Useage: Quicktrans.library is a replacement for mathtrans.library. Just re-name it and put it wherever you formerly kept mathtrans.library. The libary offsets are identical. In addition there is a 10^X function, (I call it expten), with offset -132. Speed: Trig functions are typically 2 to 2.5 times as fast as corresponding mathtrans.library functions ( except sincos, but it's rarely used). Logarithmic, exponential and hyperbolic functions are closer to 3 times as fast. Tieee and Fieee run in about 58% of the corresponding mathtrans.library time. Sqrt is the slowest, but it still runs in about two thirds of the mathtrans.library time. Method: Most of the functions are calculated using the first few terms of the appropriate Taylor series approximation and corresponding data tables. Much of the speed increase comes from using scaled-up integers rather than floating point representations wherever possible. Accuracy: Since the Motorola ffp format used in mathtrans.library uses 24 bits for the mantissa, the mathtrans library provides at best between 6 and 7 decimal digit accuracy. I scale up by a factor of 2^24, resulting in just about the same accuracy. For numbers of absolute value greater than one my results almost always agree with mathtrans.library in at least the first 6 digits, with the seventh digits usually within 4 of each other. For numbers with absolute value less than 1, my results almost always agree with mathtrans.library within about 0.0000004. The only exception I'm aware of is the tangent function very close to odd multiples of pi/2, but neither library is very accurate there. Who cares if the tangent of 1.57 is 1255 or 1256? Distribution: Use the library in whatever way you want to, but not for commercial use without my written permission. If you distribute it, please keep this document with it. Bug reports, comments, questions, etc. to: Martin F. Combs 2989 Sundance Circle Las Cruces, NM 88001 USA (505) 522-6408