Estimating Many Constants With a Coin

By Shalosh B. Ekhad and Doron Zeilberger


.pdf    .tex   



First Written: Aug. 20, 2026

This version: Aug. 20, 2026

Exclusively published in the Personal Journal of Shalosh B. Ekhad and Doron Zeilberger and arxiv.com


In this charming note, Jim Propp told us how to estimate π (or rather π/4) by tossing a fair coin. In another charming article, F. Thomas Bruss and Dvy Paindaveine, compute many other constants using a loaded coin. Here we generalize it even further, and show the usefulness of Wilf-Zeilberger algorithmic proof theory.


Maple packages


Sample Input and Output for Propp.txt

  • If you want to see exact values of the expectation of RisingFacorial(#H,r)/RisingFacorial(#Tosses,r) when the number of Heads exceeds the number of Tails for the first time for r from 1 to 100, where the coin is loaded with Pr(H)=3/4

    then the input gives the output.

  • If you want to see exact values of the expectation of #H/#Tosses,when the #Heads-#Tails=d+1 for the first time for d from 1 to 70, where the coin is loaded with Pr(H)=3/4

    then the input gives the output.


    Sample Input and Output for MultiPropp.txt

  • If you want to see exact (well, approximate, by truncating the relevant infinite sums after 200 terms) values of the expectation of x[i]/(x[1]+...+x[k]) when travelling in the k dimensions, Manhattan lattice, starting at the originm using a fair k-sided die until you get, for the first time out of
    x[1] ≥ x[2] ≥ ... ≥ x[k] ≥ 0
    followed by estimates obtained by simulation (using 3000 trials, and averaging) for k from 2 to 5 and i from 1 to k,

    then the input gives the output.


    Personal Journal of Shalosh B. Ekhad and Doron Zeilberger

    Doron Zeilberger's Home Page