Papers from the computer science community to read and discuss.

@zeeshanlakhani @newsomc - you absolutely need to review this for spelling, grammar, etc.

!!Con's was taken from PyCon's - and I took !!Con's portion about the Hacker School rules. Please make sure all attributions look legit.

Thanks

Resolves #59.

via http://www.fourmilab.ch/etexts/einstein/specrel/specrel.pdf

Not sure about the right directory to place it in but certainly relevant to dist systems. Attempted to follow the naming scheme. Pondering a PR that has an index for all these papers.

I've added a few papers on the theory of dimensionality reduction. A lot of this work is used to theoretically justify the performance of a variety of machine learning techniques and heuristics such as random forests, approximate least-squares and compressed sensing. The papers included are:

1. "The Fast Johnson-Lindenstrauss Transform" (N. Ailon, B. Chazelle) — Only paper not from arXiv, was in a SIAM Journal and taken directly from B. Chazelle's personal website [0]
2. "A Sparse Johnson-Lindenstrauss Transform" (A. Dasgupta, R. Kumar, T. Sarlós) — arXiv/cs/1004:4240; STOC 2010 proceedings
3. "Towards a unified theory of sparse dimensionality reduction in Euclidean space" (J. Bourgain, S. Dirksen, J. Nelson) — arXiv/cs/1311.2542; Accepted in an AMS Journal but unpublished at the moment

I'd be willing to give a talk on one of these papers (presumably not the last one, though, as it is quite deep and hard to cover in a short talk!)

[0] https://www.cs.princeton.edu/~chazelle/pubs/FJLT-sicomp09.pdf

While these aren't all technically "research papers", they are still amazing, exotic and free educational reading resources i've collected over the last few years, so I'll link them just in case you guys want to display some of them.

Operating Systems: Three Easy Pieces

Communicating Sequential Processes (CSP)

Unixca's historical links archive

Programming UNIX Sockets in C - Frequently Asked Questions

C Craft

Ken Thompson Q&A

The Cathedral and the Bazaar

What every programmer should know about memory

Linux Technology Reference

1024cores

switch

command center

asktog

The Art of Unix Programming

stitz zeager math books

That should be enough for right now. I still have a bunch more, so if you guys want me to post more interesting links just let me know you guys are interested via a reply in this issue's comments.

Idea of state machines is very useful in modeling and thinking about stateful systems. Statecharts, which are extensions of state machines, alleviate some of the problems of state machines, mainly state explosion. They are the basis for SCXML standard by W3C.

http://www.inf.ed.ac.uk/teaching/courses/seoc/2005_2006/resources/statecharts.pdf

I'm not sure which category would it go into though.

• [x] Adds Haskell section
• [x] Includes a great paper culled from a lecture by Simon Peyton Jones titled "Tackling the Awkward Squad: monadic input/output, concurrency, exceptions, and foreign-language calls in Haskell". Published in 2010 by Microsoft Research. This is a very readable and useful paper for anyone learning Haskell and interested in using it beyond academia.

Paper Title:

These are all the Mathematics and Machine Learning papers that were submitted in https://github.com/papers-we-love/papers-we-love/pull/550

Reasons for including paper

Good papers and nice commentaries. I've made changes relative to the PR discussion.

This should be good now, but would appreciate a second pair of eyes before merging.

Peter Alvaro from UC Berkeley will present the paper "Using Reasoning about Knowledge to Analyze Distributed Systems" by Joseph Halpern.

Please offer up questions, comments, or additional information regarding this paper. For this particular meetup we will try a round of moderated discussions. Pre-submit any topics in this thread.

Thursday, August 21, 2014 - RSVP information here - http://www.meetup.com/papers-we-love-too/events/168821902/

The directory affective-computing was renamed to follow naming convention.

Paper Title:

Fast Error-bounded Lossy HPC Data Compression with SZ

Paper Year:

2016

Reasons for including paper

This is the initial version of SZ. SZ is a state-of-the-art lossy compression technique widely used in scientific community.

Paper Title:

Significantly Improving Lossy Compression for Scientific Data Sets Based on Multidimensional Prediction and Error-Controlled Quantization

Paper Year:

2017

Reasons for including paper

This is SZ-1.4.

Paper Title:

Error-Controlled Lossy Compression Optimized for High Compression Ratios of Scientific Datasets

Paper Year:

2018

Reasons for including paper

Paper Title:

Fixed-Rate Compressed Floating-Point Arrays

Paper Year:

2014

Reasons for including paper

ZFP is another leading data compression technique widely adopted in scientific community. Its ability of random access to data blocks makes it a strong tool for compressing and decompressing scientific simulation data.

What

I'm suggesting we adopt a NodeSchool like chapter system, where each chapter gets their own repository in order to help organizers coordinate efforts and cut down on time and resources spent figuring out management.

Why

Help Organizers coordinate

I'm currently looking into how to host a Papers We Love in Boston. It's not difficult to do - find a speaker, find a space, let people know it exists, set up a MeetUp group. However, as I'm working with a coorganizer, we have to figure out a way to talk to each other outside of Twitter direct messages. I'd like to be able to use a GitHub repository to use issues for discussion of how to do this, largely because that would allow a way for community members to vote on papers closer to the source, because it's what I'm used to doing with NodeSchool and other organizations, and because it doesn't use Meetups rather poor tools for discussion.

Cross-pollinate ideas about how to coordinate

It would be good to have a standard way of communicating with coorganizers so that chapters can learn from each other how to better organize, too.

People are already doing this

Some chapters already have their own GitHub repositories, or organizations: papers-we-love-rtp, Munich, Bangalore, Hyderabad. Some individuals set up issues for their talks, too, like these in Chicago, and London. I think keeping this in the PapersWeLove organization would be more useful.

Chapter histories could be maintained

Finally, Boston's PapersWeLove chapter used to exist, and now it doesn't. Without some sort of history - which would have happened if there were closed issues on the GitHub repo, instead of just a deleted Meetup group - I don't know why it failed to continue, what caused the chapter to fall apart, and how those issues might be alleviated (or indeed, not. Maybe Boston isn't a good place for Papers We Love.).

How

Set up a repo to help organizers, like this nodeschool one. Create a team called chapter-organizers, and give members access to edit their new chapter repo, e.g. https://github.com/paperswelove/boston.

The existing Code of Conduct is modeled after CoC's used at conferences within the US. PWL is a loose confederation of meet-ups with an international reach and has slightly different needs. The updated CoC still explicitly states that PWL meetups and online communities should be harassment-free environments, but that it is up to the local organizers of live events to maintain that environment.

Hi,

I'm starting research on distributed computing frameworks for deep learning. I've read some papers like tensorflow and horovod. I'm wondering which category these kind of papers fall into. Shall I put them in "distributed system"?

Thanks, JW

When talkign about #559 I realized that there were other folders that have maybe gotten wooly.

Particularly, these concepts are similar enough, and contain overlapping papers:

• https://github.com/papers-we-love/papers-we-love/tree/master/garbage_collection
• https://github.com/papers-we-love/papers-we-love/tree/master/memory_management

I wanted to solicit thoughts before I PR'd it.

Perhaps memory_management can hold a GC group of papers?

Problem

We have little in the way of machine readable data about the contents of the repo

Potential solution

Device a simple JSON format based on BibTex, but with additional fields, that we can generate and consume

Related to #52 #20 #310

As per this recent GitHub post there is a new option for managing large files in a repo.

• Someone should probably test this first in a fork.
• A list of commands / instructions for use will be needed
• Also, the contributing info will need to be updated.
• If contributing and use is too difficult, then this may not be for us.

I can do this eventually; I just wanted to write out some ideas for how to rapidly create adoption, if this looks like a worthwhile thing.

Not to be confused with recently published papers, but a list of say, the 10 most recently added papers to this repository, regardless of when they were published. I visit the website occasionally, but don't really have a good idea as to which papers were recently found.