Tried to write a fast natural log function. Didn’t meet the required error tolerance, but it was a fun experiment.pic.twitter.com/wd8HAUiFFY
-
@warrenm How are you getting the coefficients for the polynomial approximation?
@alexkan I picked the cubic matching the value and derivative of 1 + ln(x) at x = 1 and x = 2. Least squares might do better.
-
-
@alexkan Meant 1 + log2(x) there, not ln.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
@warrenm Figured it was something like that. I’d suggest LolRemez, but I'm having trouble getting correct results out of it at the moment. -
@alexkan Yeah, Eberly’s latest book actually has a section on applying Remez to logarithms, but I haven’t implemented it yet. -
@warrenm Ah, LolRemez works if compiled -O0 with libstdc++. Try 0.15824870, -1.05187502, 3.04788415, -1.15360271—should meet your tolerance! -
@alexkan Bingo, tolerance met with room to spare! Thanks so much for the help!
End of conversation
New conversation -
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.