Mathematical and Scientific Libraries on MCSR Systems

Various scientific and mathematical library packages are available on the various MCSR computing platforms.
These packages are described below.

• ARPACK
• BLAS
• LINPACK
• EISPACK
• LAPACK
• IMSL

On sweetgum, many of the most useful math/stat libraries are implemented in the older complib.sgimath , and the newer Scientific Computing Software Library (SCSL) package: scs. SCSL is also available as a shared library on redwood at /usr/lib/libscs.so .

ARPACK is a collection of Fortran77 subroutines designed to solve large scale eigenvalue problems. On sweetgum, ARPACK version 2 is implemented in the libraries /usr/local/appl/ARPACK/libarpack_SGI.a (sequential) and /usr/local/appl/ARPACK/libarpack_SGI_mp.a(parallel) . On mimosa, ARPACK is implemented in the library packages /usr/local/apps/ARPACK_PARPACK/ARPACK/libarpack_SUN4.a and /usr/local/apps/ARPACK_PARPACK/ARPACK/parpack_MPI-SUN4.a .

Mimosa ARPACK Compilation Examples

• f77 myfile.f /usr/local/apps/ARPACK_PARPACK/ARPACK/libarpack_SUN4.a
• pgf77 myfile.f /usr/local/apps/ARPACK_PARPACK/ARPACK/libarpack_SUN4.a

BLAS (Basic Linear Algebra Subprograms) are high quality "building block" routines for performing basic vector and matrix operations. On Sweetgum, BLAS is implemented in the SCSL library packages libscs.so (serial) and libscs_mp.so (parallel), located in /usr/lib32/mips4/ . Alternatively, an older implementation of BLAS is available in libcomplib.sgimath.so (sequential) and libcomplib.sgimath_mp.so (parallel), both also in /usr/lib32/mips4/ . On mimosa, BLAS is implemented in the library package /usr/lib/libblas.a , among others. On Redwood, BLAS is contained in /usr/lib/libscs.so and /usr/lib/libblas.a .

Mimosa BLAS Compilation Examples

• f77 myfile.f /usr/lib/libblas.a
• pgf77 myfile.f /usr/lib/libblas.a

Redwood BLAS Compilation Examples

• ifort myfile.f /usr/lib/libblas.a

LINPACK is a collection of Fortran subroutines that analyze and solve linear equations and linear least-squares problems. LINPACK has been largely superceded by LAPACK, which has been designed to run efficiently on shared-memory, vector supercomputers. On sweetgum, LINPACK is implemented in /usr/lib32/mips4/libcomplib.sgimath.so and /usr/lib32/mips4/libcomplib.sgimath_mp.so .

Sweetgum LINPACK Compilation Examples

On sweetgum, to compile and link a sequential Fortran program that makes calls to LINPACK, you would include the following command line parameters after the source files when invoking any of the MIPSpro Fortran compilers (e.g., /opt/MIPSpro/f90 , /opt/MIPSpro/f77 ):

• -lcomplib.sgimath
• or
• -L/usr/lib32/mips4 -lcomplib.sgimath

EISPACK is a collection of Fortran subroutines that compute the eigenvalues and eigenvectors of nine classes of matrices. (LAPACK was designed to replace and improve upon EISPACK.). On sweetgum, EISPACK is implemented in /usr/lib32/mips4/libcomplib.sgimath.so and /usr/lib32/mips4/libcomplib.sgimath_mp.so .

LAPACK (Linear Algebra PACKage) is a library of Fortran 77 routines for solving linear algebra problems, especially on shared memory multiprocessors such as sweetgum. LAPACK was designed to replace LINPACK and EISPACK by implementing more efficient versions of each of their routines into one package. Many LAPACK routines themselves make use of BLAS routines. ScaLAPACK is a version for distributed memory architectures. On mimosa, LAPACK is implemented in the library packages /usr/lib/liblapack.a and /usr/lib/liblapack.so . On sweetgum, LAPACK is implemented in the SCSL library packages libscs.so (serial) and libscs_mp (parallel), located in /usr/lib32/mips4/ . Alternatively, an older implementation of LAPACK is available in libcomplib.sgimath.so (sequential) and libcomplib.sgimath_mp.so (parallel), both also in /usr/lib32/mips4/ . On redwood, LAPACK is available in the library /usr/lib/liblapack.a .

Sweetgum LAPACK Compilation Example

On sweetgum, to compile and link a parallel Fortran program that makes calls to LAPACK and ARPACK, you would include the following command line parameters:

• -L/usr/lib32/mips4
• -L/usr/local/appl/ARPACK
• -lscs_mp
• -larpack_SGI_mp

Mimosa LAPACK Compilation Examples

• f77 myfile.f /usr/lib/liblapack.a
• pgf77 myfile.f /usr/lib/liblapack.a

Redwood LAPACK Compilation Examples

• ifort myfile.f /usr/lib/liblapack.a
• efc myfile.f /usr/lib/liblapack.a

IMSL (International Mathematical Statistical Library) has its own page.