SFG2D

Python Toolkit to analyze 2d-sfg spectra and more.

Free software: MIT license
Documentation: https://sfg2d.readthedocs.io.

Pre Install

The following is mostly a collection of commands and steps one might need to do. If you run all of them brainlessly, this will most likely not work.

Easiest is, to just install anaconda on you platform. You can get it from:

Get Anaconda https://www.continuum.io/downloads#windows tested with Anaconda 4.2.0

Otherwise you have to setup your own python environment with a command prompt and pip as package manager. Cygwin might be a way to go here, but I haven't tried.

Install git for Windows from: https://git-scm.com/download/win

The following commands can all be executed from within the Anaconda Promt.

Install

Download source with:

git clone git@github.com:deisi/SFG2D.git

or download manually and extract the folder somewhere.

From within the source folder, run:

pip install .

This installs the minimal user version of sfg2d. It currently supports:

Windows 32 and 64 bit python 3.5
Linux 64 bit and python 3.5

To install on other platforms, see the following section about installing the development version.

If you actually want to do something with the package, like really use it, install the additional requirements. From within the package folder:

pip install -r requirements.txt

If you want to use the dashboards run:

jupyter nbextension enable --py --sys-prefix bqplot

and:

jupyter dashboards quick-setup --sys-prefix

Install Development Version from Source

Some dependencies must be pre installed manually:

pip install numpy cython

Download package from github:

git clone git@github.com:deisi/SFG2D.git

From within the package folder:

pip install -r requirements.txt
pip install -r requirements_dev.txt
pip install -r requirements_fit.txt
pip install -e .

If the requirements_fit.txt installation fails. You must install iminuit and probfit manually.

If you want to use the dashboards run:

jupyter nbextension enable --py --sys-prefix bqplot

and:

jupyter dashboards quick-setup --sys-prefix

Installation on Windows

Get Anaconda as described in the Pre Install section.

Iminuit

Download from http://www.lfd.uci.edu/~gohlke/pythonlibs/#iminuit Install with:

cd Downloads
pip install iminuit-*.whl

Probfit

Currently probfit must be compiled manually. To do so, we need visualstudio. Download from http://landinghub.visualstudio.com/visual-cpp-build-tools Start installation and grab a cup of coffee. For me this took +1 hour to complete....

Install probfit by hand. Download from https://github.com/iminuit/probfit Install with:

cd probfit
pip install .

Install sfg2d with:

pip install -r requirements.txt
pip install -r requirements_dev.txt
pip install -e .

Virtualenv

On Arch Linux

Install python-virtualenv:

pacman -S python-virtualenv

Install virtualenvwrapper:

pacman -S python-virtualenvwrapper

Setup virtualenvwrapper put:

source virtualenvwrapper.sh

in ~/.bashrc or ~/.profile

Setup a virutalenv on linux with:

mkvirtualenv -a ~/SFG2D/ -p python3.5 sfg2d

Enter virtualenv with:

workon sfg2d

note: change the path at ~/SFG2D to the location of sfg2d package. matplotlib needs a plotting backend to work from within a virtualenv, thus:

toggleglobalsitepackages pyqt

or use --system-site-packages during the setup of the virtualenv. See http://matplotlib.org/faq/virtualenv_faq.html for insights.

And follow the already described installation procedure.

On Windows with Conda

Run:

conda create -n sfg2d python+3.5 anaconda

Enter virtual env with:

activate sfg2d

And follow the above described installation procedure.

Jupuyter notebooks and virtualenvs

Up to now I have only done this under linux. To run jupyter/ipython kernel in the virtualenv I adopted the info from https://help.pythonanywhere.com/pages/IPythonNotebookVirtualenvs/

First create a new kernel with:

ipython3 kernelspec install-self --user

Now edit this kernel to use the sfg2d virtualenv by first moving it with:

mv ~/.local/share/jupyter/kernels/python3 ~/.local/share/jupyter/kernels/sfg2d

And then edit the ~/.local/share/jupyter/kernels/sfg2d/kernel.json and adjust the content to be similar to:

json
{
 "argv": [
  "/home/malte/.virtualenvs/sfg2d/bin/python3",
  "-m",
  "ipykernel",
  "-f",
  "{connection_file}"
 ],
 "display_name": "sfg2d",
 "language": "python"
}

The value of the display_name field is what jupyter will know the kernel by. The important line is the first argument of the argv. This must be the full path to the python3 binary within the virutalenv.

Test setup by running a notebook server:

jupyter notebook

Create a New Notebook and choose the sfg2d kernel from the drop-down menu and try to run:

import sfg2d

If there is trouble with missing PyQt, install it system wide and then link PyQt4 with the virtalenv. PyQt4 cant be installed via pip. e.g.:

ln -s /usr/lib/python3.5/site-packages/PyQt4 ~/.virtualenv/sfg2d/lib/python3.5/site-packages/

Officesetup

Requirements are installed into the home folder with pip3 using:

pip3 install --user --upgrade -r requirements.txt
pip3 install --user --upgrade -r requirements_dev.txt

Then created a virtual env with:

mkvirtualenv --system-site-packages --python=/usr/bin/python3.5 -a ~/sfg2d -r ~/sfg2d/requirements.txt sfg2d

And installed sfg2d from within the virtualenv:

pip install -e .

Because requirements are installed into the user folder outside of the virtualenv, we need to install the javascript nbextensions with:

jupyter nbextension enable --py --user widgetsnbextension

Description

This is a toolkit to analyze mostly sfg2d data with python3 using jupyter notebooks. It is not really generic, but rather specific to the problems and tasks I have to encounter here at the MPIP. It is nowhere near stable and things might change drastically at any point in time. If you want to use this I encourage you to create you own fork and work with your own version. At the time of writing, there is also almost no documentation available. I think this will change in time when things become more stable but up to now. Its not worth documenting much since it might be different next time anyway.

Features

Import data from Veronica, Viktor and .spe (version 2 and 3) files.
Data-structure based on pandas DataFrames to organize ans structure data.
A dashboard for the Viktor lab.
A minimal fit gui (dashboards/fit_starter/fit_starter.ipynb)
Import .spe spectra files
Import .ntb surface tension files

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

The .spe file importer is based on the code of James Battat, Kasey Russell and

For the structure of the module I was inspired by the Scikit packages.

deisi / SFG2D