Prove this limit using only the definition of convergence, then try the proof out with the tolerance .
Proved. Given any tolerance , every natural number works.
Guess the limit, then say exactly what must be proved
A proof from the definition needs a target before it can start: the definition only lets us CHECK a candidate limit, it never produces one. So we first find the candidate by asking what happens to the fraction when is enormous.
When is huge, the fixed numbers and are tiny luggage next to the terms that grow with . Dividing the top and the bottom by makes this visible:
As grows, anything of the form shrinks toward , so the fraction heads toward . That is a guess, not yet a proof. The proof is the rest of this page.
Measure the distance between a term and the limit
The definition talks about one quantity only, the distance . Right now it is a difference of two fractions, and we cannot see how it depends on . We simplify it until appears in exactly one place, because then we can solve for .
Remove the absolute value bars on the bottom
Bars block algebra: we cannot multiply an inequality through by and solve for while is locked inside them. A quantity that is positive equals its own absolute value, so we find where the bottom is positive and work there.
So the bottom is positive exactly when , which for natural numbers means . For those , . We are free to ignore the first few terms: the definition only cares about what happens from some position onward, and we will make sure .
Rough work: how large must n be?
A proof is READ forwards but FOUND backwards. We start from the inequality we want, , and rewrite it, losing nothing at each move, until it says plainly how large must be. That tells us which to announce. This is rough work, not yet the proof.
Announce the rule for M
The sceptic hands us ; the rough work says the terms behave once is past . So we answer with any natural number beyond that value. We must also be sure such a natural number EXISTS, and that is not free: it is a property of the real numbers.
Notice how the rule behaves: a smaller makes larger, so moves further out. A stricter sceptic forces us deeper into the sequence, but never beyond reach. That dependence of on is the heart of convergence.
This choice also guarantees , as we promised: because is positive.
The proof, written forwards
Rough work convinces US; a proof must convince the sceptic, and it has to run in the order the definition demands: first arrives, then we name , then we take ANY and show the distance is under . Here it is, with every relation justified.
Let be given. Choose with . Let . (Here is the set of natural numbers and means "is a member of".)
First, a lower bound for the bottom:
So the bottom is positive, which lets us use the bar-free distance, and it is LARGER than . Now the distance itself:
Test the rule with a real tolerance
A rule for that has never been run on a number is only a claim. We play the sceptic ourselves, pick , and check the terms on both sides of the position our rule gives.
The smallest natural number greater than that is . The table computes the terms around that position exactly.
| under ? | |||
|---|---|---|---|
| no | |||
| yes | |||
| yes | |||
| yes |