The Fundamental Theorem of Calculus

Antiderivatives

For example, x2x^2 is an antiderivative of 2x2x, since the derivative of x2x^2 is 2x2x. The functions x2+2x^2 + 2 and x2−2x^2 - 2 are antiderivatives of 2x2x as well. A constant term has derivative 00, so adding one doesn't change the derivative.
This follows from the Mean Value Theorem: two functions with the same derivative on an interval differ by a constant there. Their graphs are vertical translates of each other, with the same slope at every xx.
Three antiderivatives of f(x) = 2x: from top to bottom, y = x² + 2, y = x², and y = x² − 2. At x = 1 each has slope 2.
Each basic derivative formula, read backward, gives an antiderivative. Since the derivative of xn+1x^{n+1} is (n+1)xn(n + 1)x^n, dividing by n+1n + 1 gives an antiderivative of xnx^n. That works for every nn except n=−1n = -1, where it would divide by zero. Since the derivative of cos⁡x\cos x is −sin⁡x-\sin x, an antiderivative of sin⁡x\sin x is −cos⁡x-\cos x.
The row for 1x\frac{1}{x} fills the gap the power rule leaves at n=−1n = -1. For x>0x > 0, the derivative of ln⁡x\ln x is 1x\frac{1}{x}. For x<0x < 0, ln⁡∣x∣=ln⁡(−x)\ln|x| = \ln(-x), whose derivative is −1−x=1x\frac{-1}{-x} = \frac{1}{x}.

The Fundamental Theorem

To see why, divide [a,b][a, b] into nn subintervals of width Δx\Delta x. The change in FF from aa to bb is the sum of its changes over the subintervals. By the Mean Value Theorem, the change over the ii-th subinterval is F′(ci) Δx=f(ci) ΔxF'(c_i)\,\Delta x = f(c_i)\,\Delta x for some cic_i in it. So
F(b)−F(a)=∑i=1nf(ci) ΔxF(b) - F(a) = \sum_{i=1}^{n} f(c_i)\,\Delta x
for every nn. The right side is a Riemann sum, and as n→∞n \to \infty it approaches the integral of ff from aa to bb. The left side doesn't depend on nn, so it equals that integral.
The difference F(b)−F(a)F(b) - F(a) is written with a bar,
F(x)∣ab=F(b)−F(a).F(x)\Big|_{a}^{b} = F(b) - F(a).
Any antiderivative gives the same difference, since the constant cancels: (F(b)+C)−(F(a)+C)=F(b)−F(a)\bigl(F(b) + C\bigr) - \bigl(F(a) + C\bigr) = F(b) - F(a). So we take C=0C = 0.

Absolute values

No row of the table fits an absolute value ∣g(x)∣|g(x)|. When gg is continuous, split [a,b][a, b] where gg changes sign. On each piece ∣g(x)∣|g(x)| is g(x)g(x) or −g(x)-g(x), and the theorem applies to each piece.

Average value

When f≥0f \ge 0, the right side is the area of a rectangle on [a,b][a, b] of height f(c)f(c). In the figure, the part of the region above the rectangle, in darker blue, fills the part of the rectangle above the curve, in amber.
The rectangle of height f(c) on [a, b] has the same area as the region under the curve.
To see why the theorem holds, let mm and MM be the smallest and largest values of ff on [a,b][a, b], which exist by the Extreme Value Theorem. Every Riemann sum ∑f(ci) Δx\sum f(c_i)\,\Delta x lies between ∑m Δx=m(b−a)\sum m\,\Delta x = m(b - a) and M(b−a)M(b - a), so its limit, the integral, does too. Dividing the integral by b−ab - a gives a number between mm and MM. The values mm and MM are taken at points of [a,b][a, b]. The Intermediate Value Theorem on the interval between those two points gives a cc where ff takes that number.
The height f(c)f(c) is a limit of averages of sampled values of ff. Take nn subintervals of width Δx\Delta x and a sample point cic_i in each. Since 1n=Δxb−a\frac{1}{n} = \frac{\Delta x}{b - a}, the average of the nn sampled values is
1n∑i=1nf(ci)=1b−a∑i=1nf(ci) Δx,\frac{1}{n}\sum_{i=1}^{n} f(c_i) = \frac{1}{b - a}\sum_{i=1}^{n} f(c_i)\,\Delta x,
and as n→∞n \to \infty the right side approaches the integral divided by b−ab - a, which is f(c)f(c).
By the Mean Value Theorem for Integrals, ff takes its average value at least once on [a,b][a, b].