LAMMPS Input file Running on HPC Cluster

I have created batch_submit.sh file in my home (rs0043) to submit my input files into the cluster node (sp0045). compilers are loaded accordingly. and then I submit the input file named as in.sp.0.001.mp by using the following command

lmp_serial<in.sp.0.001.mp

For multiple file submission by a single batch_submit I have used batch_submit_all_20.sh where I have created few variables those are substituted by desired values by running the bash file named multijob_within_script_for_multijob. This bash file is self-explanatory and it is run by the following command,

sh multijob_within_script_for_multijob


DONE.

LAMMPS Input File RUNNING

Here I am explaining about the things one needs to know just to run the input file with varying the star data file and strain rate.

To run the lammps input file: (FOR WINDOWS)

1.       Put the input file (i.e. in.sp.0.002.mp) inside a folder. Remember the input file name will always start with “in.” . You can put whatever name you want after that. I tried to use the strain rate value to remember the strain rate value I used inside.

2.       Copy the star data file and paste inside the same folder.

3.       Copy the lammps executable (lmp_serial) and paste into the same folder with input file.

4.       Copy the libOpenCL.dll file and paste it to the same folder location.

5.       Now open the “Run” command

6.       Type “cmd” and hit “enter”

7.       A command prompt will pop up

8.       Navigate to the folder where you have pasted all the above files

9.       Then type lmp_serial<input_file_name (ie. Lmp_serial<in.sp.0.002.mp) and hit “Enter”. The simulation will start running.

(To run in LINUX/Mac):

1.       Same procedure. (except it does not need the .dll file)

2.       In the command prompt after navigating to the desired folder instead of typing lmp_serial<input_file name you just have to type ./lmp_serial<input_file_name. Here you have to create your executable by yourself and you could name it as you like. (I named the executable same as the Windows for remembering easily).

“DONE”

The input file has two things to change that would cause different different output.

1.       Star file name

2.       Strain rate values

STAR FILE NAME

The star data file is generated by Python script and it contains the coordinates of star particles.

The input file Line number 10 (link attached) has command :

read_data  data.star fix molprop NULL Molecules

here data.star is the star data file and the same named star data file should be present in the same folder with input file. I you want to change the name of star data file you have to change in both place (in folder and inside the input file).

STRAIN RATE VALUES

In the line number 118 has the fllowing command:

variable srate1 equal 0.002

the value 0.002 is the strain rate. (I used this value to name input file, remember !)

https://drive.google.com/file/d/1xNu5pZfmT_G8MrzPWFIGuErbAqmthZxP/view?usp=sharing

P-CARD EXPENSES

Subject:	MyExpenses-- please share this with everyone on your team who has a p-card

All,

We have run into the same issue with P-card charges that we did with the travel… YOU HAVE TO KNOW ALL OF THE FUNDING NUMBERS IN ORDER TO SAVE ANYTHING IN THE RECORD. 

So, if you don’t know all 5 numbers, in the short term, we have a few options:

1.       You can fill out the p-card form (make sure that it tells us where to pay, and a SPECIFIC BUSINESS PURPOSE), attach the receipt and turn it in as usual… we will do the rest.

2.       You can upload or email your receipt to your account (receipt@chromerfile.com), making sure that you write EVERYTHING on that receipt that would be on the p-card form… please name it so that we know what charge it goes to…

3.       You can call us and we will give you the last 2 numbers that you must have in order for the charge to save.

Here’s the issue… we still have to get all of the p-card information into MyExpenses, and you still have to mark “received”, sign and date the receipt.  However, you cannot put any of this information into MyExpenses without knowing the entire funding. 

We are open to procedural suggestions, but the above options are the most efficient that we can come up with right out of the chute. 

PLEASE NOTE THAT WE WILL NO LONGER BE SENDING YOU A LIST OF THE CHARGES THAT NEED FUNDING—THIS LIST IS IN YOUR ACCOUNT, SO YOU CAN SEE EVERY CHARGE THAT HAS HIT.

We will help you in whatever way we can… please don’t let these charges sit out there and accumulate!

Thanks,

Pam

Group Presentations Spring Semester 2018

Date:                    speaker：                                              topic: 

Feb 1st                    Rongguang                                            friction


Feb 8th                    Hengxu                                   2D DDD simulation of pillar reloading


Feb 15th                  Stefanos                                   ARO/DOE proposal


Feb 22th                  Michael/Hengxu/Stefanos       Machine learning on indentation (3D DDD)             


Mar 1st                    Michael                             Machine learning on pillar compression (2D DDD)           

Mar 22th                  Josh                                      Crack growth


Mar 29th                 Ryder                                 Nanoindentation of Cu/Ni/Al


April 12th              Ryder/Stefanos                 Nanoindentation of different grain orientations (EBSD)


April 19th              Salehin                                          Jamming


April 26th              Hengxu                                     3D DDD indentation with flat punch


The group meeting is planned to start at 3:30PM (Thursdays).

We have skipped some group meetings due to the planned travels:
March 4th~March 19th, Stefanos and Hengxu, trip to LA (APS and UCLA)

April 2nd~April 6th, Stefanos and Hengxu, trip to Phoenix (MRS Spring)

May 1st~June 10th , Stefanos, trip to Europe (Stockholm, Paris, Groningen)

Using Globus

• Go to htpps://www.globus.org/ 
• Click on Log in 
• Select organization (West Virginia University) 
• You will be redirected to your mix account in order to log-in.
• A new screen will pop-up. Click on endpoint and type wvu#hpcdtn (both right and left panels)
• The pre-selected path is /gpfs/. For datadepot use /depot/
• You can transfer files from hpc to datadepot storage or other public available endpoints

SHARING FILES? 

• Connect online to globus
• Select endpoint and set path to /depot/ 
• Go to the directory you want to share and select the specific folder.
• Click on share (you will need to have permission from whomever is operating your endpoint.

• A new screen will pop-up:

• Host path is the name of the folder you selected for sharing (change it by clicking on browse - required )
• Share display name is the name of the shared folder you want others to see (required).
• If you want add a description and some keywords for the people you're sharing the folder.
• If all goes well, a new screen will pop-up:

• The path will reflect your choice of shared folder. (/) means you want to share all subfolders that exist inside the folder you selected for sharing. If you want, click on browse and select a specific subfolder.
• Select if you want to make your shared folder available to: Single user,group,public,all globus users.
• For single user you will have to type their e-mails. For groups you will have to type in the group name. Click on add.
• Select if the users you're sharing to will have readable and writable permissions.
• Click on add permission.
• Your folder is shared!

UPLOADING FILES FROM YOUR PC? MAKE IT AN ENDPOINT!

• If you want to transfer files from your campus storage platform to your personal computer/laptop you will have to connect your PC to globus (endpoint).
• Info for MAC OS X and how to make your Mac an endpoint: https://docs.globus.org/how-to/globus-connect-personal-mac/
• Info for Linux OS and how to make your linux machine an endpoint: https://docs.globus.org/how-to/globus-connect-personal-linux/
• The globus personal connect application should always be running if you want your PC to be an endpoint and move files from storage platforms to your PC and vice-versa!

installatio of OOF2

Details of installation for different system can be found here:

https://www.ctcms.nist.gov/oof/oof2/prerequisites.html


For Mac:

Two commonly used package managers on OS X are MacPorts and fink. Both should work. If you have neither installed already, it will be easier to use MacPorts.
Neither fink nor MacPorts will install the version of swig that OOF2 2.1.11 or earlier requires. See Installing SWIG, below, if you need it to install it. If you aren't building your own OOF2 extensions, and if you use --skip-swig when building OOF2, you don't need to install SWIG. If you're using OOF2 2.1.12 or later, you don't need to install SWIG.
To run OOF2, you will need to install X11, which Apple no longer supplies, but which can be downloaded from Mac OS Forge. It's slightly easier to get X11 from MacPorts, if you're using MacPorts.
When the instructions below tell you to type a command, type it in a Terminal window. You can copy and paste the lines from this page.

Macintosh OS X with MacPorts

1. Install MacPorts from http://www.macports.org/install.php, following their instructions. It's easiest to use the .pkg installer. If you get error messages when running "sudo port -v selfupdate", open a new Terminal window and try again.
   
2. Type these commands. Some of them may take a while to run, depending on whether or not your computer already has some of the dependencies installed.
   

         sudo port install pkgconfig
         sudo port install xorg-server
         sudo port install py27-numpy -atlas 
         sudo port install py27-pygtk
         sudo port select python python27
         sudo port install libgnomecanvas
         sudo port install google-perftools
   

   Note: If you prefer to use Python 2.5 or Python 2.6 instead of Python 2.7, replace the '27's in the commands with '25's or '26's.
   Note: The numpy line is there to tell MacPorts to install the version of numpy that doesn't use atlas. Atlas is a substitute for the blas libraries distributed with OS X, and takes a very very long time to install. If you either already have atlas installed or don't mind waiting for it to build, you can skip the py27-numpy installation step completely. In that case, the default, atlas-using, version of numpy will be installed automatically when you install py27-gtk.
   
3. The version of ImageMagick provide with older versions of MacPorts on older versions of OS X was incompatible with OOF2. You really should update your system, but if for some reason you're using OS X 10.6 or earlier, do this:
   
   1. Type these commands:
      

            mkdir -p ~/ports/graphics/ImageMagick
            cp /opt/local/var/macports/sources/rsync.macports.org/release/tarballs/ports/graphics/ImageMagick/Portfile \
                   ~/ports/graphics/ImageMagick
            cd ~/ports
            sudo portindex 

      Note: If you're using MacPorts version earlier than 2.0.4, replace the path in the second line with

      /opt/local/var/macports/sources/rsync.macports.org/release/ports/graphics/ImageMagick/Portfile

      
      
   2. Using any text editor (not a word processor!) edit ~/ports/graphics/ImageMagick/Portfile, appending these three lines:
      

            variant no_openmp description {Disable OpenMP} {
               configure.args-append   --disable-openmp
            } 

      
      
   3. Type this command:
      

            sudo port install ImageMagick +no_openmp

      
      
   
   
4. If you're using OS X 10.7 or later, type this commmand:
   

         sudo port install ImageMagick

   
   
5. If you're using Python 2.6 or 2.7, and you get this message when trying to build OOF2
   

     Can't find pygtk-2.0! Version 2.6 or later required

   then you need to work around a MacPorts bug by setting the PKG_CONFIG_PATH environment variable to /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/pkgconfig before building. Replace the 2.7 with 2.6 if you're using python 2.6. For example, if you're using the bash shell, type

     export PKG_CONFIG_PATH=/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/pkgconfig
   

   You can put the line in your .profile file in your home directory to make the change permanent, if you find yourself rebuilding OOF2 frequently for some reason.
   

Macintosh OS X with fink

We haven't used fink recently, so these instructions may be out of date. Please let us know if they are.

1. Install Xcode from Mac OS Forge.
2. Install fink, by going to http://www.finkproject.org/download/srcdist.php and following the instructions there. If the bootstrap script asks you to choose a mode and you're on a 64-bit capable Mac, choose 64bit-only.
   If you already have fink installed in 32 bit mode, you don't need to reinstall everything just to change to 64 bit mode. You may need to force ImageMagick to build in 32 bit mode in step 5.
   
3. Enable the "unstable" fink packages.
   (a) Run "sudo fink configure", and answer "y" when asked about the unstable tree.
   (b) Run these commands:
   

          sudo fink selfupdate
          sudo fink selfupdate-rsync
          sudo fink index -f
          sudo fink scanpackages 

   
   
4. Use fink to install the libraries that ImageMagick will require by typing these commands:
   

          sudo fink install libjpeg
          sudo fink install lcms
          sudo fink install libpng14-shlibs 

   
   
5. Build ImageMagick without OpenMP, which is incompatible with OOF2 on OS X. (NOTE! It may not be necessary to do this step. We haven't tried building OOF2 with fink using a recent version of either fink or OS X. You could try installing ImageMagick with fink by typing sudo fink install imagemagick7-dev and skipping this step. You'll know if it worked if you can successfully load and display an image in OOF2.)
   
   • Go to http://www.imagemagick.org. Click "Download". Choose a site to download it from. Download the file with the latest version number and the .bz2 suffix.
     
   • Uncompress and untar the package:
     

            tar -xjf ImageMagick-<version>.tar.bz2
     

     where <version> is 6.6.7-10 or something like that.
     
   • Run these commands:
     

             cd ImageMagick-<version>
             ./configure LDFLAGS=-L/sw/lib CPPFLAGS=-I/sw/include --disable-openmp

     
     
   • Check the output from the configure command. If it says that support for some image format that you need hasn't been included, you probably need to go back to step 4 and install more packages from fink. The "Packages" link on the fink website will let you search for the package name.
     (If the output indicates no support for formats for which you are sure the libraries have been installed, and if it's possible that you have a 32 bit version of fink, try reconfiguring ImageMagick with CPPFLAGS="-I/sw/include -m32" (including the quotation marks!) instead of CPPFLAGS=-I/sw/include. The OOF developers haven't tested this. Please let us know if it works or not.)
     
   • Build and install ImageMagick by running these commands:
     

             make
             sudo make install 

     
     
   
   
6. Use fink to install the other packages that OOF2 requires by typing these commands, some of which may take a while to run:

          sudo fink install python 
          sudo fink install pygtk2-gtk-py27-dev
          sudo fink install libgnomecanvas2-dev
   

Installing SWIG on OS X

NOTE: for OOF2 version 2.1.12 and later, it is no longer necessary to install swig.
The only obstacle to installing swig 1.1 build 883 on a Mac is that the Mac's file system is not case sensitive, and swig's build procedure tries to make a subdirectory named SWIG and a file named swig in the same directory. You can install swig on a case-insensitive disk, you just can't build it there. There are at least three ways to get around this inconvenience:

• Build swig on a disk with a different type of file system. Format the disk with the Unix File System, using Apple's Disk Utility application.
• Use Disk Utility to create a disk image, formatted with the Unix File System. After you open the disk image, it will act just like a real disk, and you can use it to build swig.
• Hack the swig source files so that it doesn't have a case problem. If you don't mind a second hand swig, you can download our modified version here:
  • swig1.1-883-mac.tar.gz

NetBSD

OOF2 version 2.0.5 and 2.1.1 or later will work on NetBSD. OOF2 version 2.1.0 does not work. These commands were sufficient to install OOF2 on a 32-bit version of NetBSD 5.1. Execute them from a root account. echo /usr/X11R7/lib >> /etc/ld.so.conf pkg_add pkg-config pkg_add py-gobject-shared pkg_add py26-gtk2 pkg_add libgnomecanvas pkg_add ImageMagick pkg_add freetype2 pkg_add fontconfig The first line may not be necessary if you set LD_LIBRARY_PATH. Installing the prerequisites will provide a program called "python2.6" in /usr/pkg/bin. You can use this to build oof2, but you'll have to invoke it explicitly: python2.6 setup.py build --skip-swig --library-dirs=/usr/pkg/lib install You could also create an alias or symbolic link called "python" that points to /usr/pkg/bin/python2.6, and then you can follow the instructions in README.md without change.

---

Installing Anaconda and Pymks

Go to https://www.continuum.io/downloads and click on your OS distribution (Mac,Linux,Windows).

For MacOS:

Download the graphical installer for Python 2.7 and install anaconda.

For Linux:

Download the Python 2.7 installer.

Open a terminal and type
cd ~/path_of_downloaded_anaconda_file/
bash Anaconda    (and press the tab key to get the full name)

The final command will run the installation and you're going to be asked a series of questions.
In general, Anaconda will be installed in /home/your_username/anaconda2.

If you want to install it somewhere else you will need to give a full path like: /home/your_username/Python_tools/anaconda2

For all other questions type yes (that is critical for smooth installation and easy use of Anaconda)

After you've installed Anaconda follow the instruction for PyMKS:

MacOS and Linux:

Open a terminal and type:

conda --add-channels pymks
pip install pymks
conda install vtk
pip install mayavi

With these 4 commands you have full functionality of Pymks in Jupyter-notebook using:
import pymks

If you want to run the jupyter notebook you can type in terminal jupyter-notebook, or anaconda-navigator and you can see the full set of tools installed with Anaconda.

MacOS users can open the the Anaconda Navigator in Applications and run Jupyter Notebook from there.