From a Table to a Trend
A limit table gives function values at inputs close to a target input. The next task is to interpret the pattern in those values. Are the outputs settling near a particular number, even if they do not move steadily in one direction? Or do they keep changing by substantial amounts as the inputs get closer?
This distinction matters because a limit is about the behavior of function values near an input, not about one isolated calculation. A table can suggest a candidate limit, but the pattern must be read carefully: outputs may fluctuate while still getting closer to one value, and a short list may conceal what happens at still closer inputs.
Measure the Deviations, Not Just the Direction
Suppose the target input is \(a\), and the sampled outputs appear to approach a number \(L\). For each sample, compare the input's distance from the target, \(|x-a|\), with the output's distance from the candidate, \(|f(x)-L|\). These distances help separate a genuine settling pattern from a sequence of values that merely happens to pass near \(L\).
For example, outputs might alternate above and below \(L\), yet their distances from \(L\) could steadily shrink. That is compatible with a stabilizing trend. By contrast, outputs that continue jumping between values far apart do not provide evidence that they are settling near one number. These observations guide a conjecture; a finite table alone cannot establish what happens for every sufficiently close input.
Worked Example: Values Settling Near Two
Worked Example: Values Settling Near Two
Consider \(f(x)=\dfrac{x^2-1}{x-1}\), for \(x\ne 1\). Use inputs close to \(1\) to identify a candidate limiting value.
For each input, calculate the numerator and denominator, then divide:
| \(x\) | \(x^2-1\) | \(x-1\) | \(f(x)\) | Distance from 2 |
|---|---|---|---|---|
| 0.9 | \(0.81-1=-0.19\) | \(0.9-1=-0.1\) | \((-0.19)/(-0.1)=1.9\) | \(|1.9-2|=0.1\) |
| 0.99 | \(0.9801-1=-0.0199\) | \(0.99-1=-0.01\) | \((-0.0199)/(-0.01)=1.99\) | \(|1.99-2|=0.01\) |
| 1.01 | \(1.0201-1=0.0201\) | \(1.01-1=0.01\) | \(0.0201/0.01=2.01\) | \(|2.01-2|=0.01\) |
| 1.1 | \(1.21-1=0.21\) | \(1.1-1=0.1\) | \(0.21/0.1=2.1\) | \(|2.1-2|=0.1\) |
The inputs are \(0.1\) or \(0.01\) away from \(1\). The corresponding outputs are \(0.1\) or \(0.01\) away from \(2\). On both sides of the target input, the samples therefore suggest \(2\) as the candidate limit. Notice that \(f(1)\) is not defined by the given formula, but this table concerns values at inputs near \(1\), not the value at \(1\).
Check: For \(x\ne1\), factor \(x^2-1=(x-1)(x+1)\). Thus \(f(x)=x+1\). Substitution gives \(f(0.9)=1.9\), \(f(0.99)=1.99\), \(f(1.01)=2.01\), and \(f(1.1)=2.1\), matching every table entry.
Why Shrinking Deviations Support a Candidate
The first example has an especially clear relationship between input and output distances. Factoring the numerator is valid when \(x\ne1\), because only then can the common factor \(x-1\) be cancelled:
Now write a nearby input as \(x=1+h\), where \(h\ne0\). Substituting into the simplified expression gives \(f(1+h)=2+h\). Subtracting the candidate value \(2\) shows exactly how far the output is from that candidate:
So, in this example, the output's distance from \(2\) equals the input's distance from \(1\). Any sampled input closer to \(1\) has an output just as close to \(2\). This explains why the table's pattern is not a coincidence. It also illustrates the useful reasoning behind distance comparisons: a smaller input offset may produce a smaller output deviation.
Worked Example: Fluctuations That Shrink
Worked Example: Fluctuations That Shrink
Consider \(g(x)=4+(x-1)\cos\!\left(\dfrac{\pi}{x-1}\right)\), for \(x>1\). Sample at \(x=1+\dfrac{1}{n}\) for \(n=2,3,4,5\). The inputs get closer to \(1\), but the outputs will not move in just one direction.
At these inputs, \(x-1=\dfrac1n\), so \(\dfrac{\pi}{x-1}=n\pi\). Since \(\cos(n\pi)=(-1)^n\), the calculations are:
| \(n\) | \(x=1+1/n\) | \(x-1\) | \(\cos(n\pi)\) | \(g(x)\) | Distance from 4 |
|---|---|---|---|---|---|
| 2 | \(1+1/2=1.5\) | \(1/2\) | 1 | \(4+(1/2)(1)=9/2=4.5\) | \(|4.5-4|=0.5\) |
| 3 | \(1+1/3=4/3\) | \(1/3\) | \(-1\) | \(4+(1/3)(-1)=11/3\) | \(|11/3-4|=1/3\) |
| 4 | \(1+1/4=1.25\) | \(1/4\) | 1 | \(4+(1/4)(1)=17/4=4.25\) | \(|4.25-4|=0.25\) |
| 5 | \(1+1/5=1.2\) | \(1/5\) | \(-1\) | \(4+(1/5)(-1)=19/5=3.8\) | \(|3.8-4|=0.2\) |
The outputs alternate above and below \(4\): \(4.5,\ 11/3,\ 4.25,\ 3.8\). They are not moving steadily in one direction. But their distances from \(4\) are \(1/2,\ 1/3,\ 1/4,\ 1/5\), which shrink as the inputs get closer to \(1\). These samples support \(4\) as a candidate limiting value despite the alternating pattern.
Check: At each sample, \(g(x)-4=(1/n)\cos(n\pi)\), so \(|g(x)-4|=(1/n)|\cos(n\pi)|=1/n\), because \(|\cos(n\pi)|=1\). This gives the four distances in the table and confirms the calculations.
When Values Keep Changing
Now compare shrinking fluctuations with outputs that remain far apart. Let \(q(x)=\sin\!\left(\dfrac{1}{x-1}\right)\), for \(x>1\). Choose the inputs \(x=1+\dfrac{2}{k\pi}\), where \(k\) is an integer greater than \(2\). Then \(\dfrac{1}{x-1}=\dfrac{k\pi}{2}\). For odd \(k\), the sine values alternate between \(1\) and \(-1\):
| \(k\) | \(x=1+2/(k\pi)\), approximately | \(1/(x-1)\) | \(q(x)\) |
|---|---|---|---|
| 5 | 1.1273 | \(5\pi/2\) | 1 |
| 7 | 1.0909 | \(7\pi/2\) | \(-1\) |
| 9 | 1.0707 | \(9\pi/2\) | 1 |
| 11 | 1.0579 | \(11\pi/2\) | \(-1\) |
The inputs shown get closer to \(1\), while the outputs continue to switch between \(1\) and \(-1\). Their difference is always \(2\), so this pattern does not show the outputs settling near one fixed number. A finite table cannot rule out every possible candidate by itself, but persistent large changes are a warning sign, not evidence of stabilization.
Compare this with the previous example. There, the outputs also changed direction, but the changes were confined to smaller and smaller distances from \(4\). Here, the outputs keep returning to two values separated by \(2\). The key question is not simply, “Do the values change?” It is, “Do the changes become small relative to a plausible fixed value?”
A Common Mistake: Demanding Monotonic Values
Worked Example: Correcting a Misread Trend
A learner sees the outputs \(4.5,\ 11/3,\ 4.25,\ 3.8\) from the previous example and concludes: “They go down, then up, then down, so they cannot approach a limit.” That conclusion confuses steady movement with settling.
Check the distances from the candidate \(4\):
Each distance is smaller than the preceding one, even though the outputs alternate around \(4\). Therefore the correct interpretation is that these samples support a stabilizing trend near \(4\), not that they prove there is no limit.
The reverse mistake is also possible: deciding that a limit exists merely because several outputs happen to be close together. A handful of rounded values may hide larger changes at closer inputs. Look for a repeated pattern in the distances, and use exact values where possible. Numerical evidence is most useful when the calculations and the trend agree.
A Practical Reading Strategy
When interpreting a table, work through these questions in order:
Check that the sampled inputs are getting closer to the input whose nearby behavior is being investigated.
Look for a number the outputs seem to cluster around. Do not require the outputs to approach it from only one side or to move monotonically.
For each sample, estimate or calculate the output's distance from the candidate. Ask whether those deviations are shrinking or remain substantial.
Use closer inputs and, when available, exact calculations. A table is evidence for a conjecture, not a guarantee about every nearby input.
Why Trend Detection Matters
In applications, function values may represent measured quantities, model outputs, or computed approximations. A pattern that appears to settle can help identify a useful value to investigate further. But treating a few samples as conclusive can be misleading: the inputs may not be close enough, rounding may hide changes, or a function may behave differently between the points sampled.
Good numerical reasoning therefore has two parts. First, describe the evidence accurately: state the candidate and explain how the output deviations change. Second, keep the conclusion appropriately limited: the table suggests a trend, but additional reasoning is needed to establish a general claim. That distinction lets a numerical pattern guide further work without overstating what it shows.
Check Your Understanding
For each question, explain what the sampled pattern supports and what it does not establish by itself.
- Outputs are \(5.2,\ 4.9,\ 5.03,\ 4.98\) at inputs increasingly close to \(a\). What candidate value might they suggest? What distance comparisons would help assess the pattern?
- Why can outputs alternate above and below a candidate while still showing a stabilizing trend?
- For \(g(x)=4+(x-1)\cos\!\left(\dfrac{\pi}{x-1}\right)\) at \(x=1+1/n\), what are the output and its distance from \(4\) when \(n=6\)?
- In the table for \(q(x)=\sin\!\left(\dfrac{1}{x-1}\right)\), what feature indicates that the sampled outputs are not settling near one value?
- Why does observing a stabilizing pattern in a finite table not, by itself, prove that a limit exists?
