PiCoPiC

PiCoPiC - PiCoPiC is a Code of Particle in Cell

Motivation

PiCoPiC is a successor of PDP3 (and its further incarnation) project. Developed as electromagnetic particle-in-cell code for collisionless plasma simulation, PDP3 designed as single-thread code. Late versions of PDP3 are partially parallelized, but it can not be fully parallelized and can not be pupulated to large number of threads (it still effective with 2-8 threads). Such critical limitation of PDP3 was a motivation to replace the project as new fully parallel PIC code, called PiCoPiC.

General Description

PiCoPiC still uses same algorithms as PDP3, but separates the simulation area to numerous domains, which are fully independent from each other, except synchronization points, when edges of their currents and fields are "stitched" with overlaying.

Build

user@host$ ./autogen.sh
user@host$ ./configure [OPTIONS]
user@host$ make

Blackbox Tests

user@host$ make test

Configuration

PiCoPiC uses JSON format of configuration files (default configfile is PiCoPiC.json). Example of config located in the root of this repository.

Output and data analysis

PiCoPiC follows the same principles for data output, as PDP3 data array output and python library and tools collection (python scripts and jupyter notebooks) for analysis of such data. Python library can be found in lib/python/picopic subdir and tools collection - in tools directory.

System And Software Requirements

C++ compiler: gcc 6.X or later (8.X is recommended)
autotools (autoconf, autoheader)
git (to get sources)
hdf5 library (1.12.X is recommmended)
'make' util
libgomp (OpenMP spec. version 3.0+).
python3 + jupyter + matplotlib + numpy for visualization (anaconda - python scientific environment is recommended)
doxygen, latex (texlive-full), imagemagick, pandoc, pandoc-citeproc (optional, for documentation only)

HOWTO (linux, debian/ubuntu example)

0. INSTALL PREREQUIRED SOFTWARE

Install prerequired software with standard package management system:

root@host# apt-get install build-essential autoconf

HDF5 library installation:

HDF5 format is the only output format for now (plaintext is deprecated). So, you should build HDF5 library as prerequirement. libhdf5 v.1.12.X is recommended. But, v.1.10.3 (shipped with current debian stable distribution) or higher can be used. In case of use different HDF5 version from default in your OS distribution, HDF5 should be compiled manually as prerequirement. PiCoPiC's build scripts contain helper for it (0.b item).

0.a. install HDF5 with package management system (perform 0.a, 0.b or install in some other way, at your option):

user@host$ sudo apt-get install libhdf5-cpp-100 libhdf5-dev

0.b. install HDF5 globally with "make" helper (perform project configuration first. See CONFIGURE AND BUILD PROJECT section):

user@host$ sudo apt-get install wget
user@host$ sudo make hdf5-install

1. GIT CLONE PROJECT

user@host$ git clone https://gitlab.com/my-funny-plasma/PIC/picopic.git
user@host$ cd picopic
user@host$ git submodule update --init --recursive # require to enable external libraries

2. CONFIGURE AND BUILD PROJECT

# change your current directory to project's root directory
user@host$ cd /path/to/picopic/
# 0.b.: optional step to install HDF5 library with built-in helper and reset project's build directory
user@host$ ./autogen.sh && ./configure && sudo make hdf5-install && sudo make distclesn
# end of 0.b.
user@host$ ./autogen.sh
user@host$ ./configure [OPTIONS] # use ./configure --help to view full options list
user@host$ make [COMPILE FLAGS] # see below about COMPILE FLAGS

3. TEST (optional)

Functional (end-to-end) testing:

user@host$ make test

Unit testing (not implemented yet):

user@host$ make test-unit

4. RUN

After compilation finished, you just need binary file PiCoPiC and PiCoPiC.json. You can copy this files to somewhere (if you skipped installation), edit PiCoPiC.json and run ./PiCoPiC.

user@host$ /path/to/PiCoPiC [ -h ] | [ -f /path/to/PiCoPiC.json ] # used PiCoPiC.json from current directory, if calling without any options
# or (much better), launch it as high-priority process
user@host$ nice -20 /path/to/PiCoPiC [ -h ] | [ -f /path/to/PiCoPiC.json ] # give some power to PiCoPiC!

NOTE: it can take several days or weeks, and several hundred gigabytes of diskspace (yep, it is science, my deer friend).

5. VISUALIZATION (generate plots or animation)

After your application finished modeling, you can build some visual model from generated data. Use python with matplotlib and numpy. Anaconda as python distribution is recommended.

...but... if you don't want to use anaconda...:

root@host# apt-get install python3 python3-matplotlib python3-numpy python3-scipy

Pre-required software (for animation):

root@host# apt-get install ffmpeg

Animation generation:

user@host$ /path/to/repository/with/picopic/tools/movie_3_E_RHObeam_r_z_map.py /path/to/data/directory [OPTIONS] # --help to get list of available options

Interactive data analysis:

Data analysis tools made as jupyter (ipython) notebooks. You can find them in tools subdir. Please, run jupyter from tools dir also

user@host$ cd /path/to/PiCoPiC
user@host$ jupyter <notebook|lab> [tools/<notebook name>.ipynb]

Non-interactive notebook launch:

To run python notebook in non-interactive mode, please, use script tools/nbrun.sh from project root directory

user@host$ /path/to/PiCoPiC/tools/nbrun.sh /path/to/PiCoPiC/tools/<notebook name`>.ipynb [data_path=\'/path/to/data/directory\'] [other_notebook_variable=other_notebook_value]

6. DOCUMENTATION GENERATION (optional)

Pre-required software:

root@host# doxygen texlive-latex-base texlive-latex-extra imagemagick graphviz

Generate:

user@host$ make doc

Find documentation in:

PDF doc/app/latex/refman.pdf - application doc/vis/latex/refman.pdf - visualization
HTML doc/app/html/index.xhtml - application doc/vis/html/index.xhtml - visualization

Hacking

Coding Style

Indentication: 2 spaces
Braces: BSD style
Naming:
- classes - CamelCase
- files - camelCase, same as class name, begining from lower case
- functions - snake_case, lower case
Spacing:
- after/around operators
- no space after unary operators
- no space before the postfix increment & decrement unary operators
- no space around the . and -> structure member operators
Declaring pointer data or a function that returns a pointer type: preferred to use of * is adjacent to the data name or function name and not adjacent to the type name
Comments: '//' at line begining
File naming: lowercase

GIT

Before working

Install git hooks

user@host$ cp tools/git_hooks/pre-commit .git/hooks

Make sure, that you have clean repository and there are no unstaged untracked and uncommited files.

user@host$ git status

NOTE: If you have such files, you commit, remove or move out of your working directory (with PiCoPiC), than reset repository Also, you can just reset repository, or remove your local changes in other way

user@host$ git reset --hard

Sync with upstream

user@host$ git fetch origin

Working

Checkout to new branch

user@host$ git checkout -b <your-branch-name> # use only `-` as word delimiters for better compatability

Make changes, write new code, add new features, experiment etc.
Commit your changes (do it as often, as you press Ctrl+s ;) ) and push to upstream

user@host$ git add . # or git add your/selected/files
user@host$ git commit -m "<your comment>"
user@host$ git push origin <your-branch-name>

After working

When you finish some logical step of your work, you should merge your changes to master branch (as stable code). You can do it, using "Pull request" in github

Go to https://gitlab.com/my-funny-plasma/PIC/picopic/merge_requests
Press "New merge request" and choose "master" as base branch and "your-branch-name" as compare. Scroll to view changes against master branch
When all is ok, press "Create pull request" and confirm.
Wait for travis testing ends and (if they passed), merge your changes to master branch (squash and merge).

Debug

Linux:

build project with DEBUG option

user@host$ ./autogen.sh && ./configure --enable-debug && make

run with gdb

user@host$ gdb ./PiCoPiC
(gdb) run ## or perform some modifications first than run, e.g. set breakpoints

view data during simulation run HDF5 format does not allow access to data, when it opened in read-write mode by other application. PiCoPiC allows to pause the simulation and unlock HDF5 data file. It accepts user signals SIGUSR1 and SIGUSR2 to pause-unlock and unpause-lock data file. The algorithm to do it is:

use@host $ killall -USR1 PiCoPiC // pause simulation-unlock
2020-10-10 19:19:19.42 ( 346.397s) [main thread     ]            PiCoPiC.cpp:70    INFO| Paused, unlocking data file
...........
## use PiCoPiC's visualization tools, HDFCompass or other tools to operate with HDF5 format etc
use@host $ killall -USR1 PiCoPiC
2020-10-10 19:19:42.19 ( 361.530s) [main thread     ]            PiCoPiC.cpp:77    INFO| Continue, locking data file

use experimental features If you want to experiment and use unsafe features, disabled by default, use "--enable-experimental" key during configuration:

user@host$ ./confugure --enable-experimental [other options] && make

Tools

Quick analytic calculator of plasma (aka Langmur) frequency, wake wavelength, Debye length etc. from PiCoPiC.json file

user@host$ ./tools/quick_parameters_calculator.py <path/to/PiCoPiC.json>

See VISUALIZATION section also

You can edit jupyter notebooks with jupyter browser editor (opens with jupyter notebook or jupyter lab commands), atom (plugin: https://atom.io/packages/jupyter-notebook), emacs (plugin: https://github.com/millejoh/emacs-ipython-notebook), spyder (plugin: https://github.com/spyder-ide/spyder-notebook) etc

Bugs/Workarounds

NOTE: It's recommended to disable HDF data file locking to avoid unreadable files, during incorrect application exit or other emergency stuff. Use export HDF5_USE_FILE_LOCKING=FALSE before launching PiCoPiC, jupyter or other tools to disable locking.

Known Issues

some plasma instabilities with unknown nature found. Required to investigate root cause.

cosmonaut-ok / picopic Goto Github PK

picopic's Introduction