Installation Notes for Linux: (1) We use BLAS and LAPACK linear algebra FORTRAN libraries. (Should a linear algebra package different from LAPACK be used, you will need to write your own interface code in src/linalg.c and change init.mk accordingly.) Check that BLAS and LAPACK libraries are installed on your system. They are included in various linux distributions, e.g. as /usr/lib/libblas.a and /usr/lib/liblapack.a The homepages of both packages are http://www.netlib.org/blas/ and http://www.netlib.org/lapack/ respectively. Hardware optimized versions of BLAS and LAPACK can also be used, see e.g. ATLAS http://math-atlas.sourceforge.net/ Note that the calling names of the FORTRAN routines in our LAPACK interface code in TSFIT/src/linalg.c are valid for linux and gcc (lower case names with underscore appended at the end). These may need adjustment on a different platform. (2) Download TSFIT from http://www.staff.uni-giessen.de/odavydov/tsfit/index.html and type gunzip -c tsfit-1.00.tar.gz | tar xvf - This will unpack the package into the directory TSFIT. (3) Edit lapack.mk to edit LINALG_INC which gives the paths to the BLAS and LAPACK libraries. LINALG_LIB is left blank in lapack.mk, but it might be needed if a different linear algebra library is used. (4) Edit init.mk to adjust it to your system. LINALG_INTERFACE is set to LAPACK and lapack.mk is included, both to be replaced if a different linear algebra library is used. Choose whether gcc or g++ compiler is used by editing the definition of CC in init.mk (5) From TSFIT directory, type make This will create the library lib/libTSFIT.a and example binaries in the directory examples. (6) To test the installation, type cd examples example1 Test output will be created (output1.txt) which contains grid evaluations of approximate Franke test function computed using scattered data from f1_ds3. Exact evaluations are in f1.txt. The Matlab/Octave script (script.m) allows to compare output1.txt with f1.txt, producing graphical output and the following text: spline = output1 max = 0.0130 rms = 0.0023 mean = 0.0014 Files README_EXAMPLES and FUNCTIONS contain further information about examples and high level functions used in them. (7) Examples 5-8 require additional data sets that can be downloaded at http://www.staff.uni-giessen.de/odavydov/tsfit/data_sets.tar.gz Store data_sets.tar.gz in TSFIT/examples, and type gunzip -c data_sets.tar.gz | tar xvf - ************************************************************************************************** Installation Notes for Windows Visual Studio C++ 2010: (1) Install BLAS/LAPACK libraries for Windows on your computer. We have tested the following pre-built static 32 bit library: libblas.dll, libblas.lib, liblapack.dll, liblapack.lib downloadable from http://icl.cs.utk.edu/lapack-for-windows/libraries/VisualStudio/3.4.1/Dynamic-MINGW/Win32/ In addition, MinGW runtime DLLs libgfortran-3.dll, libquadmath-0.dll and libgcc_s_dw2-1.dll will be needed at run time. See http://icl.cs.utk.edu/lapack-for-windows for further details. NOTE: This BLAS/LAPACK library is built with Fortran name mangling assumed in TSFIT (lower case plus an underscore: see dgesdd_ dgglse_ dgemv_ dgemm_ in linalg.c). If a different name mangling convention is used, please edit linalg.c accordingly. (2) Create a new Win32 Console Project (3) Add TSFIT source files: Store them on your computer, then right click in the Project Explorer window on Source Files, choose Add, Existing Item and add all *.c and *.h files of src as well as one of the example files, e.g. examples/example1.c (4) Link the header files directory: Choose Project, Properties, in the new window go to VC++ Directions, Include Directories and add the path to the src directory of TSFIT on your computer (5) Link BLAS/LAPACK libraries to your project: Choose Project, Properties, in the new window go to Linker, then (a) go to General, Additional Library Directories and add the path to the directory where BLAS/LAPACK lirary files are stored, (b) go to Input, Additional Dependencies and add the names of *.lib files of the BLAS/LAPACK library, in the above case libblas.lib, liblapack.lib (6) Add preprocessor definitions: again choose Project, Properties, in the new window go to C/C++, Preprocessor, and in Preprocessor Definitions add LAPACK and LINALG_INTERFACE=LAPACK (7) Turn off "precompiled headers:" in C/C++ as above, choose Precompiled Headers, and turn them off. (8) Make sure that BLAS/LAPACK DLLs (and MinGW runtime DLLs if using MinGW) and input data files (e.g. examples/f1ds3 for examples 1-4) are either in the path or in the working directory where the executable is stored by Visual Studio.