MULTEM

Introduction

MULTEM is a collection of routines written in C++ with CUDA to perform accurate and fast multislice simulations for different TEM experiments as: HRTEM, STEM, ISTEM, ED, PED, CBED, ADF-TEM, ABF-HC, EFTEM and EELS. It is developed by Ivan Lobato (Ivanlh20@gmail.com).

Currently, there are three supported ways to use MULTEM:

C++: using the library itself
Matlab: using the mex interface
GUI: using the user graphical interface

The library is under heavy development and subject to change. The Matlab interface is the recommended way for researchers.

Remarks

In order to use the GPU capability of MULTEM, you need a Nvidia Graphic card with compute capability greater than 2.0 and CUDA 10.0 installed in your operating system. You can check the compute capability of your graphic card using the following nvidia website: https://developer.nvidia.com/cuda-gpus.

Using precompiled GUI interface

The precompiled GUI interface is only available for Windows operating system.

Go to https://github.com/Ivanlh20/MULTEM/releases and download MULTEM_binary.7z.
Execute vc_redist.x64.exe located in gui_bin folder.
Execute multem.exe.

Using precompiled mexfiles for Matlab

The precompiled mexfiles are only available for Windows operating system and Ubuntu 18.04-based Linux distributions.

Go to https://github.com/Ivanlh20/MULTEM/releases and download MULTEM.zip.
Execute vc_redist.x64.exe located in mex_bin folder. (Windows only)
Add the following folders to the Matlab path: crystalline_materials, matlab_functions and mex_bin.
Run the examples located in 'mex_examples_multem'.

Building MULTEM for Matlab

The following steps work using Matlab 2020b and CUDA 10.0. It assumes that Visual studio 2017 community, g++7.5 or Clang(Xcode 10.x) compiler is installed in your operating system. Additionally, Multem also requires fftw3, blas and lapack libraries.

Firstly, a C++ compiler must be set for Matlab by executing the following comand: mex -setup cpp. Be aware that Matlab 2020b only supports the above compilers.
Then add the following folders to the Matlab path: crystalline_materials, matlab_functions and mex_bin.
Run the compile_mex_multem.m script. This will create the required executable files to run the examples.
Run the examples located in mex_examples_multem.

Troubleshooting

If MULTEM does not compile with the above procedures, one of the following procedures might fix it
Currently (v2.2.3) some of the files in the Thrust library, that ships with Cuda 10.0 are incompatible with the Multem source code. To compile Multem, this library should be replaced with the Version of Cuda 8.0. The files can be found in the MULTEM repository.

for Windows:
- Verify the installation of Visual studio 2017 community.
- Verify the installation of Cuda 10.0 (https://developer.nvidia.com/cuda-downloads).
- Replace the Thrust library folder in location: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0\include\thrust
for Linux:
- Verify that gcc-7.5 and g++7.5 are the default compilers installed in your operating system. In Ubuntu, it can be installed by executing the following commands:
```
sudo apt-get update
sudo apt-get install gcc-7.5 g++-7.5
```
- Verify the correct installation of Cuda 10.0 (https://developer.nvidia.com/cuda-downloads).
- Verify the installation of fftw3 libraries. In Ubuntu, it can be installed by executing the following command:
```
sudo apt-get install libfftw3-dev libfftw3-doc
```
- Verify the installation of blas and lapack libraries. In Ubuntu, it can be installed by executing the following command:
```
sudo apt-get install libblas-dev liblapack-dev
```
- Replace the Thrust library folder in location: /usr/local/cuda/include/thrust
Verify the installation path of cuda 10.0, fftw3, blas and lapack. Their installation paths should be specified in the ilm_mex.m.

Please cite MULTEM in your publications if it helps your research:

    @article{LVAV16_1,
      Author = {I.Lobato and S.Van Aert and J.Verbeeck},
      Journal = {Ultramicroscopy},
      Title = {Progress and new advances in simulating electron microscopy datasets using MULTEM},
      Year = {2016},
  	  volume  = {168},
      pages   = {17-27}
    }
    
     @article{LD15_2,
      Author = {I. Lobato and D. Van Dyck},
      Journal = {Ultramicroscopy},
      Title = {MULTEM: A new multislice program to perform accurate and fast electron diffraction and imaging simulations using Graphics Processing Units with CUDA},
      Year = {2015},
  	  volume  = {156},
      pages   = {9-17}
    }

if you use our parameterization of the electronscattering factors, please cite the following article:

	@Article{LD14_1,
  	Title = {{An accurate parameterization for the scattering factors, electron densities and electrostatic potentials for neutral atoms that obey all physical constraints}},
  	Author = {I. Lobato and D. Van Dyck},
  	Journal = {Acta Crystallographica Section A},
  	Year = {2014},
  	Pages = {636-649},
  	Volume = {70}
  }

dnjohnstone/MULTEM