed at the above guesses) to find the "basin of attraction". The (x,y,z) inputs were increased and decreased until a final point (within +/-0.0001) was reached were the program could no longer converge within 50 iterations: Minimum Default Maximum x -2.2291 1.0 2.4907 y -4.1362 -1.0 0.6747 z -4.8617 -0.25 0.6135 To check the newt3 results, I modified the newt2 Maple calls from class notes to solve the 3-dimensional case: check3 := proc() local f1,f2,f3,r1,r2,r3,ans; f1 := x^2 + y^3 + z^4 - 1; f2 := sin(x*y*z) - x - y - z; f3 := cos(x) - y*z; ans := fsolve({f1,f2,f3},{x,y,z}); r1 := evalf(subs(ans,f1)); r2 := evalf(subs(ans,f2)); r3 := evalf(subs(ans,f3)); print(f1); print(f2); print(f3); print(ans); print(r1); print(r2); print(r3); end: which gave the following output: bytes used=1003632, alloc=786288, time=0.76 bytes used=2003796, alloc=1310480, time=1.65 bytes used=3005824, alloc=1441528, time=2.60 2 3 4 x + y + z - 1 sin(x y z) - x - y - z cos(x) - y z {x = 1.397799436, z = -.1748111822, y = -.9846925831} -8 -.11*10 -9 .2*10 -9 .2*10 Thus, Maple agrees with newt3 to the precision of the returned (r1,r2,r3).