Archimedean Clayton copula entropy

The place for a comment was too short for the following, so this became an answer.

Well, removing a factor (namely the factor that makes live too short to compute...) simplifies things in a high measure, but we still have a mess... For the above i see only some first steps, but then things still get complicated. The $\theta$ is hard to type, there will be a $t$ instead. We have $$ \begin{aligned} H &=- \iint_{[0,1]^2} \frac {(1+t)(uv)^{-t}} {(u^{-t} + v^{-t} -1)^{2+1/t}}\;\ln \frac {(1+t)(uv)^{-t-1}} {(u^{-t} + v^{-t} -1)^{2+1/t}} \; \frac {du}u \; \frac {dv}v \\ &= \iint_{I^2} \frac {(1+t)UV} {(U + V -1)^{2+1/t}} \; \ln \frac {(1+t)(UV)^{-(1+t)/t}} {(U + V-1)^{2+1/t}} \; \frac 1{t^2} \; \frac {dU}U \; \frac {dV}V \\ &= \frac {1+t}{t^2} \iint_{I^2} \frac 1{(U + V -1)^{2+1/t}} \;\ln \frac {(1+t)(UV)^{-(1+t)/t}} {(U + V - 1)^{2+1/t}} \; dU\; dV\ . \end{aligned} $$ We substituted $U=u^{-t}$, $V=v^{-t}$, so $\frac {dU}U$ is $(-t)\frac {du}u$, and $\frac {dV}V$ is $(-t)\frac {dv}v$, so that we obtain a better looking expression.

The integral is now over $I^2$, where $I$ is $[1,\infty]$, because of the sign of $-t$ in $U=u^{-t}$.

Now under the logarithm we split four factors. And have to compute correspondingly four integrals.

• The integral in $\ln(t+1)$ is the simplest. It is in fact an integral in $W=(U+V)\ge 2$. We have working with $a=2+1/t$ $$\iint_{I^2} \frac 1{(U + V -1)^a} \; dU\; dV = \int_2^\infty\frac{W-2}{(W-1)^a}\; dW \\ =\frac 1{a^2-3a+2} =\frac 1{a-2}-\frac 1{a-1} = t-\frac t{1+t} \ . $$
• The integrals in $\ln U$ and $\ln V$ are equal, it is enough to compute only one of them. Again pass from $(U,V)$ to $(U,W)$, where $W=U+V$ formally. So we have to integrate something like $$ \iint_{\substack{1\le U<\infty\\2\le W<\infty\\1+U\le W}} \frac 1{(W -1)^a} \ln U \; dU\; dW $$ We can first integrate in $U$ from $1$ to $W-1$, but some $\log(W-2)$ will come into play and the work begins.
• The final integral can also be arranged in a better form, passing from $(U,V)$ to $(U,W)$ as above, but then we have to compute something like $$ \int_{2}^\infty \frac{W-2}{(W-1)^a}\ln(W-1)\;dW\ . $$

Now try to apply integration by parts to get rid of the logarithmic term. For special values of $t$ (and $a$) this may be computed, but i will stop here.