Thanks for the functions in Wolfram language! I stayed away from doing it in Wolfram because sometimes it 'catches' computations with very small/big numbers and convert it automatically to arbitrary precision, moreover, I wanted to play a bit with library-linking c code. But your functions seem to work fine! Thanks for sharing! Arrayplot is a good way of visualizing what happens.
You're indeed right! 'most of the work' is done by shifting the exponent part of the number. The algorithm can actually be generalised for any exponent between -1 and 1, see also discussion here. Perhaps, as a follow up, I will compute the magic constant as a function of the powerÂ…
