Before you begin
- Finite and one-sided limits; algebraic factoring; trigonometric functions in radians.
Learning objectives
- Apply quotient and composition rules only when their hypotheses hold.
- Resolve removable algebraic indeterminacies.
- Use bounding arguments and distinguish growth at infinity.
Substitution is often the last line of a limit calculation, but it is not the definition of a limit. It works for polynomials because their limits obey algebraic rules; it works for a rational expression at points where the denominator has a nonzero limit. Recognizing those conditions lets us distinguish an ordinary calculation from one that needs more work.
The algebra of finite limits
Suppose and , with finite. Then limits commute with sums, differences, constant multiples, and products. Thus and . For quotients we additionally require , giving . Positive integer powers follow by repeated products.
For composition, if and is continuous at , then , provided the composition is defined nearby. Continuity will receive a careful definition shortly. Root functions are continuous on their real domains, with one-sided care at endpoints. Hence
The denominator approaches , not zero. That single check justifies the final division.
Zero over zero is a prompt to investigate
For at , substitution gives the meaningless quotient . It does not give an answer of zero, infinity, or “undefined limit.” For ,
so the limit is . The two expressions agree throughout a punctured neighborhood. Limits depend on exactly that nearby behavior.
Compare , , and at zero. All display on substitution, yet their behaviors are respectively a constant , a limit , and incompatible unbounded one-sided limits. The notation indeterminate form records insufficient information, not a result.
Conjugates handle some square-root cases. For near ,
because multiplying numerator and denominator by exposes . The limit is . Multiplying by the conjugate is legal nearby because is nonzero.
Squeezing every nearby value
The squeeze theorem says that if for all sufficiently near allowed inputs, and both bounding functions approach , then also approaches . The inequalities must hold throughout the relevant neighborhood, not only at selected table entries.
Since , we have
Both outer expressions tend to zero. This supplies a conclusion even though the middle function oscillates infinitely often near zero.
An important trigonometric limit is , with angles measured in radians. On , the unit-circle area comparison gives . Dividing yields . As , both bounds tend to ; symmetry gives the left side. This argument uses elementary geometric continuity of cosine, not a derivative formula for sine. Also,
These results will justify trigonometric derivatives. Degrees would introduce a conversion factor; the radian convention is mathematical infrastructure, not a cosmetic preference.
Unbounded inputs and dominant terms
To study a rational expression as , divide numerator and denominator by a suitable highest power. For example,
If the denominator degree is larger, the ratio tends to zero. If the numerator degree is larger, polynomial growth remains; its sign and direction must still be checked. A memorized degree rule should summarize this calculation rather than replace it.
For radicals, , not always . Therefore tends to as and as . Forgetting the absolute value changes the second answer.
Exercises
- Evaluate .
- Evaluate .
- Find and justify the factor .
- Use squeezing to evaluate .
- Find the limits of at positive and negative infinity.
- Explain why the quotient law cannot establish directly.
- Find .
- Give an example showing that can exist even when neither factor has a finite limit.
Worked solutions
- Factor the numerator as ; the limit is .
- The conjugate gives near zero, with limit .
- Write . The bracket tends to in radians, giving .
- The absolute value is bounded by , so the limit is zero.
- Dividing by gives in either direction. The nonzero denominator limit justifies the quotient step.
- The denominator limit is zero, violating that law’s nonzero-denominator hypothesis.
- Write . For negative , , giving .
- Let and for . Their product has limit at zero, although the first factor has no finite limit there. To make neither factor converge, take and ; their product is identically while both factors oscillate.