Could someone, please, comment on my code?

Here it is:

Fib.cpp

#include "Fib.h" unsigned fibonacciLinear(const unsigned& n) {     unsigned val[] = {0,1,1};      for(unsigned i(2); i<=n; ++i)     {val[i%3] = val[(i-1)%3] + val[(i-2)%3];}      return val[n%3]; }  unsigned fastRecursiveFibonacci(const unsigned& n) {     Matrix2x2 res =      {         1, 0,          0, 1     };      fastRecursiveFibonacci(res, n);      assert(("Fibonacci number was computed wrong", res._21 == fibonacciLinear(n)));      return res._21; } void fastRecursiveFibonacci(Matrix2x2& res, const unsigned& n) {     static Matrix2x2 m =      {         1, 1,          1, 0     };      if(n > 0)     {         unsigned lN(n);         if(n&1)         {             res*=m;              --lN;         }         else         {             m*=m;              lN/=2;         }         fastRecursiveFibonacci(res, lN);     } } 

Fib.h

//#define NDEBUG #include <cassert> struct Matrix2x2 {     int _11, _12, _21, _22;      void Matrix2x2::operator*=(const Matrix2x2 matrix)     {         int new11(matrix._11*_11 + matrix._21*_12);         int new12(matrix._12*_11 + matrix._22*_12);         int new21(matrix._11*_21 + matrix._21*_22);         int new22(matrix._12*_21 + matrix._22*_22);          _11 = new11;         _12 = new12;         _21 = new21;         _22 = new22;          return;     } };  unsigned fibonacciLinear(const unsigned&);  unsigned fastRecursiveFibonacci(const unsigned&); void fastRecursiveFibonacci(Matrix2x2&, const unsigned&); 

The things which seem to be a bad practice are using a static function variable and a bad recursion style. Could you recommend any other approaches?