wn_matrix releases

• Version 0.20 (2022/02/26)
  • New:
  • A routine to multiply two matrices together has been added to the API.
  • Fix:
  • Some documentation typos have been fixed.


• Version 0.19 (2017/07/13)
  • Internal:
  • Default WnChar is now simply char. Change with environment variable WN_XML_CHAR.
  • Local path has been added to regression scripts.


• Version 0.18 (2016/07/21)
  • New:
  • User can now explicitly set the string type for xmlStrPrintf format with the environment variable WN_XML_CHAR.
  • Examples Makefile now uses the default compiler standard.
  • Internal:
  • Includes have been moved outside extern "C" to prevent template specialization warnings.
  • Format for size_t type is now set to %zu for appropriate compilers.


• Version 0.17 (2016/06/14)
  • Fix:
  • Information links have been updated.
  • Internal:
  • Xml string formats have been updated for recent changes in libxml2.
  • Internal callback structures have been renamed.


• Version 0.16 (2015/05/22)
  • Fix:
  • Definitions have been made more specific to module to prevent collisions with other libraries.


• Version 0.15 (2014/06/21)
  • Fix:
  • Missing element row and column assignments have been added to the element copier.


• Version 0.14 (2012/07/05)
  • New:
  • It is now possible to use a string lookup to avoid printing out of matrix element index strings. This may speed up lookup of elements in large matrices. To use this, compile with the flag -DWN_USE_MATRIX_LOOKUP.
  • Fix:
  • Some typos have been fixed.
  • The code now uses extern "C" for C++ compilers.
  • Reorganization:
  • The examples Makefile has been cleaned up and simplified.


• Version 0.13 (2011/05/09)
  • New:
  • Code is now hosted at SourceForge.net.
  • A new API routine allows users to update matrix elements.
  • The examples/ directory now includes a regression test script generator.
  • Reorganization:
  • XML schemas are now at SourceForge.net.
  • Examples directory has moved to same level as src/.
  • Example codes now have descriptive names.


• Version 0.12 (2010/07/19)
  • Fix:
  • Casts in the data sort routine have been fixed to allow a clean compile with g++.
  • Unreliable floating-point equality comparisons have been fixed.


• Version 0.11 (2010/07/03)
  • Fix:
  • A bug in WnMatrix__getYale() introduced in version 0.8 has been fixed.
  • A number of documentation typos have been fixed.
  • Reorganization:
  • Extraction of a row or column or conversion to other sparse format is now done with sorted arrays rather than with lists. This greatly speeds up these operations for large matrices. The cost is somewhat greater memory required to store matrix elements.


• Version 0.10 (2009/08/21)
  • New:
  • Routines to remove and insert rows and columns have been added to the API.
  • An example to demonstrate the new API routines has been added to the distribution.
  • Fix:
  • Some typos in technical reports and documentation have been fixed.


• Version 0.9 (2009/04/03)
  • Fix:
  • A number of casts have been fixed to eliminate warnings from 4.3 series GNU compilers (thanks to Yeunjin Kim for reporting these warnings).


• Version 0.8 (2009/03/12)
  • New:
  • A new "arrow" matrix structure, WnMatrix__Arrow, has been added along with an efficient Gaussian elimination solver, WnMatrix__Arrow__solve( ), based on it.
  • A new technical report describes the WnMatrix__Arrow structure and its associated solver.
  • New examples demonstrate the WnMatrix__Arrow API.
  • Fix:
  • A number of typos in the API documentation have been fixed.
  • Reorganization:
  • The routine WnMatrix__Coo__writeToXmlFile( ), WnMatrix__Csr__writeToXmlFile( ), WnMatrix__Yale__writeToXmlFile( ), and WnMatrix__write_gsl_vector_to_xml( ) now allow the user to provide a format code for the outputting the matrix element or vector component value. Since the format code is required (even for the default--%g), this is a backwards incompatibility. Examples have been modified to demonstrate these changes.
  • Matrix element assignments and retrievals in some of the examples have been simplified.
  • Unsigned longs have been changed to size_t's in some examples.
  • Rows and columns are now correctly sorted because matrix indices are now compared by their long int values instead of their string values.
  • Internal:
  • Routines to retrieve Coo, Csr, and Yale forms now iterate over lists of rows with non-zero elements. This should speed up these routines for particularly sparse matrices.
  • Some internal parameters that are size_t's are now more consistently treated. This should make compilation with 64-bit machines smoother.


• Version 0.7 (2008/06/05)
  • New:
  • Module is now dependent on gnu gsl.
  • Routines have been added to the API to allow the user to validate input vector data as xml, to parse in vector data from xml, and to write gsl_vector data to an xml file.
  • Routines have been added to the API to allow the user to get the size of and the data array of a gsl_vector.
  • A new routine WnMatrix__solve() has been added to the API to allow the user to solve matrix equations using gsl lu decomposition routines.
  • WnMatrix__getDenseMatrix() routine has been replaced with WnMatrix__getGslMatrix. This is creates a backwards incompatibility for wn_matrix, and users should upgrade to this new version.
  • WnMatrix__getMatrixTimesVector() and WnMatrix__getTransposeMatrixTimesVector() have been replaced with WnMatrix__computeMatrixTimesVector() and WnMatrix__computeTransposeMatrixTimesVector(). This is creates a backwards incompatibility for wn_matrix, and users should upgrade to this new version.
  • Example codes have been added to demonstrate the new API routines.
  • Fix:
  • Several typos in the documentation and in the example files have been fixed.
  • Reorganization:
  • Routines that return data as a new double array now return them as a gsl_vector structure.


• Version 0.6 (2008/03/28)
  • Fix:
  • An index offset problem in WnMatrix__getDiagonalElements() has been fixed.
  • Several typos in the documentation have been fixed.
  • Missing examples have been added to the cleanall command in the examples Makefile.
  • Reorganization:
  • Copyright and license information is now only included in README.txt files for distribution directories to shorten file lengths.


• Version 0.5 (2007/12/02)
  • New:
  • Routines to output matrix data in coordinate, compressed sparse row, and Yale sparse matrix formats in XML files have been added to the API.
  • A routine to read in matrix data from an XML file has been added to the API.
  • Examples 3 and 4 in the tutorial have been rewritten to demonstrate the new API routines.
  • New examples demonstrate the new XML API routines.
  • A new version of the technical report briefly describes the new API routines.
  • Fix:
  • A misallocation of memory in the routine WnMatrix__getCsr() has been fixed so that there is no longer a problem if the number of rows exceeds the number of non-zero matrix elements.
  • Several typos in the documentation have been fixed.
  • Reorganization:
  • FILE pointers in several examples have been renamed.
  • Several examples no longer declare arrays to retrieve coordinate, CSR, or Yale sparse matrix data but rather retrieve the data directly.
  • A dot-slash has been added to the command-line execution call in the examples tutorial. We thank Cal Jordan for this suggestion.


• Version 0.4 (2007/10/05)
  • New:
  • Routines to get coordinate, compressed sparse row, and Yale sparse matrix versions of the matrix have been rewritten. These replace previous versions. This is a backwards incompatibility, so users should certainly upgrade.
  • Examples 3 and 4 in the tutorial have been rewritten to demonstrate the new API routines.
  • A new version of the technical report briefly describes the new data structures.
  • The data_pub directory in the distribution no longer includes the output files.
  • Fix:
  • A memory leak in WnMatrix__getColumn has been fixed.
  • Reorganization:
  • The examples tutorial now presents the line numbers in the example codes of the API routines demonstrated.


• Version 0.3 (2007/07/23)
  • New:
  • Routines to get a copy of a matrix and the transpose of a matrix have been added to the API.
  • Example 9, which demonstrates the new API routines, has been added.
  • Fix:
  • API routines WnMatrix__clear and WnMatrix__scaleMatrix have been fixed.
  • Example codes no longer have any cases of incorrectly assigning integers instead of double as values to matrix elements.
  • Makefile now compiles with GNU GSL recommended flags for gcc.
  • Example in documentation for API routine WnMatrix__Line__getNumberOfElements is now correct.
  • Documentation for API routine WnMatrix__getDiagonalElements is now correct.


• Version 0.2 (2007/07/06)
  • New:
  • Initial Release


• Version 0.1 (2006/11/08)
  • New:
  • Internal Release only.