MINND

Neural network for performing inversion of MOSES data. MOSES is a slitless EUV spectrograph, designed to capture spatial-spectral images of the sun in EUV wavelenghts.

Installation Instructions

These instructions were tested using Linux Mint 18, using an nvidia GTX980 graphics card

Caffe has several dependencies, install the graphics card drivers only if you need to do GPU training. Otherwise you can use the CPU.

Installing the GPU dependencies (nvidia GPU only)

Installing the nvidia Drivers

Use the Linux Mint driver manager to install the latest nvidia drivers. The current version is nvidia-367.

Installing the CUDA libraries

Download the latest drivers. Select the buttons that describe your target platform. We used the Ubunut 16.04 .deb file. Once downloaded, enter the following commands to install the drivers

sudo dpkg -i cuda-repo-ubuntu1604-8-0-local_8.0.44-1_amd64.deb
sudo apt-get update
sudo apt-get install cuda

add the following lines to your .bashrc file to complete the installation

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
export CUDA_ROOT=/usr/local/cuda

Testing the CUDA libraries

Navigate to the CUDA samples directory

cd /usr/local/cuda-8.0/samples/

Use the deviceQuery and bandwidthTest programs to test if CUDA and the drivers installed correctly. Here are the instructions on how to compile and run deviceQuery

cd 1_Utilities/deviceQuery
sudo make
./deviceQuery

If you would like to test OpenGL, for example on the Mandelbrot sample, you have to provide the location of the OpenGL libraries since Linux Mint is not a supported distribution.

cd /usr/local/cuda-8.0/samples/2_Graphics/Mandelbrot
sudo make GLPATH=/usr/lib/
./Mandelbrot

Installing the cuDNN Libraries

The cuDNN libraries provides more graphics acceleration to Caffe.

Download them here using your nvidia developer account.

Select the cuDNN 5.1 Library for Linux, and extract it. Open and copy *.h to $CUDA_ROOT/includes and *.so* to $CUDA_ROOT/lib64.

Installing Caffe

Installing dependencies

Thre requried depencies for Caffe can be found at the following webpage These dependencies should be downloaded from the apt-get repositories before attempting to compile Caffe

sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler libboost-all-dev libopenblas-dev libgflags-dev libgoogle-glog-dev liblmdb-dev

Installing Caffe

Download the latest version of Caffe from the git repository

git clone https://github.com/BVLC/caffe.git

Prepare Makefile.config

Navigate into the cloned directory and modify the file Makefile.config.example.

cd caffe`
xdg-open Makefile.config.example

If using cuDNN, then uncomment the line xdg-open Makefile.config.example.

For Linux Mint 18 support, modify INCLUDE_DIRS in Makefile.config

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/

Next, select the OpenBLAS library by chaning the line BLAS := atlas to BLAS := open

and save the file as Makefile.config.

Prepare the Makefile

Modify the Makefile to work in Linux Mint by changing the line

LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_hl hdf5

to

LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial

We are finally ready to compile Caffe.

Compile Caffe

Compile Caffe using make:

make all -j8
make test -j8
make runtest -j8

Installing MINND

Dependencies

MINND relies on Qt4 for image display. Install using the apt repositories

sudo apt-get install qt4-qmake qt4-dev-tools qt4-default 

SPD_Meeting_2017

Repository containing documents and programs in preparation for a poster to be presented at the Solar Physics Division meeting 2017.

Installation Instructions

These instructions were tested using Linux Mint 18.1, using an nvidia GTX980 graphics card

Matlab

This guide assumes you have a working copy of Matlab. If you are a MSU student, you can download matlab here.

Sunpy

This project relies on Sunpy for AIA dataset acquisition. Install Sunpy using the following commands

sudo apt install pip
pip install pip install setuptools pytest sunpy[all]