LA_library/ChangeLog

28.06.2016 minor changes in Makefile.am to work smoothly with recent autotools
28.06.2016 fixed determinant sign issue due to dgesv ipiv counting from 1 in nonclass.cc
11.11.2015 autotool files changed to support separate only-static ATLAS libraries
05.03.2015 operator>> for fourindex_ext
27.10.2014 added interface for Lapack SVD of complex matrices
27.10.2014 diagonalset() for rectangular matrices
04.11.2013 added conjugateme() for corder=1 in complex diagonalize() and gdiagonalize() to get correct eigenvector convention for C-storage
30.10.2013 macros for consistent lowercase and uppercase of character parameters for case sensitive BLAS and LAPACK; case-insesitivness of these in vec.cc and mat.cc (cublas still not treated)
29.10.2013 included la_traits.h in nonclass.c to get correct extern "C" of cblas
14.10.2013 added operator% to bivector
01.06.2012 more efficient clear() by detachonly parameter to copyonwrite()
14.03.2012 fixed overflow in product of dimensions of NRMat and NRSMat
13.03.2012 symmetry of integrals with different spins added to fourindex.h
23.02.2012 included unistd.h 
23.02.2012 fixed max read/write size in multiget and multiput to 1GB
24.01.2012 Improved DIIS (L.V.)
19.01.2012 Fixed location for gpu in nonclass.cc
02.06.2011 In oplus() =0 replaced by clear() to work on non-square matrices (J.P.)
01.02.2011 Added trace2 for complex matrices by L. Veis
18.01.2011 Minor bugfixes and compatibility with Intel C++ compiler by Roman Curik
28.12.2010 Generalized diagonalization and functions of general complex matrices 
21.12.2010 Changed to size_t in matrix put,get to prevent overflow
08.12.2010 Deallocate method added to LA_traits, used for memory saving in matrix exp
24.11.2010 Added checking for integer overflow in exptimes
23.10.2010 Fixed dependency on atlas for static libraries in configure.ac 
27.09.2010 Seed of CSRMat class added.
22.09.2010 Allowed formal operator[] on gpu matrices
22.09.2010 Added submatrix() to SparseSMat
22.09.2010 Zgerc and zgeru implemented in non-cblas version
21.09.2010 Fixed a bug in laerror macro causing sometimes compilation problems
08.09.2010 RELEASE 0.6
08.09.2010 Doxygen documentation for  NRVec, NRMat, NRSMat classes contributed by M. Sulc
08.09.2010 Extended CUBLAS support for NRVec, NRMat, NRSMat classes contributed by M. Sulc
08.09.2010 Minor bugfixes and improvements contributed by M. Sulc
25.06.2010 Added proof-of-concept CUBLAS support for NRVec, NRMat, NRSMat 
24.06.2010 Fixed a memory leak existing when MATPTR  was defined
18.06.2010 added autoconf support for BLAS+LAPACK compiled with 64-bit integers and for CUBLAS
11.06.2010 interface to fortran BLAS+LAPACK compiled with 64-bit integers contributed by L. Veis
25.02.2010 linear_solve_x and multiply_by_inverse contributed by M. Sulc
17.01.2010 miscellaneous extensions contributed by M. Sulc
17.01.2010 bugfix in NRMat::operator *= +=  -= for non-square matrix
15.01.2010 added check for cblas.h and clapack.h 
13.01.2010 fixed ::fscanf to be compilable by gcc 4.4
10.01.2010 outer product with sparse result otimes_sparse added
09.01.2010 typedef ROWTYPE for NRVec and SparseSMat for assignable elements/rows via operator[]
09.01.2010 SparseSMat adapted to rectangular matrices (except gemm)
07.01.2010 Fixed a bug in SparseSMat::axpy
07.01.2010 Pivoted Cholesky decomposition for positive semidefinite SparseSMat-rices
05.01.2010 Cholesky decomposition lapack interface  implemented
04.01.2010 NRMat * SparseSMat
28.12.2009 added put and get for sparsesmat, fixed put/get bug in traits (contrib. L. Veis)
15.12.2009 added possibility for a user-defined initial guess in davidson
20.11.2009 removed #include config.h from LA_traits.h
19.11.2009 fixed a bug in SparseMat::norm() and SparseSMat::norm()
19.11.2009 added SparseSMat::operator* for NRVec and SparseSMat::diagonalof
13.11.2009 RELEASE 0.5
13.11.2009 libtool employed to create also shared library version
12.11.2009 library enclosed in the namespace LA
12.11.2009 #defines for numerical parameters changed to const doubles
12.11.2009 added SparseSMat class for efficient exponentiation of sparse complex symmetric matrices
12.11.2009 removed unnecessary casts in complex vec,mat,smat routines
12.11.2009 fixed bug in complex NRSMat operator +=,-=
11.11.2009 changed default SPARSEEPSILON threshold value
10.11.2009 removing 'using namespace std' from the main include, leaving this to the user's discretion
10.11.2009 NRVec::otimes introduced
04.11.2009 improved efficiency of complex matrix exp
04.11.2009 fixed return type for complex sparsemat::norm
19.10.2009 added modified GS orthomormalization with optional metric
15.10.2009 added support for long long
07.10.2009 RELEASE 0.4
06.10.2009 autoconf/automake for user convenience
13.09.2009 RELEASE 0.3
04.09.2009 fixed a bug with missing copyonwrite() inside clear()
02.09.2009 fixed a numerical problem in exptimes() for large values of scale
04.06.2009 fixed a regression due to use of normtype
28.05.2009 removed the export keyword unsupported in g++
28.05.2009 fixed normtype in la_traits
28.05.2009 randomize() added for complex
28.05.2009 Added interface for complex hermitean  matrix diagonalization
06.02.2009 constructors of complex vec/mat/smat from real ones
20.01.2009 optional _LA_count_check
20.01.2009 NRMat(const NRVec...) constructor with optional offset parameter
18.12.2008 cout.flush() for verbose iterative methods
17.12.2008 improvements in laerror for exception handling
26.11.2008 RELEASE 0.2
xx.11.2008 checked compilation with gcc-4.3.2
xx.08.2008 fixed wrong permutation symmetry in previously unused (and untested) fourindex class
16.04.2008 replaced the dasum() wrongly used for trace
14.03.2008 support for (formal) matrix resize(n,0)
11.03.2008 added cblas_idamax replacement for non-cblas
05.03.2008 fixed transposed bug in inverse() with non-cblas
26.02.2008 INITIAL RELEASE 0.1