Why Locate Extrema Numerically?
Graphs can make the shape of a function visible, but a graph alone rarely gives a precise extremum location. A calculator’s numerical maximum or minimum command can estimate the input \(x\) and output \(f(x)\) where a high or low point occurs. The estimate is useful evidence, not a guarantee that the point is a local extremum.
The key is to understand what the tool searches. It works within a specified interval, and it reports a high or low value it finds there. If the search interval cuts off the graph, the reported point might be a boundary of that interval rather than an interior turning point. After using the tool, inspect nearby values and the graph’s behavior around the result.
Local Extrema and the Search Interval
Let \(f\) be defined on a domain containing \(c\). The point \(c\) is a local maximum input if, for all \(x\) sufficiently close to \(c\), \(f(c)\geq f(x)\). It is a local minimum input if, for all \(x\) sufficiently close to \(c\), \(f(c)\leq f(x)\). The corresponding function values \(f(c)\) are called local maximum and local minimum values.
In this tutorial, local extrema mean interior turning points: there must be domain values on both sides of \(c\) to compare. An endpoint can be the highest or lowest point on a specified interval without being an interior local extremum. That distinction matters because a calculator’s search interval also has endpoints.
A graphing calculator may label its command “maximum” or “minimum,” but the command does not determine the mathematical context for you. You select an interval or provide left and right bounds; the calculator then estimates a high or low value in that region. On some devices, the command starts from a cursor position and uses nearby points to define its search. In either case, the search bounds and displayed graph window affect what it can report.
A Reliable Calculator Procedure
Choose a window that shows the curve on both sides of the suspected high or low point. A window that cuts off the curve can hide the evidence needed to recognize an interior extremum.
Use bounds on either side of the suspected feature. Avoid placing a bound directly at the apparent turning point, and do not use an interval so wide that another extremum could confuse the search.
Record both coordinates, \((x,f(x))\), and note the precision displayed. Treat the coordinates as estimates unless the function or a separate argument establishes exact values.
Evaluate the function at inputs slightly smaller and larger than the reported input. Check whether the point is interior to the domain and search interval, and whether the nearby values support a local maximum or minimum.
Nearby values do not replace a proof when a problem asks for exact justification. They do provide an important check on what the calculator has found. If the values increase on both sides, for instance, the reported point is not behaving like a local maximum.
Why a Differentiable Interior Extremum Has a Horizontal Tangent
There is a useful calculus reason to expect the derivative to be zero at a smooth interior extremum. This result can help interpret what a graphing tool locates, although a maximum/minimum command itself does not require you to calculate a derivative.
For a local maximum, nearby inputs to the right satisfy \(f(c+h)-f(c)\leq 0\) when \(h>0\). Dividing by positive \(h\) gives a difference quotient at most zero. For nearby inputs to the left, \(h<0\), and the same numerator is nonpositive, so the quotient is at least zero. If the derivative exists, both one-sided limits equal the same number. That number must be both nonpositive and nonnegative, hence zero.
For a local minimum the inequalities reverse: the quotient is nonnegative from the right and nonpositive from the left, so the derivative is again zero if it exists. The conditions matter. A local extremum can occur where the derivative does not exist, and a point with derivative zero need not be an extremum. A graphing tool’s estimate should therefore be interpreted alongside the graph and nearby values, not as a standalone proof.
Worked Example: Locating a Smooth Minimum
Worked Example: Locating a Smooth Minimum
Use a numerical minimum command to estimate the minimum of \(f(x)=(x-2)^2+1\) near \(x=2\), and check the result with nearby values.
Choose a graph window that includes inputs on both sides of \(2\), then search, for example, on \([1,3]\). A calculator may report approximately \(x=2.000\) and \(f(x)=1.000\). The function value at the reported input is
Check one input on each side. With \(x=1.9\),
With \(x=2.1\),
Both nearby values exceed \(f(2)=1\), matching the behavior of a local minimum. The graph and numerical estimate are consistent with the exact point \((2,1)\). The calculator’s displayed decimals are estimates; here, the formula itself verifies the exact coordinates.
Worked Example: Searching for Local Extrema of a Cubic
Worked Example: Searching for Local Extrema of a Cubic
For \(f(x)=x^3-3x\), use focused searches to estimate a local maximum near \(x=-1\) and a local minimum near \(x=1\). These should be treated as two separate searches so that each one isolates a single turning point.
For the maximum, search on \([-1.5,-0.5]\). A calculator may report approximately \(x=-1.000\) and \(f(x)=2.000\). Substitution gives
Check inputs on both sides. At \(x=-1.1\),
At \(x=-0.9\),
Both are less than \(2\), supporting a local maximum at \((-1,2)\). For the minimum, search separately on \([0.5,1.5]\). A calculator may report approximately \(x=1.000\) and \(f(x)=-2.000\), and substitution confirms
The nearby values are
Each is greater than \(-2\), as expected near a local minimum. The two estimates and nearby-value checks agree with the exact points \((-1,2)\) and \((1,-2)\).
Worked Example: When the Tool Reports a Search-Boundary Minimum
Worked Example: When the Tool Reports a Search-Boundary Minimum
Let \(f(x)=x^2\), defined for all real \(x\), and suppose the graph window and minimum search are restricted to \([1,3]\). A minimum command may return the left boundary, approximately \(x=1\), with \(f(x)=1\). This is indeed the smallest value on the restricted interval: for every \(x\in[1,3]\), \(x^2\geq1\).
However, the command’s result is not a local minimum of the function on its full domain. Check nearby inputs on both sides of \(1\):
Since \(0.81<1\), values just to the left of \(1\) are lower than \(f(1)\). Thus \(x=1\) is not a local minimum on the full domain. The true local minimum is at \(x=0\), where \(f(0)=0^2=0\), but \(x=0\) lies outside the search interval and the displayed window.
Common mistake: “The calculator says the minimum is \(x=1\), so \(x=1\) is a local minimum.” This conclusion confuses the minimum within \([1,3]\) with a local extremum of the function. The correction is to inspect values on both sides in the function’s domain and to check whether the reported point is interior to the search interval. In this case, the nearby value \(f(0.9)=0.81<1=f(1)\) disproves the local-minimum claim.
Reading the Result Carefully
When interpreting a numerical extremum, keep three locations distinct: the function’s domain, the graphing window, and the interval searched by the calculator. The domain says which inputs are mathematically allowed. The graph window says which part is visible. The search interval says where the numerical command looks. These regions may not be the same.
| What the result suggests | What to check |
|---|---|
| An interior local maximum or minimum | Compare nearby values on both sides; confirm the point is interior to the domain and search interval. |
| A high or low value at a search bound | Determine whether the tool stopped at a bound rather than finding an interior turning point. |
| A feature near the edge of the visible graph | Expand or shift the window so the curve can be inspected on both sides. |
| A rounded coordinate from the calculator | Retain appropriate precision, and use substitution or other mathematical reasoning if an exact answer is required. |
Numerical maximum and minimum tools are most useful when the search region is chosen deliberately and the result is checked. A displayed point is a prompt to investigate: evaluate the function nearby, look at the curve on both sides, and decide whether the point is a genuine interior local extremum, a domain endpoint, or simply the edge of the chosen search region.
Check Your Understanding
For each question, explain what the calculator’s result does and does not establish.
- A minimum command reports \(x=3\) on a search interval \([3,5]\). What should you check before calling \(x=3\) a local minimum?
- A graph appears to have a local maximum near \(x=2\). What is one useful choice for the calculator’s search interval, and why?
- For \(g(x)=(x+1)^2\), a tool estimates a minimum at \(x=-1.001\). What nearby function values could you check to assess the estimate?
- Can a point where \(f'(x)=0\) fail to be a local extremum? Explain why a numerical search should not be treated as proof by itself.
- How can a graph window hide the information needed to decide whether a reported point is an interior extremum?
