Skip to content

Latest commit

 

History

History
34 lines (23 loc) · 1.75 KB

README.md

File metadata and controls

34 lines (23 loc) · 1.75 KB
Raw

math

This is a collection of mathematical classes useful in my robotics projects. It contains 3D transforms, rotations, vectors, and some miscellaneous math methods.

Matrix and Vector Usage Patterns

There are API patterns to the matrix and vector classes in library: results pre-allocation and method chaining.

Pre-allocation of Results

Pre-allocate results, and use the result as the object of the method call. The pre-allocation allows the caller to completely control object allocation and thus reuse objects. This can (usually) avoid excessive allocation and garbage collection as a result of mathematical routines. Also by having the result object be the base of the call, it follows mathematical conventions of writing "x = a * b", for example:

Vec3d x = new Vec3d(); // preallocate result
x.cross(a, b); // compute cross product of a and b and store result in x

Method Chaining

Where possible, methods return this. This allows for method chaining, where the result from the first operation is used immediately as an operator in the second. For example:

Vec3d v = new Vec3d();
v.sub(a,b).mul(0.5).add(c); // compute v = (a-b) * 0.5 + c

Double / Float Versions

This code is primarily written in double precision math. The make-float-versions.sh script converts the double versions to float version. This is a very simple sed-script based conversion which relies upon a few rules:

  • For all floating-point values, use a decimal, e.g., instead of "0", use "0.0". The script will append "f" to make them "0.0f".
  • Follow the existing classes for computing hash codes based upon doubles, in particular, make sure to accumulate a long, and then return the XOR'd value in the same form.

License

BSD 2-Clause License