The Derivative

We begin with the definition of the derivative of a function.
Let IR be an interval and let cI. We say that f:IR is differentiable at c or has a derivative at c if limxcf(x)f(c)xc exists. We say that f is differentiable on I if f is differentiable at every point in I.
By definition, f has a derivative at c if there exists a number LR such that for every ε>0 there exists δ>0 such that if |xc|<δ then |f(x)f(c)xcL|<ε. If f is differentiable at c, we will denote limxcf(x)f(c)xc by f(c), that is, f(c)=limxcf(x)f(c)xc. The rule that sends c to the number f(c) defines a function on a possibly smaller subset JI. The function f:JR is called the derivative of f.
Let f(x)=1/x for x(0,). Prove that f(x)=1x2.
Let f(x)=sin(x) for xR. Prove that f(x)=cos(x).
Recall that sin(x)sin(c)=2sin(xc2)cos(x+c2) and that limx0sin(x)x=1. Therefore, limxcsin(x)sin(c)xc=limxc2sin(xc2)cos(x+c2)xc=limxc(sin(xc2)xc2)cos(x+c2)=1cos(c)=cos(c). Hence f(c)=cos(c) for all c and thus f(x)=cos(x).
Prove by definition that f(x)=x1+x2 is differentiable on R.
We have that f(x)f(c)xc=x1+x2c1+c2xc=x(1+c2)c(1+x2)(1+x2)(1+c2)(xc)=1cx(1+c2)(1+x2). Now limxcf(x)f(c)xc=1c2(1+c2)2. Hence, f(c) exists for all cR and the derivative function of f is f(x)=1x2(1+x2)2.
Prove that f(x)=α if f(x)=αx+b.
We have that f(x)f(c)=αxαc=α(xc). Therefore, limxcf(x)f(c)xc=α. This proves that f(x)=α for all xR.
Compute the derivative function of f(x)=|x| for xR.
If x>0 then f(x)=x and thus f(x)=1 for x>0. If x<0 then f(x)=x and therefore f(x)=1 for x<0. Now consider c=0. We have that f(x)f(c)xc=|x|x. We claim that the limit limx0|x|x does not exist and thus f(0) does not exist. To see this, consider xn=1/n. Then (xn)0 and f(xn)=1 for all n. On the other hand, consider yn=1/n. Then (yn)0 and f(yn)=1. Hence, limnf(xn)limnf(yn), and thus the claim holds by the Sequential criterion for limits. The derivative function f of f is therefore defined on A=R{0} and is given by f(x)={1,x>01,x>0. Hence, even though f is continuous at every point in its domain R, it is not differentiable at every point in its domain. In other words, continuity is not a sufficient condition for differentiability.
Suppose that f:IR is differentiable at c. Then f is continuous at c.
To prove that f is continuous at c we must show that limxcf(x)=f(c). By assumption limxcf(x)f(c)xc=f(c) exists, and clearly limxc(xc)=0. Hence we can apply the Limits laws and compute limxcf(x)=limxc(f(x)f(c)+f(c))=limxc(f(x)f(c)(xc)(xc)+f(c))=f(c)0+f(c)=f(c) and the proof is complete.
Let f:IR and g:IR be differentiable at cI. The following hold:
  1. If αR then (αf) is differentiable and (αf)(c)=αf(c).
  2. is differentiable at and .
  3. is differentiable at and .
  4. If then is differentiable at and
Parts (i) and (ii) are straightforward. We will prove only (iii) and (iv). For (iii), we have that Now because is differentiable at . Therefore, To prove part (iv), since , then there exist a -neighborhood such that for all . If then Since , it follows that and the proof is complete.
We now prove the Chain Rule.
Let and be functions such that and let . If exists and exists then exists and .
Suppose that there exists a neighborhood of where . Otherwise, the composite function is constant in a neighborhood of , and then clearly differentiable at . Consider the function defined by Now Hence, is differentiable at and therefore is at . Now, and therefore Therefore, as claimed.
Compute if Where is continuous?
When , is the composition and product of differentiable functions at , and therefore is differentiable at . For instance, on , the functions , and are differentiable at every . Hence, if we have that Consider now . If exists it is equal to Using the Squeeze Theorem, we deduce that . Therefore, From the above formula obtained for , we observe that when is continuous since it is the product/difference/composition of continuous functions. To determine continuity of at consider . Consider the sequence , which clearly converges to . Now, . Now, for all and therefore . The sequence does not converge and therefore does not exist. Thus, is not continuous at .
Compute if Where is continuous?
When , is the composition and product of differentiable functions, and therefore is differentiable at . For instance, on , the functions , and are differentiable on . Hence, if we have that Consider now . If exists it is equal to and using the Squeeze Theorem we deduce that . Therefore, When , is continuous since it is the product/difference/composition of continuous functions. To determine continuity of at we consider the limit . Now using the Squeeze Theorem, and similarly using the Squeeze Theorem. Therefore, exists and is equal to , which is equal to . Hence, is continuous at , and thus continuous everywhere.
Consider the function Show that .

Exercises

Use the definition of the derivative of a function to find if . Clearly state the domain of .
Use the definition of the derivative of a function to find if . Clearly state the domain of .
Let be defined by
  1. Show that is differentiable at and find .
  2. Prove that if then is not differentiable at .
Let for . Determine whether , exist and if yes find them. Hint: Consider writing as a piecewise function and use the definition of the derivative.
If is differentiable at , explain why Give an example of a function and a number such that exists but does not exist.

The Mean Value Theorem

Let be a function and let .
  1. We say that has a relative maximum at if there exists such that for all .
  2. We say that has a relative minimum at if there exists such that for all .
A point is called a critical point of if . The next theorem says that a relative maximum/minimum of a differentiable function can only occur at a critical point.
Let be a function and let be an interior point of . Suppose that has a relative maximum (or minimum) at . If is differentiable at then is a critical point of , that is, .
Suppose that has a relative maximum at ; the relative minimum case is similar. Then for , it holds that for and some . Consider the function defined by for and . Then the function is continuous at because . Now for it holds that and therefore . Similarly, for it holds that and therefore . Thus .
If has a relative maximum (or minimum) at then either or does not exist.
The function has a relative minimum at , however, is not differentiable at .
Let be continuous on and differentiable on . If then there exists such that .
Since is continuous on it achieves its maximum and minimum at some point and , respectively, that is for all . If is constant then for all . If is not constant then . Since it follows that at least one of and is not contained in , and hence by Theorem 6.2.2 there exists such that .
We now state and prove the main result of this section.
Let be continuous on and differentiable on . Then there exists such that .
If then the result follows from Rolle's Theorem ( for some ). Let be the line from to , that is, and define the function for . Then and , and thus . Clearly, is continuous on and differentiable on , and it is straightforward to verify that . By Rolle's Theorem, there exists such that , and therefore .
Let be continuous on and differentiable on . If for all then is constant on .
Let . Now restricted to satisfies all the assumptions needed in the Mean Value Theorem. Therefore, there exists such that . But and thus . This holds for all and thus is constant on .
Show by example that Theorem 6.2.7 is not true for a function if is not a closed and bounded interval.
If are continuous and differentiable on and for all then for some constant .
Use the Mean Value theorem to show that for all .
Suppose that and let so that . By the MVT, there exists such that , that is . Now and therefore . Therefore, . The case can be treated similarly.
The function is increasing if whenever . Similarly, is decreasing if whenever . In either case, we say that is monotone.
The sign of the derivative determines where is increasing/decreasing.
Suppose that is differentiable.
  1. Then is increasing if and only if for all .
  2. Then is decreasing if and only if for all .
Suppose that is increasing. Then for all with it holds that and therefore . Hence, this proves that for all . Now suppose that for all . Suppose that . Then by the Mean Value Theorem, there exists such that . Therefore, since it follows that . Part (ii) is proved similarly.

Exercises

Use the Mean Value Theorem to show that In general, suppose that is such that exists on and is continuous on . Prove that is Lipschitz on .
Give an example of a uniformly continuous function on that is differentiable on but whose derivative is not bounded on . Justify your answer.
Let be an interval and let be differentiable on . Prove that if for then is strictly increasing on .
Let be continuous on and differentiable on . Show that if then exists and equals . Hint: Use the definition of , the Mean Value Theorem, and the Sequential Criterion for limits.
Let be continuous and suppose that exists on . Prove that if for then is strictly increasing on .
Suppose that is continuous on and differentiable on . We proved that if for all then is constant on . Give an example of a function such that for all but is not constant on .
Let be differentiable. Prove that if on then is injective.

Taylor's Theorem

Taylor's theorem is a higher-order version of the Mean Value Theorem and it has abundant applications in numerical analysis. Taylor's theorem involves Taylor polynomials which you are familiar with from calculus.
Let and suppose that is such that the derivatives , , ,, exist for some positive integer . Then the polynomial is called the th order Taylor polynomial of based at . Using summation convention, can be written as
By construction, the derivatives of and up to order are identical at (verify this!): It is reasonable then to suspect that is a good approximation to for points near . If then the difference between and is and we call the th order remainder based at . Hence, for each , the remainder is the error in approximating with . You may be asking yourself why we would need to approximate if the function is known and given. For example, if say then why would we need to approximate say since any basic calculator could easily compute ? Well, what your calculator is actually computing is an approximation to using a (rational) number such as and using a large value of for accuracy (although modern numerical algorithms for computing trigonometric functions have superseded Taylor approximations but Taylor approximations are a good start). Taylor's theorem provides an expression for the remainder term using the derivative .
Let be a function such that for some the functions are continuous on and exists on . Fix . Then for any there exists between and such that where
If then and then can be chosen arbitrarily. Thus, suppose that , let , and define the function by Since exists on then exists on . Moreover, since for then for . Now and therefore since by Rolle's theorem there exists in between and such that . Now we can apply Rolle's theorem to since and , and therefore there exists in between and such that . By applying this same argument repeatedly, there exists in between and such that . Now, and since then from which we conclude that and the proof is complete.
Consider the function given by . Use based at to estimate and give a bound on the error with your estimation.
Note that and so the estimate of using is . To determine we need . We compute Therefore, Now and therefore The error is which is unknown but we can approximate it using Taylor's theorem. To that end, by Taylor's theorem, for any there exists in between and such that Therefore, for , there exists such that Therefore, a bound for the error is since .
Let be the sine function, that is, .
  1. Approximate using centered at and give a bound on the error.
  2. Restrict to a closed and bounded interval of the form . Show that for any there exists such that if then for all .
(a) It is straightforward to compute that and . Thus, by Taylor's theorem for any there exists in between and such that The estimate for is By Taylor's theorem, there exists such that and Now since for all , we have (b) Since has derivatives of all orders, for any we have by Taylor's theorem that where is in between and . Now, the derivative of of any order is one of or , and therefore . Since then and therefore . Therefore, for all we have Consider the sequence . Applying the Ratio test we obtain Therefore, by the Ratio test . Hence, for any there exists such that for all . Therefore, for all we have that for all .
Taylor's theorem can be used to derive useful inequalities.
Prove that for all it holds that
Let . Applying Taylor's theorem to at we obtain where and is in between and . Now, if then and then , from which it follows that . If on the other hand then and then , from which it follows that . Hence, the inequality holds for . Now if then Hence the inequality holds for all .

Exercises

Use Taylor's theorem to prove that if then Then use these inequalities to approximate and , and for each case determine a bound on the error of your approximation.
Let be such that exists for all and for all (such a function is called infinitely differentiable on ). Suppose further that there exists such that for all and all . Let be the th order Taylor polynomial of centered at . Let , where . Prove that for any fixed there exists such that for it holds that for all . Hint: is continuous on for every .
Euler's number is approximately . Use Taylor's theorem at on and the estimate to show that, for all ,
Let be the cosine function . Approximate using centered at and give a bound on the error of your estimation.