A root of f(x) = 0 is a zero of f — trap it with a sign change, then halve the interval to any accuracy.
Every equation you have ever solved — linear, quadratic, exponential — can be read as a single question about a function: where does its graph cross the x-axis? Those crossings are the equation's roots, and they are exactly the function's zeros. This lesson unifies that view, proves that a change of sign must hide a zero between, and then forges that idea into a reliable machine — bisection — that corners any root to as many decimal places as you please. Along the way you will meet your first taste of continuity and approximation, the quiet engine behind everything a calculator does when it shows you a root.
Suppose you want to solve the equation x² − 2 = 0. Set f(x) = x² − 2 and the question becomes: for which inputs does the output equal zero? Graphically, an output of zero means the point sits on the x-axis — its height is 0. So solving f(x) = 0 is the same as finding where the curve y = f(x) meets the x-axis. Each such x earns two names that mean the very same number.
A number r is a zero of the function f if f(r) = 0, and it is a root of the equation f(x) = 0 — the two words describe the identical fact, just from the function's side or the equation's side. On the graph, r is an x-intercept: the place where y = f(x) touches or crosses the x-axis.
Three quick examples show how every kind of equation folds into this one frame. The quadratic f(x) = x² − 2 has two zeros, ±√2, because its bowl dips below the axis and rises back through it twice. The exponential g(x) = 2x − 8 has a single zero: 2x = 8 means x = 3, so the curve meets the axis exactly once. And the line h(x) = 2x − 3 has one zero at x = 32. Step through them below and watch the green dots land precisely where each blue curve crosses.
The lesson of the panel is simple but powerful: you never have to think about "solving an equation" and "finding a graph's crossings" as two different tasks. They are one task wearing two hats. Once that clicks, every equation in this book — and every one you will ever meet — becomes a hunt for zeros.
Here is a fact so natural you have always believed it without saying it aloud. Walk a curve from a point below the x-axis to a point above it without ever lifting your pencil — and somewhere along the way you must cross the axis. You cannot get from below to above while staying off the line zero. That crossing is a zero.
If f is continuous (an unbroken curve) on the closed interval [a, b], and the two endpoints have opposite signs — that is, f(a) · f(b) < 0 — then f has at least one zero somewhere in (a, b). One end is below the axis, the other above, and a connected curve has to pass through the middle.
Try it on f(x) = x² − 2 over [1, 2]. Compute the two ends directly: f(1) = 1 − 2 = −1 (below the axis) and f(2) = 4 − 2 = 2 (above it). Their product is (−1)(2) = −2 < 0, a sign change — so a zero is trapped between 1 and 2. We even know exactly which one: x = √2 ≈ 1.414, which does lie in (1, 2). The test does not tell us where the zero is, only that one is in there. That guarantee is the foothold for everything that follows.
(1) Agreeing signs say nothing. If f(a) and f(b) have the same sign, the test is silent — there may be no zero in between, or there may be two (the curve dips down and comes back up). For x² − 2 on [−2, 2], both ends equal +2, yet there are two zeros inside. A non-negative product is not a promise of "no root."
(2) Continuity is required. A curve with a jump can leap from below the axis to above it without crossing. The sign-change guarantee only holds for an unbroken, continuous f.
Once you see equations as zeros, counting solutions becomes counting crossings. The number of times y = f(x) meets the x-axis is the number of real roots of f(x) = 0 — no more, no fewer. A parabola that pokes through twice has two; one that just kisses the axis has one (a repeated root); one that floats entirely above has none.
The same eye solves a richer question: where two curves meet. The equation f(x) = g(x) asks for the x-values where y = f(x) and y = g(x) share a point — every intersection is one solution. (And it is the same idea as before, since f(x) = g(x) is just f(x) − g(x) = 0, a hunt for the zeros of the difference.)
To solve f(x) = g(x), draw both graphs and count where they cross. Each intersection's x-coordinate is one solution. This turns an equation you cannot solve by algebra — like 2x = x + 2 — into something you can simply look at.
How many solutions has 2x = x + 2? Plot the climbing exponential y = 2x and the straight line y = x + 2. They cross twice: once at x = 2 (where 2² = 4 and 2 + 2 = 4), and once near x ≈ −1.69 (where the steep tail of the exponential clips the line). So the equation has two solutions. Toggle the second example and watch a log meet a line.
Both equations resist ordinary algebra — there is no neat formula to isolate x. Yet the graph answers the counting question at a glance, and it even hands us rough locations. To turn "roughly here" into "1.414 to three decimals," we need a way to squeeze a bracket down. That machine is next.
Think of the higher–lower guessing game. A friend picks a number in [1, 100]; you guess 50; they say "higher"; now you know it is in [51, 100]; you guess 75; and each guess halves the range you still have to search. Bisection plays exactly this game with a root, using the sign-change test as the "higher or lower."
Start with a bracket [a, b] where f(a) and f(b) have opposite signs. Take the midpoint m = a + b2 and find the sign of f(m). Then keep the half whose endpoints still disagree in sign — the half that still traps the zero. The interval holding the root just got cut in half.
Walk one step on f(x) = x² − 2 starting from [1, 2]. The midpoint is m = 1 + 22 = 1.5, and f(1.5) = 1.5² − 2 = 2.25 − 2 = 0.25 > 0 (above the axis). Now compare ends: f(1) = −1 < 0 and f(1.5) = 0.25 > 0 still disagree, so the zero lives in the left half, [1, 1.5]. The right half [1.5, 2] has two positive ends — no sign change — so we discard it. One step, and the bracket shrank from width 1 to width 0.5, with √2 ≈ 1.414 safely inside.
Now just repeat. Each step halves the interval that holds the zero; the midpoint of the current bracket is your best estimate, and the error can be no larger than half the bracket's width (the zero is somewhere inside, and the midpoint sits no farther than that from either end). Stop when the bracket is shorter than the accuracy you want, and read off the midpoint.
Press Halve below to run the machine on f(x) = x² − 2 from [1, 2], step by step, and watch the bracket close in on √2 ≈ 1.4142.
The first several rows of the table the machine builds look like this — each line names the bracket [a, b], the midpoint m, the sign of f(m), and so which half to keep:
| Step | a | b | m = (a+b)/2 | f(m) | sign | keep |
|---|---|---|---|---|---|---|
| 1 | 1 | 2 | 1.5 | +0.25 | + | [1, 1.5] |
| 2 | 1 | 1.5 | 1.25 | −0.4375 | − | [1.25, 1.5] |
| 3 | 1.25 | 1.5 | 1.375 | −0.1094 | − | [1.375, 1.5] |
| 4 | 1.375 | 1.5 | 1.4375 | +0.0664 | + | [1.375, 1.4375] |
| 5 | 1.375 | 1.4375 | 1.40625 | −0.0225 | − | [1.40625, 1.4375] |
| … | … | … | … | … | … | → √2 |
Every bisection step multiplies the bracket width by ½, so after n steps the width is the original width times (½)n — roughly one more correct binary digit each time. Starting from [1, 2] (width 1), about 10 steps shrink the bracket below 0.001, since (½)10 ≈ 0.00098, and 11 steps below 0.0005. Slow and steady, but it always works once a sign change is found.
Stand back and look at the whole Functions strand you have built. The line, the parabola, the power function, the exponential, the logarithm — every one of them, ultimately, heads one way, or turns once and then heads one way again. None of them comes back to where it started. And we have just learned to read every equation through its function and to corner every root with bisection. The toolkit feels complete.
But step outside the page. A turning windmill, a swinging pendulum, the daily rise and fall of the tide, the alternating current humming in the wall — all of these return. They cycle back to where they began, again and again, forever. No power, no log, no polynomial can capture that, because those functions never repeat. Describing motion that comes back around needs a brand-new kind of function — one that is periodic.
Powers and logs rise or fall and keep going; some change returns — tides, wheels, waves. Capturing repetition needs a function built to repeat: the trigonometric functions, the next stage of the journey. With zeros, sign changes, and bisection in hand, you can now solve any equation a function can pose — including the ones the trig functions will bring.
Zero = root. A number r with f(r) = 0 is a zero of f and a root of f(x) = 0 — the place the curve crosses the x-axis.
The sign-change test. If f is continuous on [a, b] and f(a)·f(b) < 0, then a zero lies in (a, b). Agreeing signs say nothing; continuity is required.
Crossings count solutions. The real roots of f(x) = 0 are the x-axis crossings; the solutions of f(x) = g(x) are the x-coordinates where the two graphs meet.
Bisection. Halve the bracket, keep the half with a sign change, repeat. Each step cuts the width by ½; the midpoint is the estimate, with error ≤ half the width — cornering x² − 2 = 0 on √2 ≈ 1.4142.
What is the zero of f(x) = x − 5, and what root does it give for the equation f(x) = 0?
x = 5. Setting x − 5 = 0 gives x = 5, so 5 is the zero of f and the root of f(x) = 0 — it is where the line crosses the x-axis, at (5, 0).
Does f(x) = x² − 3 have a zero in the interval [1, 2]? Justify with the sign-change test.
Yes. f(1) = 1 − 3 = −2 < 0 and f(2) = 4 − 3 = 1 > 0. The product (−2)(1) = −2 < 0 is a sign change, so a zero lies in (1, 2) — in fact x = √3 ≈ 1.732.
From the graphs, how many solutions does 2x = x + 2 have?
Two. The exponential y = 2x and the line y = x + 2 cross twice: at x = 2 (both equal 4) and near x ≈ −1.69. Two intersections means two solutions.
For f(x) = x² − 2 on [1, 2], find the first midpoint and say which half bisection keeps.
m = 1 + 22 = 1.5, and f(1.5) = 0.25 > 0. Since f(1) = −1 < 0 disagrees with f(1.5) > 0, keep the left half [1, 1.5].
Take one more step from [1, 1.5]. Find the midpoint, the sign of f there, and the half kept.
m = 1 + 1.52 = 1.25, and f(1.25) = 1.5625 − 2 = −0.4375 < 0. Now f(1.25) < 0 and f(1.5) = 0.25 > 0 disagree, so keep the right half [1.25, 1.5].
Roughly how many bisection steps shrink the bracket [1, 2] below 0.001? Explain.
About 11. The width after n steps is (½)n. Since (½)10 ≈ 0.00098 and (½)11 ≈ 0.00049, you need 10 steps to dip just under 0.001 and 11 to be comfortably under 0.0005 — each step adds about one binary bit of accuracy.
Six questions to lock it in. Tap the answer you think is right.
The big idea. This closing lesson ties the entire Functions strand together with one unifying lens: solving an equation is finding a function's zeros. Linear, quadratic, exponential, and logarithmic equations all become the same graphical question — where does the curve cross the x-axis, or where do two curves meet? The existence theorem (a continuous sign change forces a crossing) gives students their first encounter with continuity and the Intermediate Value Theorem in plain language, and bisection turns that existence guarantee into a constructive, always-works algorithm — the quiet workhorse inside every "solve" button. The forward look to periodic functions motivates trigonometry as the one behavior these functions cannot model.
Watch for these misconceptions. (1) Believing that equal endpoint signs mean "no zero in between" — the test is silent there; a curve can dip below and rise back, giving two roots, or none. (2) Forgetting that the sign-change guarantee requires continuity; a function with a jump can cross from below to above without ever hitting zero. (3) In bisection, keeping the wrong half — always retain the subinterval whose endpoints still disagree in sign, computing each f-value rather than guessing. (4) Confusing a rough graphical estimate with an exact answer — bisection's strength is a guaranteed error bound (≤ half the bracket width), not a closed form.
Common Core. A-REI.D.11 (the solutions of f(x) = g(x) are the x-coordinates of the intersection points; approximate them using technology and successive approximations), A-REI.B (solve equations), and F-IF.B/C (interpret and analyze graphs, zeros, and intercepts of functions).