On 1/16/16 9:37 PM, Timon Gehr wrote:
Ivan's analysis suggests that even something significantly larger, like n/log(n)² might work as an upper bound for k.
I'm not clear on how you got to that boundary. There are a few implementation details to be minded as well (quickly and carefully approximating the log etc). Could you please make a commented pull request against my own? Thanks! -- Andrei
