Consider the series

$$S_n = a_1 + a_2(1-a_1) + a_3(1-a_2)(1-a_1)+\cdots+ a_n\prod_{i=1}^{n-1}(1-a_i)$$

$$0<a_i< 1 \;\;i=1,\ldots,n$$

For $a_i=a_j=a\;\; \forall i,j$ it is immediate to show that $S_n$ converges to $1$ as $n\to \infty$.

I also simulated drawing the $a_i$'s as i.i.d. Uniform $U(0,1)$ random variables, and $S_n$ again converged to unity pretty convincingly. But I can't determine the needed conditions and prove theoretically convergence to unity, when the $a_i$'s vary (deterministically or as random variables).

I notice that we can write

$$S_n = a_1 + (1-a_1)\Big[a_2 +(1-a_2)\big[a_3+(1-a_3)[\cdots(1-a_{n-1}) a_n]\big]\Big]$$

All these seem eerily familiar but I can't pin down where I have seen these expressions before...

QUESTION: What conditions are needed and how can we prove convergence to unity of the above infinite series?


Your sense of "eerie familiarity" with the sequence $S_n$ was justified - in fact, we have $S_n = -(1-a_1)(1-a_2)(1-a_3)\cdots(1-a_n) + 1$. Clearly as $n$ goes to infinity, under reasonable conditions we can expect the value of the product term in $S_n$ to tend to zero - and so you have your limit!

  • $\begingroup$ This is brilliant (the way you re-wrote the terms of the series). Brilliant because it is simple. Thank you. $\endgroup$ – Alecos Papadopoulos Jan 18 '17 at 21:39

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.