Suppose that I have a function $g(x): R \rightarrow R$ and suppose that the pdf of $x$ is $f(x)$. To avoid cumbersome numerical integration I approximate the expected value of $g(x)$ as

$\int_{x = -\infty}^{\infty} g(x) f(x) = \frac{1}{R} \sum_{r=1}^R g(x_r)$,

with the right hand side evaluated in $R$ random draws from $f(x)$.

Is it somehow possible to apply a similar trick when I have $\overline{x}$ as the lower bound of the integration? (the domain of $g(x)$ remains $R$)

$\int_{x = \overline{x}}^{\infty} g(x) f(x)$


1 Answer 1


Yes, just choose those $R$ draws that are more than $\bar{x}$ (but still divide by $R$). The reason this works mathematically is you can write the bounded below integral as an unbounded integral with an indicator function.

$$ \int_{\bar{x}}^{\infty} g(x) f(x) dx = \int_{-\infty}^{\infty} g(x) I_{\{x > \bar{x} \}} f(x) dx \approx \dfrac{1}{R}\sum_{r = 1}^{R} g(x_r) I_{\{x_r > \bar{x} \}}.$$

So if the draw from $f$ is more than $\bar{x}$, you evaluate $g$, otherwise, you throw it away.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.