Question

Maximize a^2 + b^2 given a^3 + b^3 = 16

Original question: Let a, b > 0 such that a^3 + b^3 = 16. Find the maximum value of C = a^2 + b^2.

Expert Verified Solution

thumb_up100%(1 rated)

Key concept: This is a constrained optimization problem with a symmetric condition. That symmetry is a big clue: the maximum usually sits on the boundary where one variable takes most of the mass.

Step by step

We want to maximize

C=a2+b2C=a^2+b^2

subject to

a3+b3=16,a,b>0.a^3+b^3=16, \qquad a,b>0.

Because the constraint and objective are symmetric, we can use the power-mean idea or check endpoints via convexity.

Let x=a3x=a^3 and y=b3y=b^3. Then x+y=16x+y=16, and

C=x2/3+y2/3.C=x^{2/3}+y^{2/3}.

Since the function u2/3u^{2/3} is concave for u>0u>0, the sum x2/3+y2/3x^{2/3}+y^{2/3} is maximized when the variables are as unequal as possible, subject to positivity. So the maximum occurs at the boundary case where one variable tends to 0 and the other tends to 16.

Thus,

a316,b30a^3\to 16,\quad b^3\to 0

or vice versa. Then

Cmax=162/3=(24)2/3=28/3=443.C_{\max}=16^{2/3}=\bigl(2^4\bigr)^{2/3}=2^{8/3}=4\sqrt[3]{4}.

So the maximum value is

443.\boxed{4\sqrt[3]{4}}.

If you want a more formal argument with Lagrange multipliers, you get the same conclusion: the only interior critical point is a=b=2a=b=2, which gives C=8C=8, but the larger value is approached when one variable becomes very small and the other approaches 161/316^{1/3}.

Pitfall alert

A common slip is to assume the symmetric point a=ba=b must be the maximum. For concave powers like x2/3x^{2/3}, symmetry is often the minimum, not the maximum. Another trap is forgetting that the problem only requires a,b>0a,b>0, so the boundary can be approached even if b=0b=0 is not allowed exactly.

Try different conditions

If the problem had added a lower bound like a,b1a,b\ge 1, then the boundary argument would change. You would check the endpoints a=1a=1 or b=1b=1 and compare with the interior point a=b=2a=b=2. In that restricted version the maximum would come from one variable at the allowed minimum and the other adjusted by the cubic constraint.

Further reading

constrained optimization, Lagrange multipliers, power mean

chat