Saola (System for Advanced Optimization of Logical Assertions)
Numerical Expression Compiler
Copyright (C) 2012 Dassault Systèmes Exalead
See the COPYING file for license information

More information at:
     https://github.com/exalead
     http://www.exalead.com

0. Goals
--------

Saola is an optimized numerical expression compiler and evaluator. It aims to
translate a regular string expression ("a+42") into an evaluable form.
The evaluable form is optimized on x86-64 (AMD64) architectures into real
machine bytecode to improve performances, and into pseudo-bytecode on other
architectures.

It has the following features:
- double and long (64-bit) numerical support
- common operators (+, -, <<, ...), parenthesis support
- built-in functions (log, pow ...), custom functions
- custom variables (const or non-const)

1. Supported features
---------------------

Numbers
-9223372036854775808 .. 9223372036854775807
                Decimal number
X.Y             Double number
X.Ye+Z          Double number, exponent form
0x0 .. 0xffffffffffffffff
                Hexadecimal number
0b0 .. 0b1111111111111111111111111111111111111111111111111111111111111111
                Binary number

Variables
[a-zA-Z0-9]*   Any defined variable (const or non-const)
                
Floating-point or Integer
-x             Unary negative
abs(x)         Absolute value of two numbers
x + y          Add
x - y          Subtract
x * y          Multiply
x / y          Divide
min(x,y)       Min of two numbers
max(x,y)       Max of two numbers
b ? x : y      Ternary conditional operator

Boolean
true           Boolean true (1)
false          Boolean false (0)
!b             Boolean negation
b && c         Boolean and
b || c         Boolean or
b ^^ c         Boolean xor
x < y          Less than (result is boolean)
x <= y         Less or equal (result is boolean)
x > y          Greater than (result is boolean)
x >= y         Greater or equal (result is boolean)
x == y         Equals (result is boolean)
x != y         Not equals (result is boolean)

Integer
x % y          Modulus
x & y          Binary and
x | y          Binary or
x ^ y          Binary xor
x << y         Left binary shift
x >> y         Right binary shift
x >>> y        Right unsigned binary shift
long(x)        Conversion from floating-point to integer
long_raw(x)    Raw IEEE754 copy from double to long

Floating-point
nan            Not a number
inf            Infinite
~x             Binary negation
acos(x)        Arc cosine
acosh(x)       Arc cosine hyperbolic
asin(x)        Arc sine
asinh(x)       Arc sine hyperbolic
atan(x)        Arc tangent
atanh(x)       Arc tangent hyperbolic
cbrt(x)        Cube root
ceil(x)        Smallest integral value not less
cos(x)         Cosine
cosh(x)        Hyperbolic cosine
exp(x)         Base-e exponential
floor(x)       Largest integral value not greater
j0(x)          Bessel functions of first kind, order 0
j1(x)          Bessel functions of first kind, order 1
log(x)         Base-e logarithm
log10(x)       Base-10 logarithm
log2(x)        Base-2 logarithm
pow(x)         Power
round(x)       Round to nearest integer away from zero
sin(x)         Sine
sinh(x)        Hyperbolic sine
sqrt(x)        Square root
tan(x)         Tangent
tanh(x)        Hyperbolic tangent
trunc(x)       Round to integer towards from zero
y0(x)          Bessel functions of second kind, order 0
y1(x)          Bessel functions of second kind, order 1
double(x)      Conversion from integer to floating-point
double_raw(x)  Raw IEEE754 copy from long to double
atan2(x,y)     Arc tangent of two numbers
fmod(x,y)      Floating-point remainder
hypot(x,y)     Euclidean distance
nextafter(x,y) Next representable floating-point value
jn(n,x)        Bessel functions of first kind, order n
ldexp(x,n)     Multiply by power of 2
yn(n,x)        Bessel functions of second kind, order n