I am trying to change the order of integration for the double integral below using the Heaviside step function. I'm following the example given in this answer.

So far:

\begin{eqnarray*} \int_{\frac{1}{4}}^{\frac{1}{2}}\int_{\frac{1}{4}}^{y_1}f\left(y_1, y_2\right)\,{\rm d}y_2\,{\rm d}y_1 & = & \int_{\frac{1}{4}}^{\frac{1}{2}}\left\lbrack\int_{\frac{1}{4}}^{1}\Theta\left(y_1 - y_2 \right) {\rm f}\left(y_1, y_2\right)\,{\rm d}y_2\right\rbrack{\rm d}y_1 \\ & = & \int_{\frac{1}{4}}^{1}\left\lbrack\int_{\frac{1}{4}}^{\frac{1}{2}}\Theta\left(\_\_\_\_\right) {\rm f}\left(y_1, y_2\right)\,{\rm d}y_1\right\rbrack{\rm d}y_2 \end{eqnarray*}

I am not sure what to do with $\Theta\left(y_1 - y_2 \right)$ at this point.


\begin{align} \int_{\frac 14}^{\frac 12} \int_{\frac 14}^{y_1} f(y_1,y_2)dy_2dy_1 = \int_{\frac 14}^{\frac 12} \int_{\frac 14}^{1} \Theta(y_1-y_2)f(y_1,y_2)dy_2dy_1 \\ = \int_{\frac 14}^{1} \int_{\frac 14}^{\frac 12} \Theta(y_1-y_2)f(y_1,y_2)dy_1dy_2 \tag{Fubini} \\ = \int_{\frac 14}^{\frac 12} \int_{y_2}^{\frac 12} f(y_1,y_2)dy_1dy_2 \tag{simplified $\Theta$} \end{align}

Note that the $1$, which is the upper limit of one of the integrals, plays no special role, really. In the last step, when we simplify $\Theta(y_1-y_2)$, this is $1$ only when $y_1 \geq y_2$. But we are only letting $y_1$ vary between $\frac 14$ and $\frac 12$, so if $y_2$ is for example $\frac 23$ then the inner integral will just be zero in the second step. That's why I had to adjust the outer limit also to account for the simplification of $\Theta$. But the answer is right.

  • $\begingroup$ So the upper limit, instead of 1 would depend on the bounds of $y_2$? $\endgroup$ Dec 11 '17 at 5:15
  • $\begingroup$ Yes, that is what I expect it to depend upon. $\endgroup$ Dec 11 '17 at 5:16
  • $\begingroup$ What if the bounds were say, something like: $0 \leq y \leq 1$, $1 \leq x \leq e^{y}$. How do I determine the bounds exactly for the Heavside function? I think that's what I'm having trouble understanding. $\endgroup$ Dec 11 '17 at 5:42
  • $\begingroup$ Oh , I see. Well, it then is a slightly different and more difficult question. I suggest you ask a different question on this site, but link to this answer and the other one for context. By the way, I am still finding it difficult to justify the use of Fubini for the heaviside function. For one, yes the integrals exist on both sides,but does it ensure absolute integrability? I mean, you could do this all the time right?(Even when Fubini does not hold, that is). $\endgroup$ Dec 11 '17 at 5:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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