Calculus III: Absolute and Conditional Extrema

Finding the highest peak or lowest valley of a function is a fundamental problem in calculus with profound implications. In engineering, this translates to optimizing performance: minimizing material cost, maximizing heat dissipation, or finding the point of greatest structural stress. For functions of two or more variables, the search for these global extrema—the absolute maximum and minimum values over a given domain—requires a systematic strategy that blends the analysis of smooth interior points with the careful scrutiny of edges and boundaries.

Critical Points and the Extreme Value Theorem

The journey to find global extrema begins with understanding where they can possibly occur. For a function $f(x,y)$ of two variables (the concepts extend to higher dimensions), we first look for critical points in the interior of the domain. A critical point occurs where the gradient vector is zero or undefined. More precisely, a point $(a,b)$ is a critical point if $f_x(a,b)=0$ and $f_y(a,b)=0$, or if one or both of these partial derivatives do not exist.

At a critical point, the function may have a local maximum, a local minimum, or a saddle point. The Second Derivative Test helps classify these points using the discriminant $D=f_{xx}{f}_{yy}-(f_{xy}{)}^2$. However, local behavior is only part of the story. The Extreme Value Theorem for Multivariable Functions provides the crucial guarantee for finding absolute extrema. It states that if a function $f$ is continuous on a closed and bounded region $D$ in ${\mathbb{R}}^n$, then $f$ attains both an absolute maximum value and an absolute minimum value somewhere on $D$. This theorem is your license to hunt; it tells you the absolute extrema exist, and they must be located either at an interior critical point or somewhere on the boundary of the region.

Systematic Boundary Analysis: The Heart of the Method

For closed, bounded regions—like a disk, rectangle, or triangle—the Extreme Value Theorem applies, and a reliable, step-by-step method emerges. The absolute extrema are found by comparing all candidate values from two distinct sources.

1. Find all interior critical points. Compute the first partial derivatives $f_x$ and $f_y$, set them equal to zero (and check where they are undefined), and solve the system of equations. Only points that lie strictly inside the region $D$ are retained as candidates.
2. Find all boundary critical points. This is the most involved step. The boundary is a curve (or a set of curves/faces in higher dimensions). You must analyze the function's behavior on this constraint. There are two primary techniques:

• Parameterization: If the boundary is a simple curve (e.g., the edge of a circle or a line segment), express it with a parameter $t$. For example, if the boundary is the circle $x^2+y^2=1$, you could use $x=\cos t$, $y=\sin t$ for $0\le t\le 2\pi$. Substitute these into $f(x,y)$ to get a single-variable function $f(\cos t,\sin t)$, then find its critical points using standard Calculus I techniques by taking the derivative with respect to $t$.
• Constraint Equations (Lagrange Multipliers Preview): If parameterization is messy, you can treat the boundary equation as a constraint. While the full Lagrange method is discussed later, the principle is to find points on the boundary curve $g(x,y)=c$ where the gradient of $f$ is parallel to the gradient of $g$. This often involves solving a system derived from the boundary condition and a modified derivative equation.

1. Evaluate and Compare. Compute the function value $f(a,b)$ at every candidate point from Steps 1 and 2. The largest value is the absolute maximum; the smallest is the absolute minimum.

This comparison between interior and boundary values is the core of solving absolute extrema problems on closed, bounded domains.

From Theory to Practice: A Complete Worked Example

Consider an engineering design problem: the temperature distribution on a rectangular microprocessor chip is modeled by $T(x,y)=x^2+y^2-2x-4y$, where the chip occupies the closed rectangular region $D:0\le x\le 3,0\le y\le 2$. We need to find the hottest and coldest spots on the chip.

Step 1: Interior Critical Points. Find partial derivatives: $T_x=2x-2$, $T_y=2y-4$. Set to zero: $2x-2=0\Rightarrow x=1$; $2y-4=0\Rightarrow y=2$. The point $(1,2)$ is a critical point. However, we must check if it's in the interior. Our region has $y\le 2$, so $y=2$ lies on the top boundary, not in the interior. Therefore, there are no interior critical points. All extrema must be on the boundary.

Step 2: Boundary Analysis. The rectangle has four edges.

• Bottom Edge ($y=0,0\le x\le 3$): Substitute: $T(x,0)=x^2-2x$. This is a function of $x$ alone. Derivative: $2x-2=0\Rightarrow x=1$. Candidate: $(1,0)$. Also check endpoints: $(0,0)$ and $(3,0)$.
• Top Edge ($y=2,0\le x\le 3$): $T(x,2)=x^2+4-2x-8=x^2-2x-4$. Derivative: $2x-2=0\Rightarrow x=1$. Candidate: $(1,2)$ (this was our earlier critical point, now confirmed on the boundary). Endpoints: $(0,2)$ and $(3,2)$.
• Left Edge ($x=0,0\le y\le 2$): $T(0,y)=y^2-4y$. Derivative: $2y-4=0\Rightarrow y=2$. Candidate: $(0,2)$ (already listed). Endpoints: $(0,0)$ and $(0,2)$.
• Right Edge ($x=3,0\le y\le 2$): $T(3,y)=9+y^2-6-4y=y^2-4y+3$. Derivative: $2y-4=0\Rightarrow y=2$. Candidate: $(3,2)$ (already listed). Endpoints: $(3,0)$ and $(3,2)$.

Step 3: Evaluate and Compare. We have 6 unique candidate points:

• $T(1,0)=1-0-2-0=-1$
• $T(0,0)=0$
• $T(3,0)=9-6=3$
• $T(1,2)=1+4-2-8=-5$
• $T(0,2)=0+4-0-8=-4$
• $T(3,2)=9+4-6-8=-1$

The absolute maximum temperature is $3$ at $(3,0)$. The absolute minimum temperature is $-5$ at $(1,2)$.

Constrained Optimization: Lagrange Multipliers

Engineering design problems are often about optimization subject to a constraint. You need to maximize fuel efficiency given a fixed engine volume, or minimize material cost while maintaining a required strength. These are constrained optimization problems, solved elegantly by the method of Lagrange Multipliers.

To find the extreme values of a function $f(x,y)$ subject to a constraint $g(x,y)=k$ (where $k$ is a constant), you solve the system of equations derived from: $$\nabla f(x,y)=\lambda \nabla g(x,y)$$ $$g(x,y)=k$$ The scalar $\lambda$ is the Lagrange multiplier. Geometrically, this method finds points where the level curves of $f$ are tangent to the constraint curve $g(x,y)=k$, as this is where $f$ cannot increase or decrease without violating the constraint.

Engineering Example: Suppose you need to design a cylindrical container with a volume of $V=16\pi$ cubic meters. Find the dimensions (radius $r$ and height $h$) that minimize the surface area $S=2\pi r^2+2\pi rh$ (and thus material cost).

• Objective Function: Minimize $S(r,h)=2\pi r^2+2\pi rh$.
• Constraint: Volume $V=\pi r^{2}h=16\pi \Rightarrow g(r,h)=\pi r^{2}h=16\pi$, or simply $r^{2}h=16$.

Set up Lagrange equations:

1. $\frac{\partial S}{\partial r}=\lambda \frac{\partial g}{\partial r}\Rightarrow 4\pi r+2\pi h=\lambda (2\pi rh)$
2. $\frac{\partial S}{\partial h}=\lambda \frac{\partial g}{\partial h}\Rightarrow 2\pi r=\lambda (\pi r^2)$
3. $r^{2}h=16$

From equation (2), assuming $r\ne 0$, we get $2\pi r=\lambda \pi r^2\Rightarrow 2=\lambda r\Rightarrow \lambda =2/r$. Substitute into equation (1): $4\pi r+2\pi h=(2/r)(2\pi rh)=4\pi h$. Simplify: $4\pi r+2\pi h=4\pi h\Rightarrow 4\pi r=2\pi h\Rightarrow h=2r$. Substitute $h=2r$ into the constraint (3): $r^2(2r)=16\Rightarrow 2r^3=16\Rightarrow r^3=8\Rightarrow r=2$. Then $h=4$. The most material-efficient design has a radius of 2 meters and a height of 4 meters.

Common Pitfalls

1. Ignoring the Boundary: The most frequent and critical error is finding interior critical points and stopping. On a closed, bounded region, the absolute extrema often occur on the boundary, not at an interior point. Always perform a complete boundary analysis.
2. Misclassifying Boundary Points as Interior: When finding critical points from $f_x=0,f_y=0$, you must verify the point lies inside the domain, not on its edge. As seen in the worked example, the point $(1,2)$ satisfied the derivative conditions but was not an interior candidate for the region $0\le y\le 2$.
3. Forgetting Endpoints/Corners in Boundary Analysis: When parameterizing a boundary edge like a line segment, you create a single-variable function. You must apply the Closed Interval Method to it, which requires evaluating the function at the endpoints of the parameter's range. These endpoints are often the corners of a region and are frequent locations for absolute extrema.
4. Misapplying the Extreme Value Theorem: The theorem requires a continuous function on a closed and bounded region. If the region is unbounded (e.g., the entire $xy$-plane) or open (e.g., $x>0$), absolute extrema are not guaranteed to exist. You must use different techniques, like analyzing limits as variables go to infinity, to determine if global extrema exist in such cases.

Summary

• The Extreme Value Theorem guarantees absolute maxima and minima exist for continuous functions on closed, bounded regions.
• The systematic method involves finding and comparing all candidates from interior critical points ($f_x=0,f_y=0$) and boundary critical points (found via parameterization or constraint analysis).
• Constrained optimization problems, central to engineering design, are solved using the method of Lagrange Multipliers, which finds points where the gradient of the objective function is parallel to the gradient of the constraint.
• Avoid common mistakes by meticulously checking if points are interior vs. boundary, analyzing entire boundaries including corners, and verifying the preconditions of the theorems you use.