Software ATS Advanced Terrestrial Simulator

编译ATS - 在MacOS下

Posted by CryoECNU on February 24, 2021

ATS 是一个非常强大的生态、水文建模工具，包括了热过程、蒸散发、表面能量平衡、积雪、生物地球化学、植被动态、变形、传输等过程。

如何编译安装ATS，下面的安装说明是直接拷贝过来的。后面会再记录一下自己的编译安装过程。

在Mac下面编译了一次，似乎没有想象那么困难，用这个脚本已经几乎是傻瓜式的编译安装过程了。 在编译之前，要安装cmake、gcc、lapack，我一般用macports安装。 然后，根据安装指南，配置相关的路径，其实也只要修改1个路径：

# EDIT THESE!
export ATS_BASE=～/Documents/ats
export ATS_BUILD_TYPE=Release
# END EDIT THESE!

后面的照抄就是了。

要在Bash里面添加这些路径，也就是修改~/.bash_profile 或者～/.bashrc

后续的编译过程也要在bash下进行，但是编译完成之后，可以在zsh下调用运行。

修改完之后，运行一下：

source ~/.bash_profile

1) 创建ats的根目录

mkdir -p ${ATS_BASE}
cd ${ATS_BASE}

2) 从github把两个库都弄下来：

 git clone -b amanzi-0.88 http://github.com/amanzi/amanzi $AMANZI_SRC_DIR
 git clone -b ats-0.88 http://github.com/amanzi/ats $ATS_SRC_DIR

3) 运行安装脚本编译amanzi

. ${ATS_SRC_DIR}/amanzi_bootstrap.sh

如果需要指定编译器的话，比如homebrew安装的gcc，都是gcc-7之类，需要编辑这个amanzi_bootstrap.sh文件：

# Known compiler lists
known_c_compilers="gcc-7 icc clang"
known_cxx_compilers="g++-7 icpc clang++"
known_fortran_compilers="gfortran-7 ifort"

注意顺序，因为程序会自动按顺序搜寻编译器。

基本上不用管它，只有个别情况下，国内网络不通顺，有些安装包下不下来，可以手动下载之后，放到：

${ATS_BASE}/amanzi-tpls/Downloads

里面。但是要注意，一定是ATS需要下载的那个链接对应的文件，否则MD5验证过不去，ATS还是要去下载原文件。

4）编译ATS：

. ${ATS_SRC_DIR}/configure-ats.sh

注意：如果需要指定编译器的话，比如homebrew安装的gcc，都是gcc-7之类，需要编辑这个${ATS_SRC_DIR}/CMakeLists.txt文件：

# which compilers to use for C and C++
set(CMAKE_C_COMPILER   /usr/local/bin/gcc-7)
set(CMAKE_CXX_COMPILER /usr/local/bin/g++-7)
set(CMAKE_Fortran_COMPILER /usr/local/bin/gfortran-7)

ATS Installation Guide

First, a word of warning – please be patient. Amanzi and therefore ATS depend upon a lot of third party libraries. This allows us to use existing, mature code to make ATS a much better software tool. It also means installing the code and its dependencies can be quite painful and time/labor intensive. Installation time is a very bimodal distribution – if it “just works” this process will take 10=20 minutes. If it doesn’t “just work” it can take much longer.

The basics of the process are these.

  1. Install pre-reqs.
  2. Download source.
  3. Use bootstrap to build Amanzi and its TPLs
  4. Use cmake to build ATS.
  5. Download test problems, test ATS.

All instructions assume you use bash. Change as needed for other shells.

  1. Ensure you have cmake and an MPI installation.
  • cmake >= 3.3 
    which cmake  # this will test if you have any cmake installed

    If not, go to: http://www.cmake.org/cmake/resources/software.html To install cmake from command line see https://cmake.org/install/

    Alternative, most binary installations are new enough, including Homebrew, Ubuntu, etc.

  • MPI 
    which mpicc # see if you have any MPI installed.
locate mpi.h # see if you have header files installed (is this enough?)

    On ubuntu, these are available with:

    sudo apt-get install libopenmpi-dev openmpi-bin  # I believe these are the sufficient conditions?

    On Mac, install mpi through homebrew or other package management systems.

  • Lapack (linear algebra package) On ubuntu, this is available with: 
    sudo apt-get install libblas3gf libblas-doc libblas-dev liblapack3gf liblapack-doc liblapack-dev # are there all needed?

    On Mac, install lapack through homebrew or other package management systems, i.e.

    brew install lapack
  1. Set up your directory structure. Life gets much easier if you get this right first.
  • I use the following: 
    # EDIT THESE!
export ATS_BASE=/my/path/to/all/things/ats
export ATS_BUILD_TYPE=Release
# END EDIT THESE!

export ATS_SRC_DIR=${ATS_BASE}/repos/ats
export ATS_BUILD_DIR=${ATS_BASE}/ats-build-${ATS_BUILD_TYPE}
export ATS_DIR=${ATS_BASE}/ats-install-${ATS_BUILD_TYPE}

export AMANZI_SRC_DIR=${ATS_BASE}/repos/amanzi
export AMANZI_BUILD_DIR=${ATS_BASE}/amanzi-build-${ATS_BUILD_TYPE}
export AMANZI_DIR=${ATS_BASE}/amanzi-install-${ATS_BUILD_TYPE}

export AMANZI_TPLS_BUILD_DIR=${ATS_BASE}/amanzi-tpls-build-${ATS_BUILD_TYPE}
export AMANZI_TPLS_DIR=${ATS_BASE}/amanzi-tpls-install-${ATS_BUILD_TYPE}
export PATH=${ATS_DIR}/bin:${AMANZI_TPLS_DIR}/bin:${PATH}
export PYTHONPATH=${ATS_SRC_DIR}/tools/utils:${PYTHONPATH}

    Note two things here – first, you may want to build both a ATS_BUILD_TYPE=Debug and ATS_BUILD_TYPE=Release builds. Debug is extremely useful for catching errors in your input files, while Release is significantly faster. I HIGHLY RECOMMEND BUILDING BOTH, then using Debug until your input files work and give reasonable results, then swapping to Release to do production runs.

    You may want to put these lines in your ~/.bashrc or similar files (~/.bash_profile on Mac OS X), or better yet use Environment modules.

    Then make your base directory and go there:

    mkdir -p ${ATS_BASE}
cd ${ATS_BASE}
  • Clone the Amanzi source for the latest release. Currently this is 0.88 
    git clone -b amanzi-0.88 http://github.com/amanzi/amanzi $AMANZI_SRC_DIR
  • Clone the ATS source for the latest release. 
    git clone -b ats-0.88 http://github.com/amanzi/ats $ATS_SRC_DIR
  1. Configure and build the Amanzi TPLs and Amanzi.
  • Run bootstrap. An example usage of bootstrap is in ${ATS_SRC_DIR} at amanzi_bootstrap.sh. If you follow the above instructions exactly, you can use that file as is, but maybe read it anyway, so you see what is happening. 
    . ${ATS_SRC_DIR}/amanzi_bootstrap.sh
  • Go get a cup of coffee. This will take a bit, from 10-15 minutes on a fast workstation to much longer on a cluster.
  1. Configure and build ATS.
  • Configure and build. Unfortunately, we have no bootstrap here, so you have to write your own cmake script. Again, if you follow these instructions exactly, you can use the sample at ${ATS_SRC_DIR} at configure-ats.sh 
    . ${ATS_SRC_DIR}/configure-ats.sh
  • Twiddle your thumbs. ~1-5 minutes?
  1. Download ats-demos to get at some examples, and run one!
  • Get the example repo: 
    cd $ATS_BASE
mkdir testing
cd testing
git clone -b ats-demos-0.88 http://github.com/amanzi/ats-demos
  • Run a test problem. 
    cd ats-demos
cd 01_richards_steadystate/richards_steadystate
ats --xml_file=../richards_steadystate.xml &> out.log
  • Visualize the results 
    cd ..    
jupyter notebook richards-steadystate.ipynb