1 comment

[ 4.2 ms ] story [ 11.3 ms ] thread
Is there any generic way to preserve loss of numerical accuracy for any given function?