Why Numerical Root Solving Needs a Check
Some equations cannot be solved conveniently by isolating the variable with algebra. A graphing calculator can estimate a solution, but the displayed decimal is only an approximation. It may depend on the interval selected, the calculator’s numerical precision, and how the equation was entered.
A reliable process has two parts: use a numerical tool to find a candidate, then test whether that candidate is consistent with the equation. For AP Calculus work, this means identifying the equation being solved, recording an appropriate approximation, and checking the result rather than treating a calculator display as self-validating.
Set Up the Equation Before Using a Calculator
For an equation such as \(P(x)=Q(x)\), define a single function by subtracting the right side from the left: \(f(x)=P(x)-Q(x)\). Then the original equation is equivalent to \(f(x)=0\). This step matters because a calculator’s zero-finding tool operates on a function, and a sign error in the subtraction changes the problem.
When more than one solution may exist, the interval is part of the setup. A graph or table can help locate a region of interest, but a calculator finding one zero does not establish that it found every zero. The selected interval should correspond to the solution requested by the question.
Move all terms to one side and write the equation as \(f(x)=0\).
Use the stated domain, graph, or a table to identify where the intended root is located.
Depending on the device, this may be named zero, root, or solve. Enter the function and an interval or starting values as required.
Substitute the reported decimal into the zero function and inspect nearby values when useful.
Why a Sign Change Can Locate a Root
A continuous function cannot move from a negative value to a positive value without taking the value zero in between. This is the Intermediate Value Theorem: if \(f\) is continuous on \([a,b]\), and a value \(L\) lies between \(f(a)\) and \(f(b)\), then \(f(c)=L\) for at least one \(c\in[a,b]\). In particular, if the endpoint values have opposite signs, there is at least one root between them.
This result justifies a useful numerical method called bisection. Start with an interval whose endpoint function values have opposite signs. Evaluate the function at the midpoint, then keep the half-interval whose endpoints still have opposite signs. Repeat. Each step preserves a bracket for at least one root, and halves the interval width.
Choose \(a\) and \(b\) so that \(f(a)f(b)<0\), with \(f\) continuous on \([a,b]\).
Let \(m=(a+b)/2\) and calculate \(f(m)\).
If \(f(m)=0\), stop: \(m\) is an exact root. Otherwise, if \(f(a)f(m)<0\), replace \(b\) by \(m\); otherwise, if \(f(m)f(b)<0\), replace \(a\) by \(m\).
After \(n\) halvings, the bracket width is \((b-a)/2^n\), where \(a,b\) are the original endpoints. Its midpoint is an approximation to a root in the bracket.
The width formula follows because each bisection step divides the current interval length by \(2\). Thus, after \(n\) steps, the length is the original length multiplied by \(1/2\), \(n\) times. Calculator root tools often use more advanced numerical methods, but a sign-changing bracket remains a valuable independent check.
Worked Example: A Polynomial Root and a Residual Check
Worked Example: A Polynomial Root and a Residual Check
Estimate a solution of \(x^3-x-1=0\), and check the calculator’s reported value \(c=1.324718\).
Define \(f(x)=x^3-x-1\). First, bracket a root. At \(x=1\),
At \(x=2\),
The function is a polynomial, so it is continuous. Since \(f(1)<0\) and \(f(2)>0\), the Intermediate Value Theorem guarantees at least one root between \(1\) and \(2\). A calculator’s zero tool, applied on that interval, reports approximately \(1.324718\).
Now substitute the candidate. The cube is approximately \((1.324718)^3=2.324718182335474\), so
The residual is about \(1.82\times10^{-7}\), which is very small. It is not exactly zero at the displayed precision, so \(1.324718\) should be reported as an approximation, not an exact root.
As a second check, evaluate on either side of the candidate. For example,
These opposite signs show that a root lies between \(1.324\) and \(1.325\). The calculator candidate lies in that interval. The residual check tests closeness at the candidate; the nearby-value check places a sign-changing bracket around it.
Worked Example: Selecting Between Two Polynomial Roots
Worked Example: Selecting Between Two Polynomial Roots
Use a calculator to estimate solutions of \(x^2=5\), and identify the positive solution to four decimal places.
Rewrite the equation as \(f(x)=x^2-5=0\). Since the equation has both a positive and a negative solution, an interval is needed to target the positive one. On \([2,3]\),
The function is continuous and the endpoint values have opposite signs, so there is a root in \((2,3)\). A calculator’s zero tool on that interval gives approximately \(2.2360679\), which rounds to \(2.2361\) to four decimal places.
Check the rounded value by substitution:
The residual is small, consistent with the four-decimal approximation. Nearby values give an additional check:
The opposite signs bracket a root between \(2.23\) and \(2.24\), and the reported positive candidate is in that interval. The original equation also has a negative solution; an interval such as \([-3,-2]\) would target it instead. The positive interval is what makes this calculator result the requested solution.
Worked Example: A Trigonometric Equation
Worked Example: A Trigonometric Equation
Estimate the solution of \(\cos x=x\) in radians near \(0.74\), and check the result.
Define \(g(x)=\cos x-x\), so the equation becomes \(g(x)=0\). Because the problem specifies radians, the calculator must be in radian mode. On the interval from \(0.739\) to \(0.740\), approximate evaluations give
The signs are opposite, and \(g\) is continuous, so the Intermediate Value Theorem guarantees a solution between these values. A calculator root tool reports approximately \(c=0.7391\). Substitution gives
This small residual is consistent with the reported approximation. The nearby values provide a bracket for a root, while the residual measures how closely the rounded candidate satisfies the equation. If the calculator were in degree mode, it would evaluate a different cosine function and the result would not check this radian equation.
What the Checks Do—and Do Not—Establish
Substitution and nearby values answer related but different questions. The residual \(f(c)\) measures how close the candidate \(c\) is to satisfying \(f(x)=0\) in the function’s output. A sign change in nearby values supports the existence of a root in the interval between them, provided the function is continuous there.
| Check | What it supports | What it does not guarantee |
|---|---|---|
| Substitute \(c\) and compute \(f(c)\) | The candidate has a small residual at the displayed precision. | That \(c\) is an exact root, or that no other roots exist. |
| Evaluate \(f(a)\) and \(f(b)\) with opposite signs | At least one root lies between \(a\) and \(b\), if \(f\) is continuous there. | That there is only one root in the interval. |
| Use a graph or calculator root tool | A candidate location for a root. | That every root has been found or that the input and mode were correct. |
A root may not change sign. For example, \(f(x)=x^2\) has a root at \(x=0\), but \(f(x)\geq 0\) on both sides of zero. Thus, the absence of a sign change does not prove there is no root; a graph, substitution, or a different interval strategy may be needed. Similarly, a small residual is useful but its meaning depends on the scale of the function and the requested precision.
Common Pitfall: Treating a Displayed Decimal as Exact
A common error is to write that a rounded candidate is a root merely because the calculator displays it under a zero or root command. For the polynomial in the first example, an incorrect check might say, “\(1.324718\) is a root because the calculator returned \(1.324718\).” That skips the substitution and overstates what the display establishes.
The correction is to compute the residual with the same function used by the calculator: \(f(1.324718)\approx 0.000000182335474\). This value is small, but not zero. The careful conclusion is that \(1.324718\) is a numerical approximation to a root. Nearby function values can then add evidence by bracketing a root, as long as continuity and the sign change are checked.
Also check that the equation was entered correctly. For \(P(x)=Q(x)\), using \(P(x)+Q(x)\) instead of \(P(x)-Q(x)\) solves a different equation. For trigonometric equations, verify radians or degrees against the problem statement. These checks are simple, but they prevent many calculator-based errors.
Check Your Understanding
For each question, state the zero function and show how you would check a calculator result.
- For \(x^3-4x=1\), write an equivalent zero equation and explain what interval information you would want before using a root tool.
- A calculator reports \(c=1.4142\) for \(x^2-2=0\). What residual would you compute to check this candidate?
- If a continuous function has \(f(a)<0\) and \(f(b)>0\), what conclusion follows, and what stronger conclusion does not necessarily follow?
- Why might evaluating the function at values on both sides of a reported root be useful?
- What should you check before solving a trigonometric equation numerically?
