Giter Club home page Giter Club logo

feast's Introduction

////////////////////////////////////////////////////////////////////////////////////////////
//////////////////////////////  The  FEAST Eigenvalue Solver     ///////////////////////////
//////////// ///////////////////////////////////////////////////////////////////////////////
Author: Eric Polizzi  
        Department of Electrical and Computer Engineering
        University of Massachusetts, Amherst                           
E-mail: [email protected] (for FEAST related questions please use:  [email protected])              
web   : http://www.ecs.umass.edu/~polizzi/feast  
/////////////////////////////////////////////////////////////////////////////////////////////


General Description: Solving Eigenvalue Problem of form AX=eBX or AX=eX for
                     obtaining eigenpairs within the search interval [Emin,Emax]


Reference: "Density-Matrix-Based Algorithm for Solving Eigenvalue Problems"
           E. Polizzi
           Physical Review B, Vol. 79, 11512 (2009)     
                                                                       

===================================================================================================
Version: 2.0   
Version Description: 

    * Solving Ax=ex or Ax=eBx, A is real symmetric or complex Hermitian, B is symmetric or Hermitian positive definite
    * SMP version (one node) and MPI-version (multi-nodes).
    * Source code and pre-compiled libraries provided for commun architectures The pre-compiled libraries do not require 
      any specific compilers (no runtime dependencies).
    * All FEAST interfaces compatible with Fortran (77 and 90) and C,
    * Reverse communication interfaces: Maximum flexibility for application-specific. Those are matrix format independent,
      inner system solver independent, so users must provide their own linear system solvers (direct or iterative) and 
      mat-vec utility routines.
    * Predefined driver interfaces for dense, banded, and sparse formats: Easy ("plug and play") and Optimized.
    * All FEAST libraries require BLAS and LAPACK packages (not included),
    * FEAST banded drivers use the SPIKE banded primitives (included)
    * FEAST sparse drivers use the MKL-PARDISO solver (not included).
    * FEAST dense drivers are not intended for performances using FEAST-SMP (scalability can be obtained using FEAST-MPI)
    * Real/Complex and Single/Double precisions
    * Examples and documentation included,
    * Utility routines for sparse systems included: users can provide their sparse systems for quick testing, timing, etc. .
 ===================================================================================================



Library: FEASTROOT (this directory)
            |
  ----------------------------------------------------
    |       |       |        |        |        |
   src     lib   include    doc    example     utility 


**********************************************************************************************
**********************************************************************************************
**************** In the following, we briefly address the  question: *************************
****************         How should you use FEAST ?                   ************************
**************** All Details can be found in the FEAST documentation *************************
**********************************************************************************************
**********************************************************************************************

===================================================================================================
ATTENTION : It is recommended to define the Shell variable "export FEASTROOT="path to this directory"
===================================================================================================


LIBRARY PATH: (choose one)
              -L/$(FEASTROOT)/lib/x64 (for 64 bits architectures)
              -L/$(FEASTROOT)/lib/"yourown" (where "yourown" directory can contain the FEAST library
                                             compiled for your own specific architectures and optionally 
                                              for FEAST-MPI your own MPI implementation)

INCLUDE PATH: It is mandatory only for C codes along with instructions that need to be added in the header C file (all that apply):
              #include "feast.h"
              #include "feast sparse.h"
              #include "feast banded.h"
              #include "feast dense.h"


COMPILATION LINKS: (Example/ choose one)
		   -lfeast                (FEAST Reverse Communication Interfaces-kernel)
                   -lfeast_dense  -lfeast (FEAST dense interfaces/drivers,  also includes -lfeast_rci)
                   -lfeast_banded -lfeast (FEAST banded interfaces/drivers, also includes -lfeast_rci)
                   -lfeast_sparse -lfeast (FEAST sparse interfaces/drivers, also includes -lfeast_rci)
                   -In general: Any FEAST-drivers combinations follows by -lfeast 
                   

DEPENDENCY:
                   - the libraries LAPACK/BLAS (not provided) are needed
                   - the library MKL-PARDISO (not provided) is only needed for -lfeast_sparse
                   - the Intel MKL library is highly recommended and contains pardiso,lapack and blas 
              

EXAMPLES of COMPILATION LINKS with DEPENDENCY: (Example/ choose one)

             include dense drivers+rci        1/ -lfeast_dense  -lfeast -lapack -lblas
             include banded drivers+rci       2/ -lfeast_banded -lfeast -lapack -lblas
             include sparse drivers+rci       3/ -lfeast_sparse -lfeast -lpardiso -lapack -lblas
             include rci                      4/ -lfeast -lapack -lblas


REVERSE COMMUNICATION INTERFACES (RCI): TFEAST_{S,H}RCI ("T" denotes S,D,C or Z)
                    -routines are then called:
                          SFEAST_SRCI, DFEAST_SRCI (Real Symmetric resp. single and double precision) 
                          CFEAST_HRCI, ZFEAST_HRCI (Complex Hermitian resp. single and double precison)
                    -Matrix free format (FEAST is independent of the matrix data format) 
	            -users must then provide their own direct/iterative solver for solving the linear 
                     system-type  MY =F
                    -A preconditioner can be also used for iterative solver
                    -user must provide his own mat-vec
                    -for Complex systems, FEAST can be accelerated if user can provide a routine for
                     solving the linear system-type  M^H Y=F (H for Hermitian) knowing only the 
                     factorization of M (case 21)

               
PREDEFINED DRIVER INTERFACES:  TFEAST_XYY 
                                          *"T"  denotes S or D for real single/double precision 
                                                        C or Z for complex single/double precision 
                                          *"YY" denotes GV for generalized eigenvalue problem
                                                        EV for standart eigenvalue problem  
                          
                    - If "X"={SY,HE} :: Dense format  (ATTENTION: current version is not intended for obtaining performance)
                      uses LAPACK inner system solver:
                    - If "X"={SB,HB} :: Banded LAPACK Format  
                      uses SPIKE primitives inner system solver (included)
                    - If "X"={SCSR,HCSR} :: Sparse Format  
                      uses PARDISO inner system solver (not included)




feast's People

Contributors

certik avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.