floating-point representation


Also found in: Wikipedia.
Related to floating-point representation: Floating point number

floating-point representation

n
(Computer Science) computing the representation of numbers by two sets of digits (a, b), the set a indicating the significant digits, the set b giving the position of the radix point. The number is the product arb, where r is the base of the number system used. Compare fixed-point representation
References in periodicals archive ?
The hardware used in ordinary deep learning uses a data format called 32-bit floating-point representation for processing calculations.
Section 3, presents a detailed description of our algorithm and a brief discussion about the floating-point representation.
n], a binary64 floating-point representation is used as shown in Figure 1 The algorithmic principle is simple and consists at each iteration, to apply a xor operation on the 32 bits of mantissa 1 of the three output elements [X.
Among the topics are mathematical preliminaries and floating-point representation, numerical integration, least squares methods and Fourier series, boundary-value problems, and linear programming problems.
If students are not aware of 64-bit floating-point representation at this point in their study, then the next step usually involves the default 32-bit floating-point representation.
When the programming assignment was returned, the 64-bit floating-point representation was discussed and the correct values revealed.
Floating-point representation of data has a smaller amount of probable error and noise.
Not infrequently in that era, hardware (and, especially, double-precision software) was sloppy enough to require penalizing by several bits from what one would naively guess by counting bits in the floating-point representation.
The concern is about scientific applications using floating-point representation, and justifiably so.
2 A common definition of the relative machine precision, or unit roundoff, is the smallest positive floating-point value, [Epsilon], such that fl(1 + [Epsilon]) [greater than] 1, where fl(x) is the floating-point representation of x.
If you are not a floating-point expert, Micromega provides a utility that converts back and forth between hexadecimal and floating-point representations.
Some specific topics include FPGA particle graphics hardware, hardware factorization based on the elliptic curve method, higher radix floating-point representations for FPGA-based arithmetic, and interleaving behavioral and cycle-accurate descriptions for reconfigurable hardware compilation.

Full browser ?