Lis 1.2.62 December 24, 2011 - Added support for preconditioning in MINRES. - Fixed error handling bug in lis_solve(). - Updated documents. Lis 1.2.61 November 30, 2011 - Added support for lis_matrix_is_assembled() in lis_matrix_setDLU_xxx(). Lis 1.2.60 November 30, 2011 - Fixed bug of lis_matrix_set_ccs() since 1.2.56, which caused segmentation fault. Lis 1.2.59 November 25, 2011 - Updated documents. Lis 1.2.58 November 22, 2011 - Updated documents. Lis 1.2.57 November 9, 2011 - Added support for flag --with-library-suffix to tag suffix to library names (thanks to Jussi Lehtola). Lis 1.2.56 November 8, 2011 - Added support for lis_matrix_is_assembled() in lis_matrix_set_xxx() for calling solver multiple times without allocating new memory (thanks to Malte Thoma). - Changed origin of Fortran subroutine lis_vector_set_size() to 1 (thanks to Kengo Tomida). Lis 1.2.55 November 7, 2011 - Changed names of configuration header files config.h and config_win32.h to lis_config.h and lis_config_win32.h to avoid conflict in Fedora packages (thanks to Jussi Lehtola). Lis 1.2.54 November 2, 2011 - Fixed bug of lis_solver_set_option() (thanks to Akio Yamamoto and Kenichiro Yoshimi). Lis 1.2.53 November 29, 2010 - Added support for Intel Composer XE and Microsoft Visual Studio 2010. Lis 1.2.52 October 20, 2010 - Fixed bug of lis_solver_minres() with flag -print (thanks to Kazuhiro Ando). Lis 1.2.51 September 16, 2010 - Updated documents. Lis 1.2.50 September 15, 2010 - Updated documents. Lis 1.2.49 September 14, 2010 - Updated documents. Lis 1.2.48 September 12, 2010 - Fixed bug in lis_input_hb.c, which caused input error with symmetric matrices in Harwell-Boeing Format (thanks to Shoji Itoh). Lis 1.2.47 September 10, 2010 - Added instruction for building shared libraries to documents. Lis 1.2.46 September 8, 2010 - Fixed bug of OMPFLAG. Lis 1.2.45 September 7, 2010 - Updated documents (thanks to Riccardo Rossi). Lis 1.2.44 August 26, 2010 - Updated configure.in. Lis 1.2.43 August 25, 2010 - Set default configuration of Mac OS X binaries to 64-bit. - Fixed bug in configure.in for PGI compilers. - Added support for Pathscale compilers (experimental). Lis 1.2.42 August 20, 2010 - Enabled AC_CHECK_HEADERS for Mac OS X. - Fixed parameters of configure.in for Mac OS X. (Note that default installation of Open MPI for Mac OS does not support Fortran.) Lis 1.2.41 August 18, 2010 - Added support for Windows and Mac OS X. - Added instructions for Visual Studio to documents (thanks to Lizhu Tong). - Fixed default location of win32 executable files. Lis 1.2.40 August 17, 2010 - Fixed bug in lis_with_fortran_mpi.icproj. - Restored flag USE_SAAMG in project files. Lis 1.2.39 August 14, 2010 - Fixed default location of win32 libraries. Lis 1.2.38 August 13, 2010 - Fixed character encoding of project files. - Fixed obsolete library name of MPICH2 in project files. Lis 1.2.37 August 12, 2010 - Moved project files to win32. Lis 1.2.36 August 11, 2010 - Fixed bug in configure.in, which caused configure error with XL Fortran. - Fixed bug in src/fortran/lisf_system.c, which caused compile error with Visual Studio. - Temporarily removed flag USE_SAAMG in project files for Visual Studio. Lis 1.2.35 July 29, 2010 - Fixed bug in test/test.sh, which caused make check error with flag --enable-quad. - Updated documents. Lis 1.2.34 July 28, 2010 - Updated documents. Lis 1.2.33 July 23, 2010 - Fixed bug in spmvtests. - Updated documents. Lis 1.2.32 July 23, 2010 - Added experimental support for flags -(e)storage and -(e)storage_block. - Updated spmvtests and documents. - Changed specification of precision flag: flags | description flags | description -------------+-------------------- -------+----------------------------- -precision | for lis_solve() ==> -f | for lis_solve() -eprecision | for lis_esolve() -ef | for lis_esolve() Lis 1.2.31 July 22, 2010 - Added support for Rayleigh Quotient Iteration. - Inner solvers for quad precision eigensolvers are switched to quad precision. - Changed order of eigensolvers: flags | eigensolvers flags | eigensolvers ------------+-------------------- ------------+----------------------------- -e {cg|4} | Conjugate Gradient ==> -e {rqi|4} | Rayleigh Quotient Iteration -e {li|5} | Lanczos Iteration -e {si|5} | Subspace Iteration -e {si|6} | Subspace Iteration -e {li|6} | Lanczos Iteration -e {cr|7} | Conjugate Residual -e {cg|7} | Conjugate Gradient -e {cr|8} | Conjugate Residual - Updated documents. Lis 1.2.30 July 21, 2010 - Fixed bug of lis_esolve(), which failed to destroy esolver in some cases with flag --enable-mpi. Lis 1.2.29 July 21, 2010 - Fixed bug of lis_esolve() in 1.2.27 and 1.2.28, which failed to return default eigenvector. - Fixed bug of configure for flag --enable-gprof with f90 compilers. - Added support for flag --enable-64bit for AIX and Solaris x86. - Updated documents. Lis 1.2.28 July 16, 2010 - Updated documents. Lis 1.2.27 July 15, 2010 - Added support for quadruple precision operations in eigensolvers. - Fixed bug in lis.vcproj. Lis 1.2.26 July 7, 2010 - Added support for lis_vector_nrmi() for computing infinity norm (thanks to Osamu Nakamura). - Updated documents. Lis 1.2.25 July 4, 2010 - Fixed bug in configure.in for gcc with --enable-gprof. Lis 1.2.24 July 3, 2010 - Added support for flag --enable-gprof for performance analysis with gprof (experimental). - Added test program for lis_esolver_get_evectors(). - Updated documents. Lis 1.2.23 July 1, 2010 - Added support for lis_output_matrix() (thanks to Jun Sato). - Changed specification of lis_esolver_get_evectors(). Lis 1.2.22 June 30, 2010 - Fixed bug in test.sh. - Updated spmvtests. - Updated documents. Lis 1.2.21 June 29, 2010 - Updated test programs. - Updated documents. Lis 1.2.20 June 28, 2010 - Added support for parameter irestart for IDR(s). - Fixed default value of parameter emaxiter for subspace and Lanczos iteration. - Updated documents. Lis 1.2.19 June 26, 2010 - Fixed default value of parameter etol to avoid breakdown. Lis 1.2.18 June 24, 2010 - Fixed minor bug in test.sh for SAAMG with flag --enable-omp. - Fixed minor bug in lis_solver.c for SAAMG with flag --enable-mpi. - Added support for three dimensional test programs. - Updated documents. Lis 1.2.17 June 21, 2010 - Fixed print bug in lis_esolver_cg.c. Lis 1.2.16 June 18, 2010 - Bug fix release. Some Fortran related interoperability problems still exist. - Fixed some errors in documents (thanks to Osamu Nakamura). - Added support for lis_output() in Fortran (thanks to Hiroshi Imamura). - Updated spmvtests (thanks to Katsunori Matsuzaki). - Fixed errors in number of nonzero entries printed by test programs test2, etest2, spmvtest1 and spmvtest2 with flag --enable-mpi. - Fixed bug in spmvtests, which sometimes fail to measure elapsed time when matrix is very small. - Fixed bug in spmvtests for single threaded environments. - Added support for flag -m to specify default mode for Lanczos and subspace iteration. - Updated documents. Lis 1.2.15 September 8, 2009 - Separated core kernel of lis_solve() to avoid overhead of handling preconditioner when called multiple times. It is now implemented as lis_solve_kernel(), to be combined with lis_precon_create() and lis_precon_destroy() (thanks to Masakazu Akiyama). - Inverse iteration and subspace iteration were rewritten using lis_solve_kernel(). Performance of Lanczos iteration, which calls inverse iteration to compute eigenvectors, is also affected. - Updated test programs to measure SpMV performance. Lis 1.2.14 August 26, 2009 - Fixed bug in configure.in, which omitted OMPFLAG when target architecture is specified with flag --enable-mpi. - Updated test programs to measure SpMV performance. - Fixed description of test program etest3 in documents. Lis 1.2.13 August 15, 2009 - Fixed specification of lis_matrix_split() to keep original data of splitted matrix (thanks to Tim Cramer). Lis 1.2.12 July 21, 2009 - Fixed minor bugs related to SpMV test programs. - Revised documents. Lis 1.2.11 July 7, 2009 - Fixed specification of lis_vector_gather() to share vector values among processes (thanks to Kouji Takatani). - Revised documents. Lis 1.2.10 July 4, 2009 - Added support for lis_vector_scatter() and lis_vector_gather() to handle vector elements, mainly for MPI configuration. - Added official support for lis_solver_get_status() and lis_esolver_get_status() to get correct status of solvers (thanks to Tim Cramer). - Fixed bug of link error for Fortran 77 based MPI compilers. - Revised documents. Lis 1.2.9 July 1, 2009 - Added official support for flag -saamg_theta to set drop criteria for AMG. - Fixed default parameter of theta for nonsymmetric AMG, which caused error with Lis 1.2.6 or higher (thanks to Kouji Takatani). - Added test programs to measure SpMV performance (experimental). - Fixed MPILIBS related bug in configure.in. - Fixed some minor bugs. - Revised documents. Lis 1.2.8 June 3, 2009 - Fixed minor bug in configure script. Lis 1.2.7 June 1, 2009 - Fixed bug in MPI configuration, which caused error on Hitachi SR16000 and IBM p575. - Fixed some minor bugs. Lis 1.2.6 March 23, 2009 - Changed location of config*.h, to be installed in $(includedir). - Fixed some minor bugs. Lis 1.2.5 December 14, 2008 - Added full functions for handling elements of vectors and matrices via Fortran interface. - Added support for matrix storage format conversion in eigensolvers. - Added support for PGI and Hitachi compilers. - Fixed bug in configure, which omitted flag -DUSE_MPI of Fortran based MPI compilers. - Fixed bug in test/test3f.F, which caused make check error with -DUSE_MPI. Lis 1.2.4 December 5, 2008 - Added support for conjugate residual method for nonsymmetric eigenproblems (experimental). - Fixed some minor bugs. Lis 1.2.3 December 4, 2008 - Fixed origin of lis_vector_get_value() in Fortran (thanks to Kengo Tomida). - Fixed some minor bugs. Lis 1.2.2 November 25, 2008 - Fixed bug in test/test2.c, which caused make check error in some cases (thanks to Takahito Miki). - Fixed bug in configure for combination of SAAMG and gfortran based MPI compiler. SAAMG can be built with gfortran 4.3 or higher. - Fixed some minor bugs. Lis 1.2.1 November 18, 2008 - Fixed origin of lis_vector_set_value() in Fortran (thanks to Yutaka Masuda). Lis 1.2.0 November 12, 2008 - Added support for eigensolvers (power iteration, inverse iteration, approximate inverse iteration, conjugate gradient, Lanczos iteration, and subspace iteration) for real sparse matrices. - Added support for MINRES (thanks to Hirofumi Notsu). - Changed names of lis_output_residual_history() and lis_get_residual_history() to lis_solver_output_rhistory() and lis_solver_get_rhistory(), respectively. - Fixed convergence criteria of TFQMR and BiCGSTAB(l) (thanks to Shoji Itoh). - Added support for NEC SX-9. - Updated config.guess and config.sub. - Added some optimization options for Fujitsu Compiler (thanks to Fujitsu Limited). - Fixed location of include statement in lisf_init.F to avoid compiling error with Fujitsu Compiler. - Fixed configure to choose mpiexec if mpirun is not available. - Fixed configure to choose Fortran 9x based MPI compiler if SAAMG is enabled. - Fixed some minor bugs. - Revised documents. Lis 1.1.2 December 24, 2007 - Added support for IDR(s) method. - Fixed some minor bugs. Lis 1.1.1 December 6, 2007 - Changed specification of lis_initialize() to avoid MPICH1 related error (thanks to Takashi Yoshida). - Fixed some minor bugs. Lis 1.1.0 October 31, 2007 - Added support for CR,BiCR,CRS,BiCRSTAB,GPBiCR,BiCRSafe and FGMRES methods. - Added support for ILUT preconditioner. - Added binary package for Microsoft Windows. - Added Lis-test for Windows. - Fixed some minor bugs. Lis 1.1.0 beta2 June 18, 2007 - Added support for BiCGSafe. - Fixed bug caused by some memory leaks. Lis 1.1.0 beta1 November 7, 2006 - Added support for quadruple precision operations in linear equation solvers. - Added support for Fortran user interface. - Added support for nonsymmetric SAAMG preconditioner (symmetric structure only). - Added support for Crout ILU preconditioner. - Added support for additive Schwarz preconditioner. - Added support for user-defined preconditioners. - Added support for Harwell-Boeing format. - Integrated Lis-AMG into Lis. - Introduced autoconf. - Changed some specifications: - Introduced LIS_SOLVER structure (integrated type parameters). - Changed declaration of matrices (thanks to David Moloney) and vectors. - Changed notation of command line options. Lis 1.0.2 April 17, 2006 - Fixed bug caused by not correctly setting value of beginning and end of line when lis_vector_create() is called. - Fixed bug caused by wrong selection of linear equation solvers specified with lis_set_options(): flags | solvers flags | solvers ----------------+--------------- ----------------+--------------- -i {tfqmr|7} | Orthomin ==> -i {tfqmr|7} | TFQMR -i {orthomin|8}| GMRES -i {orthomin|8}| Orthomin -i {gmres|9} | TFQMR -i {gmres|9} | GMRES - Fixed bug in allocation of temporary array of Orthomin method. - Fixed bug in output of residual norm in TFQMR method. - Fixed bug in array assignment of BiCG method with ILU preconditioner. - Fixed some minor bugs. Lis 1.0.1 November 6, 2005 - Added support for Lis-AMG. - Added English manual. - Added test program test1b. - Fixed some minor bugs. Lis 1.0.0 September 20, 2005 - Released first version.