Skip to content

Latest commit

 

History

History
235 lines (187 loc) · 8.49 KB

README.md

File metadata and controls

235 lines (187 loc) · 8.49 KB
Raw

SkyNEt

NOTE: we have moved to our brainspy repository. This repository will not be maintained anymore. Please visit BraiNEdarwin for the latest version. This repository houses a collection of functions and scripts used by the Darwin team of the NanoElectronics group at the University of Twente. This README serves to outline the structure of the repo and its files. Additionally it gives some notes about how to contribute. For more details, please refer to the wiki. Before you start with your experiments, please read and understand how we handle data.

Installation

To begin with, install Anaconda (python3 version). For maintaining the code we use GitHub, so please make a GitHub account. To use GitHub on your computer you can use git in the command line, or if this does not ring a bell, we recommend using GitHub Desktop. To run the code in this repo, we make use of an Anaconda environment called skynet (based on the QCoDeS environment). To install this environment on a new PC, download the file environment.yml from this repo. Open an Anaconda prompt (or just a normal command prompt) and browse to the directory where you saved environment.yml. Then run the following command to install the skynet environment:

conda env create -f environment.yml

Now all that is left to do, is add the SkyNEt repository to the skynet environment, such that all script will be able to import it. To do this, go to your Anaconda prompt and activate the skynet environment by running:

activate skynet

Note that for non-windows users this will probably be source activate skynet instead.

Now open up an iPython console

ipython

And inside the console run the following commands:

import sys
sys.path

You will now see all directories where python will look for modules if you try to import one. There should be a path that looks something like this:

~/anaconda3/envs/skynet/lib/python3.6/site-packages/

This is the directory where you should place a path configuration file named skynet.pth. This will allow python to find the SkyNEt modules when imported. To do this, follow the instructions below:

  • Go to the directory you found above, i.e. ~/this/that/site-packages/
  • Make a file named skynet.pth containing a line with the absolute path to the directory containing your SkyNEt repo
  • Start your python and check sys.path; you should see the path to the repo there

Note: by convention, you should import SkyNEt explicitly, i.e. in the python sys.path do not include the SkyNEt directory, only include the directory a level higher. Since Python will look for scripts on that path, we recommend you keep it separated from all your other scripts to avoid interference.

To finish off the installation process, there is one package left to install. Activate the skynet environment again and run the following command:

pip install nidaqmx

Now you are done with the installation process and ready to get to work! Feel free to ask any of the code maintainers (listed at the bottom of this document) for help.

Neural networks

If you want to work with neural networks, there is an extra dependency PyTorch, which you have to install yourself. As this is OS dependent, please have a look at the instructions on the website. Have a look at the various pages at the wiki.

Kinetic Monte Carlo

If you want to work with the kinetic monte carlo model, there are some other dependencies. First, follow the installation instructions which can be found on the separate repository for kmc on dopant networks, here.

This repository is included in SkyNEt as a git submodule. To start using it, run the following two commands.

git submodule init
git submodule update

If you would like to pull the latest changes from kmc_dn, run the following:

git submodule update --remote

Note: at present, the kmc_dn code is only known to work on linux.

Repository structure

The repository is structured in a few folders. The instruments folder contains instrument drivers. The modules folder holds various modules containing function and class definitions used to run experiments. The experiments folder is where all the experiments are stored. Each individual experiment has its own measurement script (.py file) and configuration class definition.

As for the branching structure, there is a master and a dev branch. The master branch is only for stable versions of the software. Important changes or new features are implemented in the dev branch, which at some point will merge into a new version in the master branch. Each user at NE uses a personal branch to run experiments and optionally change code.

Practical use

As mentioned above, make sure that there is a personal branch for you to run code in and experiment. If this is not the case, contact one of the code maintainers (see the end of this page). Assuming everything is installed correctly, this is the basic workflow for running an experiment.

  • Open an Anaconda prompt and activate the qcodes environment.
  • Browse to the directory containing the measurement script you wish to run.
  • Make sure that you are checked out to your personal branch in git!
  • Configure the configuration class definition file to your liking.
  • Run the measurement script by running the following command:
python <your_script>.py

Writing your own experiment scripts

Probably at some point you wish to write your own measurement scripts. There are a few instructions that you should follow in order to keep things coherent with the rest of the repository:

And please make sure you only work on your own branch, but feel free to suggest any experiments you think should go into the dev branch!

Setting up a new measurement PC

Usually, you will run experiments on a PC that is already setup for measurements. This section shows the steps you need to take te setup a new measurement PC (this is unneccesary for personal PCs).

First follow the installation instructions above. Then make sure that git can connect to the internet via the utwente proxy server:

  • Go to C:/users/
  • Open .gitconfig
  • Add the following
[http]
	proxy = http://proxy.utwente.nl:3128

If you want to use the ADwin for measurements, additionally do the following steps:

  • Install the ADwin software with the CD-rom in the lab
  • Install drivers for Startech USB31000S
  • Configure the ADwin via adconfig

Code contribution

If you wish to contribute to the code you are more than welcome to do so. If you see any bugs or come across issues you think need improvement, please raise an issue on this repository here. If you want to know more about writing pieces of code to aid development, please refer to the wiki.

Please be pro-active and discuss with the code maintainers if you think some code you wrote can be valuable to improve this repo; for example, if you work on an improved sampling procedure, an optimized code or found a better way to implement a function, do contact us to discuss the adaptation into our code.

Code maintainers

Currently the users with admin rights on this repo are: