Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $X_1$ and $X_2$ be random variables (not of the same distribution and not independent). Both have a zero probability of being below $-1$. Their joint density is $\rho(x_1,x_2)$. Also, they both have finite expectations.

Now, define the region $A = \{ (t_1,t_2)\in\mathbb{R}^2 \mid 0\le t_1,t_2<1 \text{ and }t_1+t_2<1 \}$, and define the function $f:A\to\mathbb{R}$ with

$$f(t_1,t_2) = \int_{-1}^\infty\int_{-1}^\infty \log(1+t_1x_1+t_2x_2)\rho(x_1,x_2)\,dx_1dx_2.$$

Can we say something interesting about $f$? For example, does $f$ have at most only one local maximum?

share|improve this question
add comment

1 Answer 1

up vote 2 down vote accepted

It does, because $f$ is strictly concave. Indeed, consider its restriction to a line through $(t_1,t_2)$ in the direction $(v_1,v_2)$. In terms of parameter $s$ this restriction is $$f(t_1+sv_1,t_2+sv_2)=\int_{-1}^\infty\int_{-1}^\infty \log(1+t_1x_1+t_2x_2+s(v_1x_1+v_2x_2))\rho(x_1,x_2)\,dx_1dx_2$$ The one-variable function $s\mapsto \log(1+t_1x_1+t_2x_2+s(v_1x_1+v_2x_2))$ is strictly concave unless $v_1x_1+v_2x_2=0$. Assuming the joint density is indeed a density and not some singular measure, we conclude that for $\rho$-a.e. $(x_1,x_2)$ strict concavity holds. Thus, integration against $\rho$ gives a strictly concave function.

If you want a more explicit argument, take the second derivative in $s$: $$\frac{d^2}{ds^2}f(t_1+sv_1,t_2+sv_2)\\=-\int_{-1}^\infty\int_{-1}^\infty (v_1x_1+v_2x_2)^2 (1+t_1x_1+t_2x_2+s(v_1x_1+v_2x_2))^{-2}\rho(x_1,x_2)\,dx_1dx_2<0$$

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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