Using a Table to Estimate a Limit
A function table can make nearby behavior easier to inspect. Instead of judging a graph’s scale or shape, we can compare specific input values with the function values they produce. The goal is to estimate what output the function is approaching as the input approaches a target.
A useful table includes inputs on both sides of the target. If the outputs from both sides settle near the same number as the inputs get closer, that number is a reasonable estimate for the limit. This is evidence, not a proof: a finite table shows only finitely many function values.
Here, “from below” means using inputs less than \(a\), and “from above” means using inputs greater than \(a\). These descriptions concern the inputs, not whether the function values are below or above the proposed limit. The outputs may lie on either side of \(L\).
How to Read the Evidence
For each row, compare the input’s distance from the target with its function value. An input such as \(0.99\) is closer to \(1\) than \(0.9\) is, because \(|0.99-1|=0.01\) while \(|0.9-1|=0.1\). Then check whether the corresponding outputs move closer to a common candidate.
The main question is not whether every output is identical. Nearby outputs often differ. Instead, look for a consistent pattern: as the inputs approach the target from below and above, do the outputs settle near the same number? Agreement between the two sides matters. Values from only one side can suggest a candidate while concealing different behavior from the other side.
A table does not establish that a limit exists or determine its exact value. It samples selected inputs, and unlisted inputs may behave differently. A sensible conclusion therefore uses language such as “the table suggests” or “the data support an estimate of,” rather than claiming that the table proves the limit.
Worked Examples
Worked Example: A Rational Function Near a Removable Gap
Estimate the limit of \(f(x)=\dfrac{x^2-1}{x-1}\) as \(x\) approaches \(1\), using inputs on both sides. The function is not defined at \(x=1\), so the table uses nearby inputs other than \(1\).
For any \(x\ne1\), factor the numerator and simplify:
Now substitute each listed input into \(x+1\). For example, at \(x=0.9\), \(f(0.9)=0.9+1=1.9\); at \(x=0.99\), \(f(0.99)=0.99+1=1.99\); and at \(x=1.001\), \(f(1.001)=1.001+1=2.001\).
| Input \(x\) | Distance \(|x-1|\) | Function value \(f(x)\) |
|---|---|---|
| 0.9 | 0.1 | 1.9 |
| 0.99 | 0.01 | 1.99 |
| 0.999 | 0.001 | 1.999 |
| 1.001 | 0.001 | 2.001 |
| 1.01 | 0.01 | 2.01 |
| 1.1 | 0.1 | 2.1 |
On the lower-input side, the outputs move from \(1.9\) to \(1.99\) to \(1.999\), approaching \(2\). On the higher-input side, they move from \(2.1\) to \(2.01\) to \(2.001\), also approaching \(2\). The table supports the estimate \(2\).
Check: The identity \(f(x)=x+1\) holds for every \(x\ne1\). As \(x\) gets close to \(1\), \(x+1\) gets close to \(1+1=2\). This agrees with the table and confirms the estimate without substituting the excluded input into the original expression.
Worked Example: A Square-Root Function Near an Input of 4
Estimate the limit of \(g(x)=\sqrt{x+5}\) as \(x\) approaches \(4\). Use inputs on both sides. The outputs below are rounded to six decimal places.
Substitute each input into the function. For \(x=3.9\), \(x+5=8.9\), so \(g(3.9)=\sqrt{8.9}\approx2.983286\). For \(x=3.99\), \(g(3.99)=\sqrt{8.99}\approx2.998333\). For \(x=4.001\), \(g(4.001)=\sqrt{9.001}\approx3.000167\).
| Input \(x\) | Radicand \(x+5\) | Function value \(g(x)\) |
|---|---|---|
| 3.9 | 8.9 | 2.983286 |
| 3.99 | 8.99 | 2.998333 |
| 3.999 | 8.999 | 2.999833 |
| 4.001 | 9.001 | 3.000167 |
| 4.01 | 9.01 | 3.001666 |
| 4.1 | 9.1 | 3.016621 |
For inputs below \(4\), the listed outputs approach \(3\) from below. For inputs above \(4\), the outputs approach \(3\) from above. The values on both sides support the estimate \(3\).
Check: The entries are consistent with the function: for instance, \(2.998333^2\) is approximately \(8.99\), and \(3.000167^2\) is approximately \(9.001\), allowing for rounding. Also, the input \(4\) gives radicand \(4+5=9\) and output \(\sqrt{9}=3\). This direct check agrees with the numerical estimate.
Worked Example: A Table That Does Not Support a Finite Estimate
Consider \(h(x)=\dfrac{1}{x}\) near the target \(0\). Calculate the values at the listed inputs.
For example, \(h(-0.1)=1/(-0.1)=-10\), while \(h(0.1)=1/0.1=10\). For the closer inputs, \(h(-0.01)=-100\) and \(h(0.01)=100\).
| Input \(x\) | Function value \(h(x)\) |
|---|---|
| -0.1 | -10 |
| -0.01 | -100 |
| -0.001 | -1000 |
| 0.001 | 1000 |
| 0.01 | 100 |
| 0.1 | 10 |
As the negative inputs get closer to \(0\), the outputs become large in magnitude and negative. For positive inputs closer to \(0\), the outputs become large in magnitude and positive. The two groups do not settle near one common finite number. This table therefore does not support a finite limiting-value estimate.
Check: Each output is the reciprocal of its input: \(1/(-0.001)=-1000\) and \(1/(0.001)=1000\). The calculations confirm that the outputs move away from a shared finite value rather than clustering near one.
Why Both Sides Matter
Inputs below and above the target provide two separate collections of evidence. If one collection suggests a value but the other suggests a different value, a single two-sided estimate is not supported. For example, seeing outputs close to \(5\) for inputs below \(a\) would not be enough if inputs above \(a\) instead produce outputs close to \(8\).
When both sides support the same candidate, confidence in the estimate increases. Yet agreement in a table remains limited evidence. The table might omit important inputs, and a handful of rows cannot cover every input close to the target. A table is a tool for forming and checking a conjecture, not a substitute for a mathematical justification.
A Short Justification for the Table Method
The method is grounded in what a limit describes: the behavior of function values for inputs near a target, rather than a calculation at just one input. A table makes a finite sample of that nearby behavior visible. If inputs increasingly close to \(a\) produce outputs increasingly close to a candidate \(L\) from both sides, the observed pattern is exactly the kind of pattern a limit estimate should capture.
For the first example, this reasoning can be made explicit. Whenever \(x\ne1\), the function equals \(x+1\). Its difference from the candidate \(2\) is therefore
Thus, for every input in the table other than \(1\), the output’s distance from \(2\) equals the input’s distance from \(1\). If an input is one hundredth of a unit from \(1\), its output is one hundredth of a unit from \(2\). This exact relationship explains why the displayed values cluster around \(2\); the table illustrates the behavior, while the identity justifies it for every allowed input.
Common Mistake: Averaging the Two Sides
A table can show very different values on the two sides, so it may be tempting to average one value from each side and call the result the limit. That procedure does not test whether the outputs themselves approach a common value.
For \(h(x)=1/x\), take inputs \(-0.1\) and \(0.1\). Their outputs are \(-10\) and \(10\). A mistaken calculation would be
and would incorrectly report \(0\) as the estimate. But closer inputs, \(-0.01\) and \(0.01\), give outputs \(-100\) and \(100\), not values close to \(0\). Averaging hides the fact that the outputs are moving away from a common finite candidate. The correction is to examine how each side’s outputs behave as its inputs get closer, not to combine two selected outputs arithmetically.
Using Estimates Carefully
When you report a table-based estimate, include enough information for someone else to see how you reached it. State the target input, identify the candidate output, and describe what the table shows on both sides. If the values have been rounded, say so; a displayed decimal may not be exact.
Keep the conclusion proportional to the evidence. A clear pattern in several rows can make an estimate plausible, but it cannot prove that all sufficiently nearby inputs behave the same way. If the two sides appear to disagree, do not force a single answer. State that the table does not support one common estimate, and check the calculations before deciding whether the pattern is genuine.
Check Your Understanding
Use the table evidence carefully. In each question, distinguish an estimate from a proof.
- A function’s values at inputs just below \(a\) are \(4.8\), \(4.98\), and \(4.998\). Its values at inputs just above \(a\) are \(5.2\), \(5.02\), and \(5.002\). What value do the data support, and what feature of both sets of values supports it?
- Why should a table for estimating a two-sided limit include inputs both below and above the target?
- For \(f(x)=(x^2-1)/(x-1)\), why is it valid to use the expression \(x+1\) at nearby inputs, but not to substitute \(x=1\) into the original quotient?
- For \(h(x)=1/x\), why is averaging \(h(-0.1)\) and \(h(0.1)\) not a valid way to estimate behavior near \(0\)?
- What can a finite table suggest about a limit, and what can it not establish on its own?
