Pi has been sequenced to its two quadrillionth bit, and the value has been found to be zero.
Yahoo engineer Tsz Wo Sze announced on his Apache developer page in August that using a MapReduce programe named DistBdp and idle computer cycles in Yahoo's Hadoop clusters, he was able to calculate the two quadrillionth bit of Pi. The technique did not involve a full count of the Pi number sequence up until the two quadrillionth bit, but instead used a series of mathematical tricks to compute a specific part of the Pi sequence. In this case, the two quadrillionth bit.
"We have calculated Pi at varying bit positions and precisions, and one of the largest computations took 23 days of wall clock time and 503 years of CPU time on a 1000-node cluster," Wo Sze said on his developer page, which gave details of the result.
Unlike traditional methods of calculating digits of Pi, which requires the calculation to process all digits up to and including the desired digit, this method used an algorithm that allows desired digits of Pi to be calculated without a need for knowledge of their direct numerical predecessors. This algorithm outputs the digits in hexadecimal and binary format, so the two quadrillionth bit of Pi — 0 — is part of a larger binary sequence that denotes an actual number.
To calculate it, spare cycles were used from Yahoo's Hadoop clusters. Yahoo is one of the world's foremost users of Hadoop, which is a MapReduce-based open source software framework that supports massively distributed applications, such as databases. According to the Hadoop user page, Yahoo has over 100,000 individual computer processing units (CPUs) in over 36,000 computers running Hadoop across multiple clusters.
Wo Sze's success has smashed the previous record - finding pi's quadrillionth bit - set by student Colin Percival in 2000.
August was a good month for global technology companies and their efforts in obscure mathematics: in August Google donated CPU cycles to a project that solved the vaunted Rubik's Cube's 'God Number'.