Fixed-point math for Python

Comments

(no comments yet)

Log in or Join to leave comments.

Article Metadata

---

Tested with

Devices(s): Nokia N95 8GB

---

Compatibility

Platform(s): S60 3rd Edition

---

Platform Security

Capabilities: )

---

Article

Keywords: math

Created: lpvalente (03 Dec 2009)

Last edited: hamishwillee (08 May 2013)

Introduction

Fixed-point math is a technique for representing floating point numbers as integers. By using this technique one is able to simulate floating point operations on hardware that don't support them (or at least on hardware that don't have dedicated floating point units, as many of current devices). This technique has been used used on many graphics applications (i.e. games).

The source code represents numbers with 32 bit integers, using 16 bits to represent the fractional part.

Source Code

import math
 
def degreeToRad (angle):
    return math.radians(angle)
 
#
def radToDegree (angle):
    return math.degrees(angle)
 
#
def int2fixed (value):
    return value << 16
 
#
def float2fixed (value):
    return int(value * 65536.0)
 
#
def fixed2int (value):
    return value >> 16
 
#
def fixed2float (value):
    return value * 0.0000152587890625  # (1/65536.0) 
 
# should be 64-bit math
def fixed_mul (op1, op2):
    r = long(op1) * long(op2)
    return r >> 16
 
# should be 64-bit math
def fixed_div (op1, op2):
    o1 = long (op1) << 16        
    r = long (o1 / long (op2) )
    return int(r)
 
# some fixed point constants
PIx = float2fixed (pi)
 
# one
ONEx = int2fixed (1)
 
# zero
ZEROx = 0