What's new or has changed in this version 2.32d vs. version 2.32c?
=================================================================
A. Linear Algebra
*****************
- some routines are rewritten to use less memory space
- some algorithms streamlined
- test functions rewritten
- added rotation matrices (rotating vectors in 2 and 3-dimensional space)
- fixed inconvenience in routines:
  procedure RealMatrix_InsertRow
  procedure ComplexMatrix_InsertRow
  procedure RealMatrix_InsertColumn
  procedure ComplexMatrix_InsertColumn
  when matrix has only one row/column
- added routines
  procedure RealMatrix_PolishMatrix
  procedure ComplexMatrix_PolishMatrix
  procedure RealVector_PolishVector
  procedure ComplexVector_PolishVector
  They replace elements in matrix/vector that are smaller than eps (= 1E-18) i.e. almost equal to zero with a
  real zero.
  Doing this can increase accuracy in some algorithms (e.g. SysOfEquations).   


B. Linear Algebra Pro
*********************
- Gauss elimination algorithm now works more efficient and accurate.
Therefore, also the hermite reduction and calculation of matrix inverse and determinant
work more accurate.
- updated Gauss-Jordan algorithm (nil rows/columns)
- fixed bug in calculation of rank of a matrix
- several routines are rewritten to use less memory space
- added:
  - LU factorization
  - QR factorization
  - completed systems of linear equations algorithm (including least-squares method
    and solving systems of complex equations)

C. Linear Algebra for Sparse Matrices
*************************************
Complete new library.


Known Bugs.
===========
The following are known bugs.
They will be fixed a.s.a.p..

1. The QR decomposition of a m*n matrix where m > n gives wrong(?) results.
   The error does not appear if m <= n.  
   Note that if m < n the columns of the matrix can become linearly dependent -> no QR.
2. The least squares method does not return a correct result for systems A*X = Y where
   A is a n*1 matrix i.e. A has only one column.
   This bug is a result of a small inconvenience in the hermite reduction of a
   1*n matrix (i.e. a matrix with only 1 row).
