We received solutions from Amrit, Adithya, Guruvignesh and Agathiyan, all from Hymers College, and Pablo from Kings College Alicante. Here is Pablo's solution:
As you can see in Graph I, the areas in purple and blue add up to the big rectangle, bf(b), minus af(a), the small rectangle.
There is a change of variable in f^{-1}(t) because the area isn't taken
from the x-axis, but from the t-axis. f^{-1}(x) is the inverse of f(x). It is obtained either by switching variables and rearranging the equation again or by reflecting it in the line y=x.
The reason why f(x) must be increasing in the interval a \leq x \leq b is shown in Graph II. If it weren't, if at some point it reached a turning point and began to decrease, the area (orange) at the top would not be accounted for.
If f(x) is always decreasing in the interval a \leq x \leq b, a similar formula comes out, but through more algebraic manipulation. The area called A in Graph III is:
A = bf(a) - bf(b) - \int_{f(b)}^{f(a)} f^{-1}(t) d t
but also, A = bf(a) - af(a) \int_a^b f(x) d x
so bf(a) - bf(b) - \int_{f(b)}^{f(a)} f^{-1}(t) d = bf(a) - af(a) \int_a^b f(x) d x
\int_a^b f(x) d x = bf(a) - af(a) - \left[bf(a) - bf(b) - \int_{f(b)}^{f(a)} f^{-1}(t) d t\right]
\int_a^b f(x) d x = bf(b) - af(a) + \int_{f(b)}^{f(a)} f^{-1}(t) d t
The difference with the first formula is that \int_{f(b)}^{f(a)} f^{-1}(t) d t is added, not subtracted. Additionally, the boundaries are swapped, as f(b) is less than f(a).
First, we can find the value of \int_1^4\sqrt t dt the conventional
way: \eqalign{\int_1^4 \sqrt x dx &= \left[\frac23 x^\frac32\right]_1^4 \\
&= \frac23 ( 4^\frac32 - 1^\frac32) \\
&= \frac{14}3}