# Techniques for evaluating probability integral

Consider the integral of a normal distribution: $$\int_a^b f(x)\,\mathrm d x=c$$

and a second integral for the expected value: $$\int_a^b x\cdot f(x)\,\mathrm dx$$ Since you know the first integral is equal to $c$, what is a good way to evaluate the second integral to find the expected value? Integration by parts doesn't seem to work.

-

$$\int_a^b \frac1c f(x)\,dx = 1.$$ Therefore $\dfrac1c f(x)$ is a probability density function on the interval $[a,b]$. It is the conditional probability density of a normal random variable given that it's in that interval. So $$\int_a^b \frac1c x f(x)\,dx$$ is the conditional expected value of a normal random variable given that it's in that interval.
Later note: To see why this is the conditional distribution, consider the problem of finding the conditional probability $\Pr(X\in S\mid a\le X\le b)$, where $X$ is a random variable whose density is $f$: $$\Pr(X\in S\mid a\le X\le b) = \frac{ \Pr(X\in S\text{ and }a\le X\le b ) }{\Pr(a\le X\le b)}$$ $$= \frac{ \Pr(X\in S\text{ and }a\le X\le b ) }{c} = \frac{\int_{S\cap [a,b]} f(x)\,dx}{c} = \int_{(\text{the relevant set})} \frac1c f(x)\,dx.$$