Seemingly Simple Probabilistic Inequality Question

30 Views Asked by At

Let's assume all random variables here take positive values. All RVs will be represented in capital letters and constants in lower case. First, we know that $$\big| \mathbb{E}[A] - \mathbb{E}[B] \big| \le k$$

Now we also know that for RVs $M$ and $N$, $\mathbb{E}[M] = 1$ and $\mathbb{E}[N]=1$. Also $M$ and $N$ are bounded. I am trying to show (or find a counter-example) that $$\big| \mathbb{E}[AM] - \mathbb{E}[BN] \big| \le kc$$ In other words $\big| \mathbb{E}[AM] - \mathbb{E}[BN] \big|$ is on the order of $\big| \mathbb{E}[A] - \mathbb{E}[B] \big|$.

So far the most promising direction seems to be use Hölder's inequality where we have that $$\mathbb{E}[|AM|]\le\mathbb{E}[|A|]\sup|M|\le \mathbb{E}[|A|]\cdot h$$ and similarly $$\mathbb{E}[|BN|]\le\mathbb{E}[|B|]\sup|N|\le \mathbb{E}[|B|]\cdot l$$

However, the subtraction is throwing things off with my straight-forward calculation. I was hoping someone had a better idea for how to progress, or of course a counter-example?