1,249

edits

Jump to navigation
Jump to search
##
Summer of Code - Getting Started
(edit)

###
Revision as of 19:54, 29 March 2020

3,126 bytes added
, 29 March
m
~~ ~~
~~How to get started: learn about Chebfun, browse Octave's bug list for classdef-related bugs.~~
~~ ~~
~~First steps: clone this Chebfun [https://github.com/cbm755/chebfun/tree/octave_dev octave_dev branch]. On that, "f = chebfun(@(x) sin(x), [-2 6])" should work with Octave 4.3.0+ and maybe even with 4.2.1. Check that "f(pi)" and "g = f + 1" work. A good first task would be to study [https://github.com/cbm755/chebfun/commit/e20b0ad2dc89cfe8e50ba461b864eff7d5bbef17 this commit], a workaround for "f.funs{1}" using "temp = f.funs; temp{1}". "2*f" is failing, can you fix it, perhaps with this workaround? Or can you make "f.funs{1}" work by changing something in "@chebfun/subsref.m"?~~

→OCS: Add link to open bugs on Savannah.

The following projects are broadly grouped by category and probable skills required to tackle each. Remember to check [[Projects]] for more ideas if none of these suit you, and your own ideas are always welcome. You can also look at our [[Summer of Code|completed past projects]] for more inspiration.

{{Note|~~these ~~These are suggested projects but you are welcome to propose your own projects provided you find an Octave mentor}}

== Summary table ==

! <br />!! !! !! !! !! !!

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#~~ode15s_~~ode15.7Bi.2Cs.7D_:~~_Matlab_Compatible_DAE_solver ~~_Matlab_Compatible_DAE_solvers | ode15{i,s} : Matlab Compatible DAE solvers]] || Carlo de Falco || Francesco Faccio, Marco Caliari, Jacopo Corno, Sebastian Schöps || Numerical || No || Medium || GSoC 2016

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Improve_logm.2C_sqrtm.2C_funm | Improve logm, sqrtm, funm]] || ? || Marco Caliari, Mudit Sharma || Numerical || [https://github.com/RickOne16/matrix No] || Hard || Independent devs 2016

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Improve_iterative_methods_for_sparse_linear_systems | Improve iterative methods for sparse linear systems]] || Marco Caliari || Carlo de Falco || Numerical || No || Hard || SOCIS 2016

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#EPA_hydrology_software_suite | EPA hydrology software suite]] || [[User:KaKiLa| KaKiLa]] || ? || Octave Forge || Yes || Medium || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#FullSWOF overland flow simulator | FullSWOF overland flow simulator]] || [[User:KaKiLa| KaKiLa]] || ? || Octave Forge || Yes || Medium || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#TISEAN_package | TISEAN: Nonlinear Time Series Analysis]] || [[User:KaKiLa|KaKiLa]] || ? || Octave Forge || [[TISEAN_package | No]] || Medium || GSoC 2015

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Octave_Package_management | Octave Package management]] || Sebastian Schöps || [[User:KaKiLa|KaKiLa]], Carnë Draug, Carlo de Falco || Infrastructure || Yes || Medium || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Symbolic_package | Symbolic package]] || Colin B. Macdonald || Mike Miller, Abhinav Tripathi || Octave Forge || [https://github.com/cbm755/octsympy Octsympy] || Medium || GSoC 2016

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#OCS | OCS package]] || Sebastian Schöps || Sebastian Schöps || Octave Forge, Numerical || Yes || Easy || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Using_Python_within_Octave | ~~Pytave project~~Pythonic package]] || Mike Miller || Colin B. Macdonald, Abhinav Tripathi || Infrastructure || No || Medium || some in GSoC 2016

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#~~Jupyter_Integration ~~JSON_encoding.2Fdecoding | ~~Jupyter integration~~JSON encoding/decoding]] || ~~Mike Miller ~~[[User:Siko1056|Kai T. Ohlhus]] || ~~Colin B. Macdonald, Ankit Raj ~~? || Infrastructure || Yes ~~Yes ~~|| Medium || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#~~Chebfun_in_Octave ~~Jupyter_Notebook_Integration | ~~Chebfun in Octave~~Jupyter Notebook Integration]] || Mike Miller || Colin B. Macdonald ~~|| ~~, [[User:~~KaKiLa~~Siko1056|~~KaKiLa~~Kai T. Ohlhus]]~~, Ankit Raj, needs core-Octave mentor/comentor ~~|| Infrastructure~~, Numerical ~~|| Yes ~~ ~~|| ~~Hard ~~Medium || Never

|-

| [[~~Summer_of_Code_Project_Ideas~~Summer of Code - Getting Started#Chebfun_in_Octave | Chebfun in Octave]] || Colin B. Macdonald || [[User:KaKiLa|KaKiLa]] || Infrastructure, Numerical || Yes || Hard || Never|-| [[Summer of Code - Getting Started#PolarAxes and Plotting Improvements | PolarAxes and Plotting Improvements ]] || ? || Rik || Graphics || Yes || Medium || Never

|}

: Difficult.

* '''Potential mentors'''

: ~~?~~Marco Caliari, Mudit Sharma

=== Improve iterative methods for sparse linear systems ===

GNU Octave currently has the following Krylov subspace methods for sparse linear systems: pcg (spd matrices) and pcr (Hermitian matrices), bicg,

bicgstab, cgs, gmres, and qmr (general matrices). The description of some of them (pcr, qmr) and their error messages are not aligned. Moreover, they have similar blocks of code (input check for instance) which can be written once and for all in common functions. The first step in this project could be a revision and a synchronization of the codes, starting from the project [~~http~~https://~~planet~~socis16octave-improveiterativemethods.~~octave~~blogspot.~~org~~com/~~#tag:blogger.com,1999:blog-1297699247151766814.post-8054019978706480250 ~~SOCIS2016]~~, whose latest patch, still to be included, ~~which is ~~[https://savannah.gnu.org/patch/?9108 here]~~already merged into Octave (cset {{cset|6266e321ef22}}).

In Matlab, some additional methods are available: minres and symmlq (symmetric matrices), bicgstabl (general matrices), lsqr (least

squares). The second step in this project could be the implementation of some of these missing functions.

The ~~reference book is available ~~[https://www-users.cs.umn.edu/~saad/IterMethBook_2ndEd.pdf ~~here~~reference book by Yousef Saad]is available online.

* '''Required skills'''

: Maybe hard the mathematical part, medium the programming part.

* '''Mentor'''

:Marco Caliari, Carlo de Falco

=== Chebfun in Octave ===

[https://www.chebfun.org/ Chebfun] is a mathematics and software project for "numerical computing with functions". Basically it approximates functions to machine precision accuracy (10<sup>-15</sup>) using piecewise Chebyshev polynomial interpolants. Operations on those functions (arithmetic, derivatives, root-finding, etc) are then overloaded and return new interpolating polynomials, which are themselves proxies for the actual solution.

Chebfun makes extensive use of classdef classes, and is one of the largest Free Software projects to do so. Unfortunately it currently only works in Matlab. This project seeks to (1) improve Octave's classdef support and (2) tweak Chebfun to work under Octave, for example, removing undocumented classdef features. The final goal is to have at least basic Chebfun features working on Octave. An additional goal would be making ~~"~~<code>pkg install chebfun.zip~~" ~~</code> work in Octave.

The impact of this project is improving Octave and allowing Chebfun to be used without proprietary software.

How to get started:

* Learn about [https://www.chebfun.org/ Chebfun]

* Browse [https://savannah.gnu.org/bugs/?group=octave Octave's bug list] for "classdef"-related bugs.

* Clone this Chebfun [https://github.com/cbm755/chebfun/tree/octave_dev octave_dev branch].

** On that, <code>f = chebfun(@(x) sin(x), [-2 6])</code> should work with Octave 4.3.0+ and maybe even with 4.2.1. Check that <code>f(pi)</code> and <code>g = f + 1</code> work.

** A good first task would be to study [https://github.com/cbm755/chebfun/commit/e20b0ad2dc89cfe8e50ba461b864eff7d5bbef17 this commit], a workaround for <code>f.funs{1}</code> using <code>temp = f.funs; temp{1}</code>. <code>2*f</code> is failing, can you fix it, perhaps with this workaround? Or can you make <code>f.funs{1}</code> work by changing something in <code>@chebfun/subsref.m</code>?

* '''Required skills'''

* '''Potential mentors'''

: Colin B. Macdonald, [[User:KaKiLa|KaKiLa]], Mike Miller (?), Carnë Draug (?), someone from Chebfun team (?).

== Adding functionality to Forge packages ==

Starting points

* [~~http~~https://www.~~univ-orleans~~idpoisson.fr/~~mapmo/soft/FullSWOF~~fullswof/ The FullSWOF Project].

* [https://arxiv.org/abs/1204.3210 FullSWOF: A software for overland flow simulation]

* [https://bitbucket.org/binello7/fswof2d Initial work on Bitbucket]

=== Symbolic package ===

Octave's [https://github.com/cbm755/octsympy Symbolic package] handles symbolic computing and other CAS tools. The main component of Symbolic is a pure m-file class "@sym" which uses the Python package [https://www.sympy.org SymPy] to do (most of) the actual computations. The package aims to expose the full functionality of SymPy while also providing a high-level of compatibility with the Matlab Symbolic Math Toolbox. The Symbolic package requires communication between Octave and Python. Recently, a GSoC2016 project successfully re-implemented this communication using the new [~~https://bitbucket.org/mtmiller/pytave Pytave tool~~[Pythonic|Pythonic package]].

This project proposes to go further: instead of using ~~Pytave ~~Pythonic only for the communication layer, we'll use it throughout the Symbolic project. For example, we might make "@sym" a subclass of "@pyobject". We also could stop using the "python_cmd" interface and use ~~Pytave ~~Pythonic directly from methods. The main goal was already mentioned: to expose the *full functionality* of SymPy. For example, we would allow OO-style method calls such as "f.diff(x)" instead of "diff(f, x)".

* '''Required skills'''

: OO-programming with m-files, Python, and possibly C/C++ for improving ~~Pytave ~~Pythonic (if needed).

* '''Difficulty'''

: easy/medium

=== OCS ===

[[Ocs package | OCS]] is a circuit simulator for Octave. The objective of this project is to update the code to use modern features of Octave (e.g. classdef), [https://savannah.gnu.org/search/?Search=Search&words=%28ocs%29&type_of_search=bugs&only_group_id=1925&exact=1&max_rows=25#options fix open bugs], increase compatibility with SPICE and improve compatibility with other Octave packages (odepkg, control etc).

* [http://octave.sourceforge.net/ocs/overview.html Package help at source forge.]

== Infrastructure ==

=== JSON encoding/decoding === [https://en.wikipedia.org/wiki/JSON JavaScript Object Notation], in short JSON, is a very common human readable and structured data format. Unfortunately, GNU Octave still lacks of builtin support of that data format. Having JSON support, Octave can improve for example it's web service functions, which often exchange JSON data these days. Another interesting applicatoin is described in another Octave GSoC project, see [[Summer of Code - Getting Started#Jupyter_Integration | Jupyter integration]]. In bug {{bug|53100}} a vivid discussion about proper JSON support took place. As JSON is a highly demanded feature for Octave, there are already several attempts to fill the gap: * [https://github.com/fangq/jsonlab jsonlab] (M-file implementation, probably slow for large JSON files)* [https://github.com/gllmflndn/JSONio JSONio] (C MEX wrapper around [https://github.com/zserge/jsmn jsmn])* [https://github.com/Andy1978/octave-rapidjson octave-rapidjson] (C++ Octave wrapper around [https://rapidjson.org/ RapidJSON])* [https://github.com/apjanke/octave-jsonstuff octave-jsonstuff] (C++ Octave wrapper around [https://rapidjson.org/ RapidJSON]) For different reasons, none of them can be directly merged into Octave core yet. Thus there is still lots of work to do. The goal of this project is to evaluate (and to cherry pick from) the implementations above, to create Matlab compatible [https://www.mathworks.com/help/matlab/ref/jsonencode.html jsonencode] and [https://www.mathworks.com/help/matlab/ref/jsondecode.html jsondecode] functions. This involves proper documentation of the work and unit tests to ensure the correctness of the implementation. * '''Minimum requirements''': Good Octave and C/C++ programming knowledge. Ability to make use of C/C++ libraries.* '''Difficulty''': Medium.* '''Mentor''': [[User:Siko1056|Kai T. Ohlhus]] === Jupyter Notebook Integration ===

[http://jupyter.org Jupyter Notebook] is a web-based worksheet interface for computing. There is a [https://github.com/Calysto/octave_kernel Octave kernel for Jupyter]. This project seeks in first place to improve that kernel to make Octave a first-class experience within the Jupyter Notebook.

In general the [https://nbformat.readthedocs.io/en/latest/ Jupyter Notebook Format] is a plain JSON document. In combination with another Octave GSoC project (see [[Summer of Code - Getting Started#JSON_encoding.2Fdecoding | JSON encoding/decoding]]), a second valuable outcome was that Octave can run (and fill) those Jupyter Notebooks on it's own. This would enable Jupyter Notebook users to evaluate long running Octave Notebooks on a computing server without permanent browser connection, which is [https://github.com/jupyter/notebook/issues/1647 still a pending issue].

* '''Minimum requirements'''

: Good Octave and Python programming knowledge.

* '''Difficulty'''

: Medium.

* '''Mentors'''

: Colin B. Macdonald, Mike Miller, ~~others?~~[[User:Siko1056|Kai T. Ohlhus]]

=== Using Python within Octave ===

[~~https://bitbucket.org/mtmiller/pytave Pytave~~[Pythonic]] allows one to call Python functions and interact with Python objects from within Octave .m file code and from the Octave command line interface. ~~Ideally, Pytave will ~~Pythonic may eventually not be a separate ~~project~~package, but rather a core feature of Octave. This project aims to improve ~~Pytave ~~Pythonic with the goal of ~~merging ~~making the ~~code into the core Octave code base~~package more stable, maintainable, and full-featured. ~~ ~~

Based on a previous summer project related to ~~Pytave~~Pythonic, this work will consist of fast-paced collaborative software development based on tackling the [https://~~bitbucket~~gitlab.~~org~~com/mtmiller/~~pytave~~octave-pythonic/issues~~?status=new&status=open pytave ~~Pythonic issue list]. You would also be expected to participate in software design decisions and discussion, as well as improve documentation, doctests , and unit tests. As an example of the sorts of ~~decision ~~decisions being made, note that Octave indexes from 1 whereas Python typically indexes from 0; in which cases is it appropriate to make this transparent to the user?

* '''Mentors'''

Retrieved from "https://wiki.octave.org/Special:MobileDiff/12616...12828"

- Not logged in
- Talk
- Contributions
- Create account
- Log in