MUMPS在CentOS下的安装

1.下载安装包MUMPS_5.0.1.tar.gz   这是目前最新版本的安装包。下载地址 http://mumps.enseeiht.fr/ 但下载需要提交申请，提交后隔天会发送到你邮箱里。或者在百度云下载 http://pan.baidu.com/s/1eQJfYRS

2. MUMPS需要连接下面这些库文件。这些安装包都可以在网上免费下载得到。

-> MPI

-> BLAS library

-> BLACS library

-> ScaLAPACK library

本人尝试了安装这些文件，但过程一波三折，最后可能是安装版本问题没有成功。由于我电脑处理器是intel的，我干脆安装Intel MKL，里面包含了这些库文件，不需要单独安装，省去了不少麻烦。

3. 安装IntelMKL。

     parellel_studio.xe_2016.tgz下载地址https://software.intel.com/en-us/qualify-for-free-software，这个包Intel只对部分人免费。如果是学生并有学校官方邮箱则可以通过Student身份申请。安装包很大有3.58G，里面包含了很多开发工具，包括icc，ifort，mpiicc,mpifort等，当然包含MKL库。为了避免后面安装MUMPS 兼容性问题，我用到了mpiicc和mpiifort编译器，所以我直接安装整个Intel安装包。安装整个包需要磁盘空间大约7G。如果在虚拟机上运行CentOS，碰到磁盘空间不足则需要磁盘扩容，参见http://blog.csdn.net/fancylovejava/article/details/8688405

将parellel_studio.xe_2016.tgz解压，进入解压包目录后 在终端运行

$./install_GUI.sh 实现界面安装，这个安装和window程序一样，但中间可以选择修改安装路径（默认是在/opt/intel）,如果之前进行了磁盘扩容，并选择空间挂载的方式，则路径选择在挂载的空间下。如/cm/opt/intel。（注：扩容磁盘空间被挂载到/cm/下，只能在这个路径下使用扩容的磁盘空间）。

4.安装MUMPS

解压MUMPS_5.0.1.tar.gz包后， 查看README和INSTALL文件，里面有一些简单的安装介绍。由于我用的Intel

MKL库。因此我选择了Make.inc文件夹下的Makefile.INTEL.PAR实现并行库编译。复制在MUMP_5.0.1根目录下，并改名Makefile.inc。内容如下。红色的是需要注意的和更改的项。蓝色的SEFTDEF路径是额外补上去的库。没有这个 库路径在编译过程中有些文件lib找不到。

#

#   This file is part of MUMPS 5.0.1, released

#   on Thu Jul 23 17:08:29 UTC 2015

#

#Begin orderings

# NOTE that PORD is distributed within MUMPS by default. If you would like to

# use other orderings, you need to obtain the corresponding package and modify

# the variables below accordingly.

# For example, to have Metis available within MUMPS:

#               1/ download Metis and compile it

#               2/ uncomment (suppress # in first column) lines

#                   starting with LMETISDIR,   LMETIS

#               3/ add -Dmetis in line ORDERINGSF

#                   ORDERINGSF   = -Dpord -Dmetis

#               4/ Compile and install MUMPS

#                   make clean; make    (to clean up previous installation)

#

#               Metis/ParMetis and SCOTCH/PT-SCOTCH (ver 6.0 and later) orderings are now available for MUMPS.

#

#SCOTCHDIR   = ${HOME}/scotch_6.0

#ISCOTCH      = -I$(SCOTCHDIR)/include   # Should be provided for pt-scotch (not needed for Scotch)

#

# You have to choose one among the following two lines depending on

# the type of analysis you want to perform. If you want to perform only

# sequential analysis choose the first (remember to add -Dscotch in the ORDERINGSF

# variable below); for both parallel and sequential analysis choose the second 

# line (remember to add -Dptscotch in the ORDERINGSF variable below)

#LSCOTCH      = -L$(SCOTCHDIR)/lib -lesmumps -lscotch -lscotcherr

#LSCOTCH      = -L$(SCOTCHDIR)/lib -lptesmumps -lptscotch -lptscotcherr -lscotch

LPORDDIR = $(topdir)/PORD/lib/

IPORD      = -I$(topdir)/PORD/include/

LPORD      = -L$(LPORDDIR) -lpord

#LMETISDIR = /local/metis/

#IMETIS      = # should be provided if you use parmetis, to access parmetis.h

# You have to choose one among the following two lines depending on

# the type of analysis you want to perform. If you want to perform only

# sequential analysis choose the first (remember to add -Dmetis in the ORDERINGSF

# variable below); for both parallel and sequential analysis choose the second 

# line (remember to add -Dparmetis in the ORDERINGSF variable below)

#LMETIS      = -L$(LMETISDIR) -lmetis

#LMETIS      = -L$(LMETISDIR) -lparmetis -lmetis

# The following variables will be used in the compilation process.

# Please note that -Dptscotch and -Dparmetis imply -Dscotch and -Dmetis respectively.

#ORDERINGSF = -Dscotch -Dmetis -Dpord -Dptscotch -Dparmetis

ORDERINGSF   = -Dpord

ORDERINGSC   = $(ORDERINGSF)

LORDERINGS = $(LMETIS) $(LPORD) $(LSCOTCH)

IORDERINGSF = $(ISCOTCH)

IORDERINGSC = $(IMETIS) $(IPORD) $(ISCOTCH)

#End orderings

########################################################################

################################################################################

PLAT      =

LIBEXT   = .a

OUTC      = -o 

OUTF      = -o 

RM = /bin/rm -f

CC = mpiicc

FC = mpiifort

FL = mpiifort

AR = ar vr 

#RANLIB = ranlib

RANLIB   = echo

# Make this variable point to the path where the Intel MKL library is

# installed. It is set to the default install directory for Intel MKL. 

MKLROOT=/cm/opt/mkl/lib/intel64

SCALAP = -L$(MKLROOT) -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64

INCPAR = -I/ cm/opt/mkl/lib/  #/usr/local/include

# LIBPAR = $(SCALAP)   -L/usr/local/lib/ -llamf77mpi -lmpi -llam

SEFTDEF = -L /cm/opt/compilers_and_libraries_2016.0.109/linux/mpi/intel64/lib/

LIBPAR = $(SCALAP)  $(SEFTDEF)  # seft add $(SEFTDEF)

#LIBPAR = -lmpi++ -lmpi -ltstdio -ltrillium -largs -lt

INCSEQ = -I$(topdir)/libseq

LIBSEQ   =   -L$(topdir)/libseq -lmpiseq

#LIBBLAS = -L/usr/lib/xmm/ -lf77blas -latlas

LIBBLAS = -L$(MKLROOT) -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core 

LIBOTHERS = -lpthread -openmp

#Preprocessor defs for calling Fortran from C (-DAdd_ or -DAdd__ or -DUPPER)

CDEFS    = -DAdd_

#Begin Optimized options

OPTF      = -O   -DALLOW_NON_INIT -nofor_main

OPTL      = -O -nofor_main

OPTC      = -O

#End Optimized options

INCS = $(INCPAR)

LIBS = $(LIBPAR)

LIBSEQNEEDED =

修改完成后执行命令make all lib。则会在    MUMPS_5.0.1/lib 目录下生成libcmumps.a    libdmumps.a   libmumps_common.a   libpord.a   libsmumps.a   libzmumps.a 这6个lib文件。（注：如果只执行make，则只会生成libdmumps.a   libmumps_common.a libpord.a 其余文件不生成）