little C++ experiment because I was curious, adding floats but rounding them probabilistically to get better accuracy / less floating point errorhttps://pastebin.com/SMpLJExD
-
Show this thread
-
Replying to @TylerGlaiel
If I understand what you want correctly, this is already a solved problem, no entropy necessary.https://en.wikipedia.org/wiki/Kahan_summation_algorithm …
1 reply 0 retweets 10 likes -
Replying to @cmuratori
I don't want anything, this was purely about curiosity. Needing to keep a separate error accumulator value around, may as well just store it as a double instead. stochastic rounding works on any operation also, not just addition
1 reply 0 retweets 1 like
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.