Skip to content
Topos Uranos

The fundamental theorem of calculus

Summary

This lesson unites the two integrals of the course: the indefinite one from the unit on antiderivatives, defined with derivatives, and the definite one from this unit, defined with sums. The fundamental theorem of calculus states that the integral of a function with a variable upper limit is continuous and that, at each point where the function is continuous, its derivative is the value of the function. It follows that every function continuous on an arbitrary interval has an antiderivative, as the unit on antiderivatives and Calculus I promised. Conversely, Barrow’s rule computes the integral of an integrable function with any function whose derivative is the integrand: it is proved with Lagrange’s mean value theorem on each subinterval of a partition, without assuming the integrand continuous, and, for continuous integrands, also with the corollary on functions with the same derivative. With it, the change of variable and integration by parts carry over to the definite integral, the integral with both limits variable is differentiated with the chain rule, and some limits of sequences are computed by reading them as Riemann sums.

Learning objectives

  1. Prove the fundamental theorem of calculus on an arbitrary interval and deduce that every function continuous on an interval has an antiderivative.
  2. Prove Barrow’s rule without assuming the integrand continuous, and recognize with examples which hypotheses cannot be omitted.
  3. Use the change of variable and integration by parts in the definite integral, with their exact hypotheses, and differentiate integrals with variable limits.
  4. Compute limits of sequences by reading them as Riemann sums of an integral.

The integral with a variable limit

The unit on antiderivatives defined the antiderivative of a function with the derivative, and this unit defined the integral with sums of rectangles. The lesson on the properties of the definite integral proved that the integral with a variable upper limit is Lipschitz. If the function is continuous at a point, the integral over a small interval starting there is almost the area of a rectangle, and the integral with a variable limit is differentiable there. For the result to hold on an arbitrary interval (open, closed, half-open or unbounded), it suffices to require the function to be integrable on each closed and bounded interval contained in it.

TheoremFundamental theorem of calculus

Let II be an interval with more than one point, ff a function defined on II and integrable on each closed and bounded interval with more than one point contained in II, and cc a point of II. Let FF be the function defined on II by the following formula:

F(x)=∫cxf(t) dtF(x) = \int_c^x f(t) \, \mathrm{d}t

Then FF is Lipschitz on each closed and bounded interval contained in II, and continuous at each point of II. Moreover, if ff is continuous at a point x0x_0 of II, then FF is differentiable at x0x_0, and F′(x0)=f(x0)F'(x_0) = f(x_0); if x0x_0 is an endpoint of II that belongs to II, the derivative is the one-sided one. The difference quotient of FF is the average of ff between x0x_0 and xx, and if ff varies little near x0x_0, that average is close to f(x0)f(x_0). The figure shows, for a positive function, the region whose area is F(x0)F(x_0) and the strip added on moving to xx, inside a rectangle of height BB; the same strip with xx closer; the rectangle of height f(x0)f(x_0) on the base of the strip; and, with xx even closer, the strip between the heights f(x0)−ε2f(x_0) - \frac{\varepsilon}{2} and f(x0)+ε2f(x_0) + \frac{\varepsilon}{2}.

Proof

  1. x,y∈[p,q]→∣F(y)−F(x)∣≤B∣y−x∣\dato{lip}{x, y \in \icc{p}{q} \rightarrow \abs{F(y) - F(x)} \leq \resaltar{B \abs{y - x}}}

    Let [p,q]\icc{p}{q} be a closed and bounded interval contained in II, with p<qp < q, and let JJ be the closed interval with endpoints min⁡(p,c)\min(p, c) and max⁡(q,c)\max(q, c), contained in II because II is an interval. The function ff is integrable on JJ, and therefore bounded, because the definition of an integrable function, from the lesson on the Riemann integral, applies only to bounded functions: ∣f(t)∣≤B\abs{f(t)} \leq B on JJ. On JJ, FF is the function of the lemma stating that the integral with a variable limit is Lipschitz, from the lesson on the properties of the definite integral, with g=fg = f; the lemma makes it Lipschitz on JJ, with constant BB, and in particular on [p,q]\icc{p}{q}.

  2. ∣x−x0∣<δ→∣F(x)−F(x0)∣≤B∣x−x0∣≤B εB+1<ε\abs{x - x_0} < \delta \rightarrow \abs{F(x) - F(x_0)} \leq B \abs{x - x_0} \leq \frac{B \, \varepsilon}{B + 1} < \resaltar{\varepsilon}

    Let x0x_0 be a point of II. If x0x_0 is not the least point of II, let pp be a point of II less than x0x_0, and if it is, let p=x0p = x_0; likewise, let qq be a point of II greater than x0x_0, or q=x0q = x_0 if x0x_0 is the greatest. Since II has more than one point, p<qp < q; let BB be the constant of the preceding move for [p,q]\icc{p}{q} and rr the smaller of the positive numbers among x0−px_0 - p and q−x0q - x_0. Every xx of II with ∣x−x0∣<r\abs{x - x_0} < r lies in [p,q]\icc{p}{q}: if x<x0x < x_0, then x0x_0 is not the least point of II, and x>x0−r≥px > x_0 - r \geq p; if x>x0x > x_0, likewise, x<qx < q. Given ε>0\varepsilon > 0, let δ\delta be the smaller of rr and εB+1\frac{\varepsilon}{B + 1}, which depends on ε\varepsilon and on x0x_0. For every xx of II:

  3. F(x)−F(x0)−f(x0) (x−x0)=∫x0x(f(t)−f(x0)) dt\dato{dif}{F(x) - F(x_0) - f(x_0) \, (x - x_0) = \int_{x_0}^{x} \left(f(t) - f(x_0)\right) \, \mathrm{d}t}

    Now let ff be continuous at x0x_0: for every ε>0\varepsilon > 0 there is a δ>0\delta > 0 such that ∣f(t)−f(x0)∣<ε2\abs{f(t) - f(x_0)} < \frac{\varepsilon}{2} for every tt of II with ∣t−x0∣<δ\abs{t - x_0} < \delta. Let xx be a point of II with 0<∣x−x0∣<δ0 < \abs{x - x_0} < \delta. The closed interval whose endpoints are the least and the greatest of cc, x0x_0 and xx is contained in II, has more than one point and ff is integrable on it; by additivity for three points in any order, from the lesson on the properties of the definite integral, F(x)−F(x0)F(x) - F(x_0) is the integral of ff between x0x_0 and xx. The integral of the constant f(x0)f(x_0) between x0x_0 and xx is f(x0) (x−x0)f(x_0) \, (x - x_0), by the proposition on the three integrals from the definition, from the lesson on the Riemann integral, and by the definition of the integral with reversed limits if x<x0x < x_0. By the linearity of the integral, which also holds with reversed limits because one multiplies by −1-1:

  4. ∣F(x)−F(x0)x−x0−f(x0)∣≤ε2<ε→F′(x0)=f(x0)\abs{\frac{F(x) - F(x_0)}{x - x_0} - f(x_0)} \leq \frac{\varepsilon}{2} < \varepsilon \rightarrow \resaltar{F'(x_0) = f(x_0)}

    Every tt of the closed interval with endpoints x0x_0 and xx lies in II and satisfies ∣t−x0∣≤∣x−x0∣<δ\abs{t - x_0} \leq \abs{x - x_0} < \delta, so that the integrand has absolute value less than ε2\frac{\varepsilon}{2}. The corollary on bounds for the integral, with −ε2-\frac{\varepsilon}{2} and ε2\frac{\varepsilon}{2} as bounds, bounds the integral in absolute value by ε2∣x−x0∣\frac{\varepsilon}{2} \abs{x - x_0}, also if x<x0x < x_0, because reversing the limits only changes the sign; and one divides by ∣x−x0∣\abs{x - x_0}. Since II is an interval with more than one point, x0x_0 is an accumulation point of II, and the inequality is the definition of the derivative at x0x_0, from the lesson on the derivative as a limit; if x0x_0 is an endpoint of II, there are points xx on one side only, and by the proposition on the derivative and one-sided derivatives, from the same lesson, it is the one-sided derivative.

Differentiating the integral with respect to its upper limit returns the integrand. With the variable limit below, the integral changes sign: if ff is continuous at x0x_0, the derivative of ∫xcf(t) dt=−F(x)\int_x^c f(t) \, \mathrm{d}t = -F(x) at x0x_0 is −f(x0)-f(x_0). Continuity at x0x_0 cannot be omitted: the solved problem “Where the integrand is not continuous” shows a function with a jump whose integral is not differentiable and another whose integral is differentiable with a derivative different from the value of the function.

CorollaryEvery function continuous on an interval has an antiderivative

Let ff be a function continuous on an interval II with more than one point and cc a point of II. Then the function F(x)=∫cxf(t) dtF(x) = \int_c^x f(t) \, \mathrm{d}t is an antiderivative of ff on II, the only one that vanishes at cc, and the antiderivatives of ff on II are the functions F+CF + C, with CC real. Continuity gives at once integrability on each closed interval and the derivative at each point.

Proof

  1. (∀x∈I)(F′(x)=f(x)),F(c)=∫ccf=0(\forall x \in I)\left(F'(x) = f(x)\right), \qquad F(c) = \int_c^c f = \resaltar{0}

    For each closed and bounded interval [p,q]\icc{p}{q} contained in II, with p<qp < q, the restriction of ff to [p,q]\icc{p}{q} is continuous, and it is integrable by the theorem stating that continuous functions are integrable, from the lesson on integrable functions. The fundamental theorem applies, and since ff is continuous at each point of II, FF is differentiable at each point of II, with the one-sided derivative at the endpoints that belong to II: this is the definition of an antiderivative of a function on an interval, from the lesson on antiderivatives and the indefinite integral.

  2. (∀x∈I)(G′(x)=f(x))⇔(∃C∈R)(∀x∈I)(G(x)=F(x)+C)(\forall x \in I)\left(G'(x) = f(x)\right) \Leftrightarrow (\exists C \in \mathbb{R})(\forall x \in I)\left(G(x) = F(x) + C\right)

    The theorem on the antiderivatives of a function on an interval, from the same lesson, gives the family of antiderivatives; one of them, F+CF + C, vanishes at cc if and only if C=0C = 0.

This corollary fulfills what the unit on antiderivatives left pending. The function e−x2e^{-x^2} is continuous on R\mathbb{R}, and ∫0xe−t2 dt\int_0^x e^{-t^2} \, \mathrm{d}t is an antiderivative of it on the whole real line, although none of its antiderivatives can be written with the elementary functions. In the lesson on integration by parts, solving for the antiderivatives SS and KK of exsin⁡xe^x \sin x and of excos⁡xe^x \cos x assumed that they exist; now this is ensured in advance, because both integrands are continuous on R\mathbb{R}. The lesson on trigonometric integrals built by hand an antiderivative of 1+cos⁡x\sqrt{1 + \cos x} on the whole real line; the corollary guarantees one for every function continuous on R\mathbb{R}. Continuity is sufficient, not necessary: the solved problem “Changing a single value”, from the lesson on antiderivatives and the indefinite integral, showed a discontinuous function with an antiderivative. And integrability is not enough: the sign function is integrable on each closed interval, by the theorem on bounded functions with finitely many discontinuities, and by the proposition stating that the sign function has no antiderivative on the real line, from the lesson on antiderivatives, it has none on R\mathbb{R}.