Lapack++ 2.5.4 released on 2010-11-10

Very minor bugfix release to update the build system for latest
autotools and also adapt to some newer gcc versions.

Lapack++ 2.5.3 released on 2009-02-09

Bugfix release.

Lapack++ 2.5.2 released on 2007-07-14

This bugfix release introduces another copy() method for all the
matrices, along with documentation of a known issue when declaring a
new matrix variable that should contain a deep copy of another
submatrix (see operator= documentation). Also, this release fixes the
broken LaIndex::set method.

Lapack++ 2.5.1 released on 2007-01-30

This bugfix release will fix problems with the new gcc-4.1.2
compiler due to different fortran return values of zdotc(). Also,
an ambiguous overloading of Blas_Mat_Mat_Mult() functions is
fixed. For MSVC compilers, the inject() method has been fixed. For
LaSVD_IP, the argument checking will only complain about wrong U
and Vt arguments if they are actually used.

Lapack++ 2.5.0 released on 2006-12-05

This release contains improved classes for symmetric banded
matrices and a completely new class for their factorization (hence
the libtool major version of the library was incremented). Some
vector classes got resize() methods where these were still
missing. The exception class was improved so that easily the
assert()s can be changed into throwing exceptions, although the
default is still the conventional behaviour of aborting the
application. The header files for non-C++ lapack functions were
improved to enable their use in plain C as well.

Lapack++ 2.4.10 released on 2006-05-17

This release eventually enables the compilation of a real dynamic
DLL on windows/MSVC, because before it could only create a static
LIB. To build this, you need libblas32.lib and liblapack32.lib
which are available in the pre-compiled setup.exe package in the
subdirectory lapackpp\lib\ of the installation location.

Lapack++ 2.4.9 released on 2006-05-15

In this release, some eigenvector function names have been
clarified, some matrix operators have been improved, and a project
file for MSVC 7.1 has been added. Also, the header files with "++"
in the name are no longer installed (though still included in the
tarball).

Lapack++ 2.4.7 released on 2006-02-24

In this release, scaling functions for matrices have been
added. Also, a serious calculation error regarding the normF of
non-square matrices has been fixed. And some newer Microsoft
Visual C++ is probably now support that hasn't been before.


LAPACK++ 2.0.0 released on 2004-07-14
------------------------

For LAPACK++ 1.1, the authors are as follows:

University of Tennessee, Knoxvilee, TN.
Oak Ridge National Laboratory, Oak Ridge, TN.
Authors: J. J. Dongarra, E. Greaser, R. Pozo, D. Walker
(C) 1992-1996 All Rights Reserved

And their last statement was: Lapack++ is no longer actively
supported. The successor to this project is that Template Numerical
Toolkit (TNT), see http://math.nist.gov/tnt for details.

Since then, the existing code base was picked up and this successor
LAPACK++ 2.0 was created by Christian Stimming <stimming@tuhh.de>.


2004-02-24 Version 1.9.3
------------------------

Some LaLinearSolve function arguments have been made const in
accordance to the actual behaviour. The documentation has been greatly
updated.


2004-02-06 Version 1.9.1
------------------------

Documentation has been greatly updated. Unexpected
LaGenMatDouble::operator=() behaviour has been fixed.


2004-02-04 Version 1.9.0
------------------------

Complex matrices are available again, through the help of the
additional class LaComplex (la::complex<double>) which contains the
necessary conversion functions.


Original release notes for Lapack++ v. 1.1a below:


WHAT'S NEW IN VERSION 1.1.a
---------------------------

Version 1.1a is a bug-fix release, reflecting changes
C++ compiler support:


o) removed generic.h dependencies (used in early C++ compilers)
o) changed X::X(&X) constructor to X::X(const &X) in utgmd.h,
		ltgmd.h, ultgmd.h, uutmgd.h,
o) reordered members in several matrix objects to coincide with
	constructor argument ordering.
o) removed unused variables (generated warnings on
		some compilers)

