32

25

24

24

# FixedPointNumbers

This library implements fixed-point number types. A fixed-point number represents a fractional, or non-integral, number. In contrast with the more widely known floating-point numbers, with fixed-point numbers the decimal point doesn't "float": fixed-point numbers are effectively integers that are interpreted as being scaled by a constant factor. Consequently, they have a fixed number of digits (bits) after the decimal (radix) point.

Fixed-point numbers can be used to perform arithmetic. Another practical application is to implicitly rescale integers without modifying the underlying representation.

This library exports two categories of fixed-point types. Fixed-point types are used like any other number: they can be added, multiplied, raised to a power, etc. In some cases these operations result in conversion to floating-point types.

# Type hierarchy and interpretation

This library defines an abstract type `FixedPoint{T <: Integer, f}` as a subtype of `Real`. The parameter `T` is the underlying machine representation and `f` is the number of fraction bits.

For `T<:Signed` (a signed integer), there is a fixed-point type `Fixed{T, f}`; for `T<:Unsigned` (an unsigned integer), there is the `Normed{T, f}` type. However, there are slight differences in behavior that go beyond signed/unsigned distinctions.

The `Fixed{T,f}` types use 1 bit for sign, and `f` bits to represent the fraction. For example, `Fixed{Int8,7}` uses 7 bits (all bits except the sign bit) for the fractional part. The value of the number is interpreted as if the integer representation has been divided by `2^f`. Consequently, `Fixed{Int8,7}` numbers `x` satisfy

``````-1.0 = -128/128 ≤ x ≤ 127/128 ≈ 0.992.
``````

because the range of `Int8` is from -128 to 127.

In contrast, the `Normed{T,f}`, with `f` fraction bits, map the closed interval [0.0,1.0] to the span of numbers with `f` bits. For example, the `N0f8` type (aliased to `Normed{UInt8,8}`) is represented internally by a `UInt8`, and makes `0x00` equivalent to `0.0` and `0xff` to `1.0`. Consequently, `Normed` numbers are scaled by `2^f-1` rather than `2^f`. The type aliases `N6f10`, `N4f12`, `N2f14`, and `N0f16` are all based on `UInt16` and reach the value `1.0` at 10, 12, 14, and 16 bits, respectively (`0x03ff`, `0x0fff`, `0x3fff`, and `0xffff`). The `NXfY` notation is used for compact printing and the `fY` component informs about the number of fractional bits and `X+Y` equals the number of underlying bits used.

To construct such a number, use `1.3N4f12`, `N4f12(1.3)`, `convert(N4f12, 1.3)`, `Normed{UInt16,12}(1.3)`, or `reinterpret(N4f12, 0x14cc)`. The last syntax means to construct a `N4f12` from the `UInt16` value `0x14cc`.

More generally, an arbitrary number of bits from any of the standard unsigned integer widths can be used for the fractional part. For example: `Normed{UInt32,16}`, `Normed{UInt64,3}`, `Normed{UInt128,7}`.

# Computation with Fixed and Normed numbers

You can perform mathematical operations with `FixedPoint` numbers, but keep in mind that they are vulnerable to both rounding and overflow. For example:

``````julia> x = N0f8(0.8)
0.8N0f8

julia> float(x) + x
1.6f0

julia> x + x
0.596N0f8
``````

This is a consequence of the rules that govern overflow in integer arithmetic:

``````julia> y = reinterpret(x)        # `reinterpret(x::FixedPoint)` reinterprets as the underlying "raw" type
0xcc

julia> reinterpret(N0f8, y + y)  # add two UInt8s and then reinterpret as N0f8
0.596N0f8
``````

Similarly,

``````julia> x = eps(N0f8)             # smallest nonzero `N0f8` number
0.004N0f8

julia> x*x
0.0N0f8
``````

which is rounding-induced underflow. Finally,

``````julia> x = N4f12(15)
15.0N4f12

julia> x*x
ERROR: ArgumentError: Normed{UInt16,12} is a 16-bit type representing 65536 values from 0.0 to 16.0037; cannot represent 225.0
Stacktrace:
 throw_converterror(::Type{Normed{UInt16,12}}, ::Float32) at /home/tim/.julia/dev/FixedPointNumbers/src/FixedPointNumbers.jl:251
 _convert at /home/tim/.julia/dev/FixedPointNumbers/src/normed.jl:77 [inlined]
 FixedPoint at /home/tim/.julia/dev/FixedPointNumbers/src/FixedPointNumbers.jl:51 [inlined]
 convert at ./number.jl:7 [inlined]
 *(::Normed{UInt16,12}, ::Normed{UInt16,12}) at /home/tim/.julia/dev/FixedPointNumbers/src/normed.jl:254
 top-level scope at REPL:1
``````

In some circumstances, it may make most sense to think of `FixedPoint` numbers as storage types rather than computational types. You can call `float(x)` to convert `x` to a floating-point equivalent that is reasonably safe for computation; in the type domain, `floattype(T::Type)` returns the corresponding type. Note that in some cases `floattype(T)` differs from `float`'s behavior on the corresponding "raw" type:

``````julia> float(UInt8)
Float64

julia> floattype(N0f8)
Float32
``````

Because of the role of FixedPointNumbers in domains such as image-processing, this package tries to limit the expansion of the number of bits needed to store results.

02/27/2014

14 days ago

272 commits