Let $f(x,y)=e^{-xy}\sin x\sin y$ for $(x,y)\in\mathbb{R_{\geq 0}^2}$ and $f(x,y)=0$ wherever else.
I want to prove that
$$\int_{\mathbb{R}}\int_{\mathbb{R}} f(x,y)dxdy$$ exists but $$\iint_{\mathbb{R}^2} f(x,y)dxdy$$ does not, the first as Lebesgue integrals over $\mathbb{R}$, and the second as a Lebesgue integral over $\mathbb{R}^2$. Furthermore I want to prove that $$\int_{\mathbb{R}}\int_{\mathbb{R}} |f(x,y)|dxdy$$ does not exist (In a way that does not involve the inexistence of the second integral as hypothesis).
For the first integral, I considered this:
First, if $y=0$, the function is inmediately Lebesgue integrable with respect to $x$. If $y\neq 0$, let's consider that $$|f(x,y)|=|e^{-xy}\sin x\sin y|=e^{-xy}|\sin x||\sin y|\leq |\sin y|e^{-xy}$$ and in this case $f(x,y)$ is dominated by a Lebesgue integrable function $|\sin y| e^{-xy}$ which has an integral $$\int_0^\infty |\sin y|e^{-xy}dx=|\sin y| \int_0^\infty e^{-xy}dx=\frac{|\sin y|}{y}$$ so the inner integral $F(y)=\int_0^\infty f(x,y) dx$ exists and
$$|F(y)|=\left|\int_{0}^{\infty} f(x,y)dx\right|\leq \frac{|\sin y|}{y}.$$
I want to prove that $F(y)$ is Lebesgue-Integrable with respect to $y$, but even if it's dominated by $\frac{|\sin y|}{y}$, it is not enough, since $\frac{|\sin y|}{y}$ is not Lebesgue-Integrable.
How can I do this?
For $y>0$, $\int_0^\infty e^{-xy}\sin x\,dx$ is the imaginary part of $\int_0^\infty e^{-(y-i)x}\,dx=1/(y-i)$. Thus $\int_0^\infty e^{-xy}\sin x\,dx=1/(1+y^2)$. The iterated integral of $f$ is thus $\int_0^\infty(\sin y)/(1+y^2)\,dy$ which is finite.
Here's an incomplete attempt to prove that $f$ is not an $L^1$ function.
We can write the (possibly divergent) double integral of $|f|$ as $$\begin{align} \int_0^\infty\int_0^\infty|\sin x\sin y|e^{-xy}\,dx\,dy &=\sum_{m,n=0}^\infty\int_0^\pi\int_0^\pi \sin x\sin y\, e^{-(x+m\pi)(y+n\pi)}\,dx\,dy\\ &=\int_0^\pi\int_0^\pi\sin x\sin y\sum_{m,n=0}^\infty e^{-(x+m\pi)(y+n\pi)}\,dx\,dy\\ &=\int_0^\pi\int_0^\pi\sin x\sin y\,G(x,y)\,dx\,dy\end{align}$$ say, where $$G(x,y)=\sum_{m,n=0}^\infty e^{-(x+m\pi)(y+n\pi)}=\sum_{m=0}^\infty \frac{e^{-xy}e^{-m\pi y}}{1-e^{-\pi(x+m\pi)}}.$$ All we need to do now is to show that $G(x,y)$ tends to infinity "quickly" enough as $(x,y)\to0$ to ensure the integral diverges. I'm not seeing how at the moment!