Numerics Updates in Maple 14
Extracting Hardware Floating Point Numbers from ArraysConstructing Hardware Floating Point Numbers
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk0">Extracting Hardware Floating Point Numbers from Arrays</Text-field>
Arrays with datatype = float[8] contain 64-bit hardware double-precision floating-point numbers. These arrays as a whole can be directly accessed and operated on efficiently both in internal and external compiled algorithms. Prior to Maple 14, without the use of option hfloat, extracting an individual element from one of these arrays returned a floating-point number using Maple's internal software float representation. Now, extracting a single element from a datatype=float[8] array will return the number using Maple's HFloat representation. No base 2 to base 10 exponent change is incurred.
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk1">Constructing Hardware Floating Point Numbers</Text-field>
The HFloat constructor has been extended to take an optional 3rd argument indicating the base of the exponent. Hardware floats can be created in the form mantissa*2^exponent, which matches the underlying binary representation. Similarly SFloatMantissa and SFloatExponent have been updated to also accept an optional base 2 argument to extract the mantissa and exponent parts of hardware floats in the same form.
See AlsoEnhancements to Symbolic Capabilities in Maple 14Index of New Maple 14 Features