I want to calculate the following product : $$\prod_{p\ prime\ ,\ p\le 10^{10}} 1-\frac{1}{p}$$

I know the approximation formula $$\frac{e^{-\gamma}}{\ln(10^{10})}$$ where $\gamma$ is the Euler-Mascheroni-constant. The result should be good to $12$ decimal digits.

Is there an efficient way (not brute force by determining all primes, which takes long with PARI/GP) to calculate this product with high precision ?


Call the product $P$, and take logs to get $\log(P)=\sum_{i=1}^n\log(1-1/p_i)=-\sum_{i=1}^n\frac{1}{p_i}-\frac{1}{2}\sum_{i=1}^n\frac{1}{p^2_i}+...\approx-\sum_{i=1}^n\frac{1}{p_i}-\gamma$

So to get a good approximation, use ever-better approximations of $\sum_{i=1}^n\frac{1}{p_i}$. As an example, the Meissel Merten constant is defined as $M:=\lim_{n\rightarrow\infty}\sum_{i=1}^n1/p_i-\ln\ln(n)\approx 0.26149$.

If you're still getting error, you can look into the error for finite sum approximations of $\gamma=\sum_{m=2}^\infty (-1)^m\frac{\zeta(m)}{m}$.

  • 2
    $\begingroup$ I do not get good approximations this way for $n=\pi(10^6)$. Please work out examples, perhaps I missed something. Additionally, the sum requires also the primes because the convergence towards Merten's constant is very very slow. $\endgroup$ – Peter Apr 3 '18 at 21:46
  • $\begingroup$ You can find a detailed discussion on how to better approximate $B$ to any accuracy here: pdfs.semanticscholar.org/0ff0/… $\endgroup$ – Alex R. Apr 3 '18 at 21:54

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.