Byron S. answered 11/16/14
Tutor
5.0
(44)
Math and Science Tutor with an Engineering Background
Hi Mandy,
You can solve this problem similarly to the ellipsoid problem I posted an answer to earlier, using Lagrange multipliers. In this case, you're trying to optimize f(x,y,z) = xy + xz with two constraints g(x,y,z) = x2 + y2 = 1 and h(x,y,z) = xz = 1. Your Lagrange equation is then
∇f = λ ∇g + μ ∇h
Again, the partial derivatives are fairly simple, and you get the three equations
∂x: y + z = 2x λ + z μ
∂y: x = 2y λ + 0 μ
∂z: x = 0 λ + x μ
From the third equation, either x = 0 or μ = 1. x ≠ 0 since xz = 1. Therefore μ = 1.
The first equation then simplifies to
y + z = 2x λ + z
y = 2x λ
Substitute this into the second equation x = 2y λ in place of y, and you get
x = 2(2x λ) λ
x = 4x λ2
Since x≠0, divide by x and
1 = 4 λ2
λ = ±1/2
Plug these values into either the first or second equations and you find that either
x = y or x = -y (and in both cases, x2 = y2)
Since x2 + y2 = 1
2 x2 = 1
x = ±1/√(2)
y = ±1/√(2)
The final condition is xz = 1. Plug in each value for x, and you get
z = ±√(2) where x and z are either both positive or both negative.
These combine to give 4 critical values to test
(1/√(2), 1/√(2), √(2))
(1/√(2), -1/√(2), √(2))
(1/√(2), -1/√(2), √(2))
(-1/√(2), 1/√(2), -√(2))
(-1/√(2), -1/√(2), -√(2))
(-1/√(2), -1/√(2), -√(2))
Plug each of these into f(x,y,z) to determine which are the maximum and which are the minimum values.