The Central Idea
A numerical table pairs selected input values with corresponding output values.
For a function \(f(x)\), a table may record
\[ x,\qquad f(x). \]By examining how \(f(x)\) changes as \(x\) changes, we can estimate roots, intersections, turning behavior, and other important features.
Successive approximation improves an estimate by repeating this process with smaller steps or narrower intervals.
What Is a Numerical Table?
A numerical table is an organized collection of input-output pairs.
For example, if
\[ f(x)=x^2-2, \]then selected values give
| \(x\) | \(f(x)=x^2-2\) |
|---|---|
| \(1\) | \(-1\) |
| \(1.5\) | \(0.25\) |
| \(2\) | \(2\) |
The table shows that the function changes sign between \(1\) and \(1.5\).
Tables as Discrete Samples
A function may be defined for infinitely many input values, but a numerical table records only selected samples.
If
\[ f(x)=x^2, \]then a table such as
| \(x\) | \(f(x)\) |
|---|---|
| \(-2\) | \(4\) |
| \(-1\) | \(1\) |
| \(0\) | \(0\) |
| \(1\) | \(1\) |
| \(2\) | \(4\) |
samples five points from a function that contains infinitely many points.
Why Tables Are Useful
A numerical table can help us:
- estimate where a function is zero;
- compare two functions numerically;
- identify increasing or decreasing behavior;
- locate intervals where a solution may lie;
- test conjectures;
- refine approximate solutions;
- organize repeated calculations.
Turning an Equation into a Function
To use a numerical table for an equation, it is often helpful to rewrite the equation in the form
\[ f(x)=0. \]Suppose
\[ x^2=3. \]Move everything to one side:
\[ x^2-3=0. \]Define
\[ \boxed{f(x)=x^2-3}. \]Now a solution of the equation is a zero of \(f\).
Finding a Root from a Table
Consider
\[ f(x)=x^2-3. \]| \(x\) | \(f(x)\) |
|---|---|
| \(1\) | \(-2\) |
| \(1.5\) | \(-0.75\) |
| \(2\) | \(1\) |
Because \(f(1.5)<0\) and \(f(2)>0\), the positive root lies somewhere between \(1.5\) and \(2\), assuming the function is continuous on that interval.
Sign Changes
A sign change occurs when one function value is negative and another is positive.
For example,
\[ f(a)<0 \]and
\[ f(b)>0. \]If \(f\) is continuous on \([a,b]\), then the Intermediate Value Theorem guarantees at least one root in the interval:
\[ \boxed{c\in(a,b)\text{ such that }f(c)=0}. \]A Sign Change Is Strong Evidence
For continuous functions, a sign change gives a rigorous guarantee that at least one zero lies between the two inputs.
This makes sign changes especially useful in numerical root-finding.
The table does not yet tell us the exact root, but it gives us a search interval.
Successive Approximation
Successive approximation improves an estimate through a sequence of increasingly precise steps.
A typical pattern is:
\[ \text{broad interval} \longrightarrow \text{narrower interval} \longrightarrow \text{still narrower interval}. \]Each stage uses previous information to focus the next calculation.
Example: Approximating \(\sqrt{3}\)
We want the positive solution of
\[ x^2=3. \]First test whole numbers:
\[ 1^2=1<3 \]and
\[ 2^2=4>3. \]Therefore,
\[ \boxed{1<\sqrt3<2}. \]Refining to Tenths
Now test tenths near the root:
| \(x\) | \(x^2\) |
|---|---|
| \(1.7\) | \(2.89\) |
| \(1.8\) | \(3.24\) |
Therefore,
\[ \boxed{1.7<\sqrt3<1.8}. \]Refining to Hundredths
Test hundredths:
| \(x\) | \(x^2\) |
|---|---|
| \(1.73\) | \(2.9929\) |
| \(1.74\) | \(3.0276\) |
Therefore,
\[ \boxed{1.73<\sqrt3<1.74}. \]Refining to Thousandths
Test thousandths:
| \(x\) | \(x^2\) |
|---|---|
| \(1.732\) | \(2.999824\) |
| \(1.733\) | \(3.003289\) |
Therefore,
\[ \boxed{1.732<\sqrt3<1.733}. \]This process can continue as far as the required precision demands.
A Table Can Track Function Values Instead
Instead of tabulating \(x^2\), we can tabulate
\[ f(x)=x^2-3. \]| \(x\) | \(f(x)\) |
|---|---|
| \(1.732\) | \(-0.000176\) |
| \(1.733\) | \(0.003289\) |
The sign change directly identifies the interval containing the root.
Step Size
The difference between successive input values is called the step size.
For the table
\[ 1.70,\quad1.71,\quad1.72,\quad1.73,\quad1.74, \]the step size is
\[ \boxed{0.01}. \]Smaller step sizes usually provide finer resolution.
Large Steps versus Small Steps
Large step sizes are useful for locating a broad region quickly.
Small step sizes are useful for refining a solution once the general region is known.
An efficient strategy is therefore:
\[ \text{coarse search} \longrightarrow \text{fine search}. \]Coarse Search
Suppose
\[ f(x)=x^3-x-2. \]Test integer inputs:
| \(x\) | \(f(x)\) |
|---|---|
| \(1\) | \(-2\) |
| \(2\) | \(4\) |
The root lies between \(1\) and \(2\).
Finer Search
Now test tenths:
| \(x\) | \(f(x)=x^3-x-2\) |
|---|---|
| \(1.5\) | \(-0.125\) |
| \(1.6\) | \(0.496\) |
Therefore,
\[ \boxed{1.5Refining Again
Test hundredths:
| \(x\) | \(f(x)\) |
|---|---|
| \(1.52\) | \(-0.008192\) |
| \(1.53\) | \(0.052577\) |
Therefore,
\[ \boxed{1.52Using Midpoints
One way to refine an interval is to test its midpoint.
For an interval
\[ [a,b], \]the midpoint is
\[ \boxed{m=\frac{a+b}{2}}. \]Evaluating \(f(m)\) can tell us which half of the interval should be kept.
This idea becomes the foundation of the bisection method.
Successive Approximation Does Not Require Equal Steps
Refinement does not always mean using a fixed decimal grid.
A method may instead generate values such as
\[ 1.5,\quad1.55,\quad1.525,\quad1.521875,\ldots \]according to a rule.
The defining feature is not equal spacing. It is that later estimates are intended to improve on earlier ones.
Tables for Comparing Two Functions
Suppose we want to solve
\[ 2^x=5x. \]We can tabulate both sides:
| \(x\) | \(2^x\) | \(5x\) |
|---|---|---|
| \(0\) | \(1\) | \(0\) |
| \(1\) | \(2\) | \(5\) |
| \(4\) | \(16\) | \(20\) |
| \(5\) | \(32\) | \(25\) |
The relative order of the two sides changes, indicating possible intersections.
Difference Tables for Equations
Instead of comparing two columns directly, define
\[ f(x)=2^x-5x. \]Then solving
\[ 2^x=5x \]becomes solving
\[ f(x)=0. \]This reduces the problem to a single output column and makes sign changes easier to identify.
Choosing Useful Input Values
A good table does not need to contain every nearby number.
Useful input choices depend on the problem.
Common strategies include:
- starting with integers;
- testing simple fractions;
- using powers of ten;
- sampling evenly spaced values;
- concentrating values near suspected roots;
- refining only where the table shows meaningful change.
Do Not Oversample Too Early
If a root might lie somewhere between \(-100\) and \(100\), beginning with steps of \(0.001\) would require an enormous number of evaluations.
A better strategy is to search broadly first and refine only after a useful region is found.
Numerical efficiency often comes from choosing the scale of the search intelligently.
Tables and Monotonic Behavior
A table can also suggest whether a function is increasing or decreasing over sampled inputs.
For example:
| \(x\) | \(f(x)\) |
|---|---|
| \(1\) | \(3\) |
| \(2\) | \(5\) |
| \(3\) | \(8\) |
| \(4\) | \(12\) |
The sampled outputs increase as the inputs increase.
The table suggests increasing behavior on the sampled region, although a finite table alone does not prove global monotonicity.
Tables and Local Behavior
Numerical tables can reveal local patterns near a particular input.
Suppose values are sampled near \(x=2\):
| \(x\) | \(f(x)\) |
|---|---|
| \(1.9\) | \(4.01\) |
| \(1.99\) | \(4.0001\) |
| \(2\) | \(4\) |
| \(2.01\) | \(4.0001\) |
| \(2.1\) | \(4.01\) |
The values suggest a local minimum near \(x=2\).
Tables Can Reveal Symmetry
Consider
\[ f(x)=x^2. \]| \(x\) | \(f(x)\) |
|---|---|
| \(-2\) | \(4\) |
| \(-1\) | \(1\) |
| \(0\) | \(0\) |
| \(1\) | \(1\) |
| \(2\) | \(4\) |
Matching outputs at opposite inputs suggest the symmetry
\[ f(-x)=f(x). \]Tables and Limits
Numerical tables are also useful for exploring what happens as inputs approach a particular value.
For example, values of
\[ f(x)=\frac{x^2-1}{x-1} \]near \(x=1\) can suggest the limiting value even though the original expression is undefined at \(x=1\).
This idea is developed more fully in calculus.
Tables and Recurrence
A table can also record a sequence generated recursively.
Suppose
\[ x_{n+1}=\frac12\left(x_n+\frac{2}{x_n}\right) \]with
\[ x_0=1. \]Then:
| \(n\) | \(x_n\) |
|---|---|
| \(0\) | \(1\) |
| \(1\) | \(1.5\) |
| \(2\) | \(1.416666\ldots\) |
| \(3\) | \(1.414215\ldots\) |
The values approach \(\sqrt2\).
Successive Approximation as a Sequence
Numerically generated estimates can be viewed as a sequence
\[ x_0,\quad x_1,\quad x_2,\quad x_3,\quad\ldots \]If the estimates approach a number \(L\), we write informally here that
\[ x_n\to L. \]Later lessons formalize convergence and stopping conditions.
Approximation from Above and Below
Some methods produce estimates on both sides of a target value.
For example,
\[ 1.414<\sqrt2<1.415. \]The lower estimate is
\[ 1.414, \]and the upper estimate is
\[ 1.415. \]This two-sided information is stronger than reporting only one rounded decimal.
Interval Width as a Measure of Uncertainty
If a solution is known to lie in
\[ [a,b], \]the interval width is
\[ \boxed{b-a}. \]A smaller interval gives a tighter location for the unknown solution.
Example: Shrinking Interval Width
Suppose a root is first located in
\[ [1,2]. \]The width is
\[ 2-1=1. \]Later refinement gives
\[ [1.52,1.53]. \]The new width is
\[ 1.53-1.52=0.01. \]The search interval is now one hundred times narrower.
Choosing a Stopping Point
Successive approximation does not continue forever in practical work.
We stop when the result is precise enough for the problem.
Possible stopping conditions include:
- a required number of decimal places has been reached;
- the interval width is below a chosen tolerance;
- successive estimates differ by less than a chosen amount;
- the residual is sufficiently small.
Example: Stopping by Interval Width
Suppose the requirement is to locate a root inside an interval of width less than
\[ 0.001. \]If the current interval is
\[ [2.2360,2.2365], \]then its width is
\[ 2.2365-2.2360=0.0005. \]Since
\[ 0.0005<0.001, \]the stopping requirement has been satisfied.
Tables by Hand
Small tables are often easy to construct manually.
A basic workflow is:
- choose input values;
- substitute each input into the function;
- record the output;
- inspect signs, magnitudes, or trends;
- choose a narrower region if needed;
- repeat.
Tables with a Calculator
Many calculators can generate function tables automatically.
The mathematical reasoning remains the same:
- define the function correctly;
- choose a useful starting input;
- choose an appropriate step size;
- interpret the resulting values;
- refine the search where needed.
Technology accelerates the arithmetic but does not replace the logic of the search.
Tables with a Spreadsheet
A spreadsheet can generate large tables efficiently.
One column may contain input values,
\[ x, \]while another contains a formula for
\[ f(x). \]Filling the formula downward evaluates the function across many inputs.
Later lessons examine spreadsheet methods in more detail.
Tables with Computer Algebra or Code
A computer algebra system or program can automate repeated evaluation.
Conceptually, the process is still:
\[ \text{choose }x \longrightarrow \text{evaluate }f(x) \longrightarrow \text{inspect} \longrightarrow \text{refine}. \]This repeated pattern is one of the simplest forms of an algorithm.
Accuracy of Table Values
Table entries may themselves be approximate.
If
\[ f(x)=\sin x, \]then calculator-generated values such as
\[ \sin(1)\approx0.8414709848 \]are finite numerical approximations.
The reliability of conclusions depends on using sufficient numerical precision.
Rounding in a Table
Excessive rounding can hide important distinctions.
Suppose two values are
\[ -0.0004 \]and
\[ 0.0003. \]If both are rounded to two decimal places, they become
\[ 0.00, \]and the sign change disappears from the displayed table.
Numerical tables should retain enough digits to support the intended analysis.
Repeated Roots Can Be Harder to Detect
Not every root produces a sign change.
Consider
\[ f(x)=(x-2)^2. \]Then
\[ f(1)>0,\qquad f(2)=0,\qquad f(3)>0. \]The function touches zero at \(x=2\) but does not change sign.
Therefore, sign-change searches can miss roots of even multiplicity.
A Sign Change Can Contain More Than One Root
A sign change guarantees at least one root for a continuous function, but it does not necessarily tell us how many roots are present.
A wide interval may contain several zeros.
Finer sampling, graphing, algebraic analysis, or additional numerical methods may be needed to distinguish them.
Tables Do Not Replace Proof
A finite table provides evidence about sampled values.
It does not automatically prove a statement about every value in a domain.
For example, a table showing
\[ f(1)>0,\quad f(2)>0,\quad f(3)>0 \]does not prove
\[ f(x)>0 \]for every real \(x\).
Common Error: Using Steps That Are Too Large
A table with widely spaced inputs may skip over important behavior.
Suppose a narrow feature occurs between \(x=2.01\) and \(x=2.02\).
A table using only
\[ x=2,\quad3,\quad4 \]may never reveal it.
Step size should match the scale of the feature being investigated.
Common Error: Refining the Wrong Interval
Once a useful interval is found, refinement should remain focused on the interval containing the target solution.
If
\[ f(1.5)<0 \]and
\[ f(1.6)>0, \]then testing values near \(4\) does not refine that particular root.
Common Error: Assuming the Closest Output Is the Root
Suppose a table contains
\[ f(1.52)=-0.008 \]and
\[ f(1.53)=0.053. \]The root is not necessarily equal to \(1.52\) merely because its output is closer to zero.
The table tells us that the root lies between the two inputs, not that one endpoint is exact.
Common Error: Confusing Input Precision with Solution Precision
Listing inputs to three decimal places does not automatically guarantee a root accurate to three decimal places.
The conclusion depends on how the values bracket or approximate the root.
Precision should be justified by the interval, residual, or other error information.
Common Error: Ignoring Continuity
A sign change only guarantees a zero between two points when continuity conditions are satisfied.
For example, a function with a vertical asymptote may change sign across a discontinuity without ever being zero.
Numerical reasoning must therefore account for the mathematical properties of the function.
A Reliable Table-Based Root-Finding Procedure
- Rewrite the equation as \(f(x)=0\).
- Choose a broad set of convenient inputs.
- Evaluate \(f(x)\) at those inputs.
- Look for zeros, sign changes, or very small residuals.
- Check that continuity assumptions are appropriate.
- Identify a smaller interval containing the target root.
- Reduce the step size or otherwise refine the search.
- Repeat until the required accuracy is reached.
- Report the final result with an appropriate approximation symbol.
Example: Full Successive-Approximation Workflow
Approximate the positive root of
\[ x^2=10. \]Define
\[ f(x)=x^2-10. \]First:
\[ f(3)=9-10=-1 \]and
\[ f(4)=16-10=6. \]Therefore,
\[ 3<\sqrt{10}<4. \]Refining the Workflow to Tenths
| \(x\) | \(f(x)=x^2-10\) |
|---|---|
| \(3.1\) | \(-0.39\) |
| \(3.2\) | \(0.24\) |
Therefore,
\[ \boxed{3.1<\sqrt{10}<3.2}. \]Refining the Workflow to Hundredths
| \(x\) | \(f(x)\) |
|---|---|
| \(3.16\) | \(-0.0144\) |
| \(3.17\) | \(0.0489\) |
Therefore,
\[ \boxed{3.16<\sqrt{10}<3.17}. \]Refining the Workflow to Thousandths
| \(x\) | \(f(x)\) |
|---|---|
| \(3.162\) | \(-0.001756\) |
| \(3.163\) | \(0.004569\) |
Therefore,
\[ \boxed{3.162<\sqrt{10}<3.163}. \]The root can now be rounded to
\[ \boxed{\sqrt{10}\approx3.162} \]to three decimal places.
Practice 1: Build a Table
For
\[ f(x)=x^2-5, \]evaluate \(f(x)\) at
\[ x=2,\quad2.2,\quad2.3. \]Use the results to identify an interval containing the positive root.
Practice 2: Locate a Sign Change
For
\[ f(x)=x^3-4, \]evaluate
\[ f(1),\qquad f(2). \]What interval contains the real root?
Practice 3: Refine to Tenths
Use a numerical table to show that
\[ \sqrt5 \]lies between \(2.2\) and \(2.3\).
Practice 4: Refine to Hundredths
Use
\[ f(x)=x^2-5 \]to show that the positive root lies between \(2.23\) and \(2.24\).
Practice 5: Step Size
What is the step size of the input sequence
\[ 1.20,\quad1.25,\quad1.30,\quad1.35,\quad1.40? \]Practice 6: Interval Width
Find the width of
\[ [1.732,1.733]. \]Practice 7: Interpret a Table
Suppose
\[ f(4.6)=-0.18 \]and
\[ f(4.7)=0.09. \]If \(f\) is continuous on \([4.6,4.7]\), what can you conclude?
Practice 8: Repeated Root
Explain why a sign-change search might fail to detect the root of
\[ f(x)=(x-3)^2. \]Practice 9: Successive Approximation
A root is first located in
\[ [2,3], \]then in
\[ [2.2,2.3], \]then in
\[ [2.23,2.24]. \]Describe how the approximation is improving.
Practice 10: Stopping Condition
A method must stop when the interval width is less than
\[ 0.0001. \]Does the interval
\[ [1.41420,1.41427] \]satisfy the requirement?
Practice Answers
Practice 1
\[ f(2)=4-5=-1, \] \[ f(2.2)=4.84-5=-0.16, \] \[ f(2.3)=5.29-5=0.29. \]Since the sign changes between \(2.2\) and \(2.3\),
\[ \boxed{2.2<\sqrt5<2.3}. \]Practice 2
\[ f(1)=1-4=-3 \]and
\[ f(2)=8-4=4. \]Therefore,
\[ \boxed{1<\sqrt[3]{4}<2}. \]Practice 3
\[ 2.2^2=4.84<5 \]and
\[ 2.3^2=5.29>5. \]Thus,
\[ \boxed{2.2<\sqrt5<2.3}. \]Practice 4
\[ 2.23^2=4.9729<5 \]and
\[ 2.24^2=5.0176>5. \]Therefore,
\[ \boxed{2.23<\sqrt5<2.24}. \]Practice 5
Consecutive inputs differ by
\[ 1.25-1.20=0.05. \]Therefore, the step size is
\[ \boxed{0.05}. \]Practice 6
\[ 1.733-1.732=0.001. \]The interval width is
\[ \boxed{0.001}. \]Practice 7
Because
\[ f(4.6)<0 \]and
\[ f(4.7)>0, \]and \(f\) is continuous on the interval, at least one root lies between the two values:
\[ \boxed{4.6Since
\[ (x-3)^2\ge0 \]for every real \(x\), the function does not change sign at \(x=3\).
It touches zero and turns upward, so a method that looks only for sign changes can miss the root.
Practice 9
The intervals become progressively narrower:
\[ 1,\qquad0.1,\qquad0.01. \]Each stage localizes the root more precisely than the stage before it.
Practice 10
The interval width is
\[ 1.41427-1.41420=0.00007. \]Since
\[ 0.00007<0.0001, \]the stopping condition is satisfied.
What to Remember
- Numerical tables organize selected input-output pairs.
- Tables can help locate roots, intersections, and local behavior.
- A sign change across a continuous interval guarantees at least one root.
- Successive approximation improves an estimate in stages.
- Large steps are useful for broad searches; small steps are useful for refinement.
- Interval width measures how tightly a solution has been localized.
- Not every root produces a sign change.
- Finite tables provide evidence but do not automatically prove global claims.
- Technology speeds up table construction but does not replace mathematical interpretation.
- Refinement should continue until the required tolerance or precision is reached.
Where We Go Next
Numerical tables introduce the basic search-and-refine pattern. The next step is to turn that pattern into explicit procedures that repeatedly update values or search intervals.
The next lesson, Iteration, Recursion, and Search Intervals, develops the language of repeated numerical processes and prepares for bracketing, bisection, fixed-point iteration, and other root-finding algorithms.