cptndarling
One of the best features of this pipeline is the fact that I can write a large chunk of the document whilst conducting research. I can jot down notes, caveats, schedule TODO items etc. and just tag the bits I want for my own reference as 'noexport', so that they don't make it into the exported tex or pdf document.
Another great feature is that I can embed code for producing tables/figures all within the document and specify whether I want the code or result or both to be included.
Finally, once I am happy I have most of the document the way I want it, I export to tex to deal with the final tweaks to finesse things -- spacing for the entire document, fine-grained adjustment of figure formatting.
This ensures that I have all the information that went into constructing the document and allows me to deal with various venue-specific constraints like submitting a source archive, or pdf as necessary.
Here are some resources for getting started with this:
- https://github.com/vikasrawal/orgpaper/blob/master/orgpapers...
- http://draketo.de/english/emacs/writing-papers-in-org-mode-a...
- http://www.clarkdonley.com/blog/2014-10-26-org-mode-and-writ...
Github link: https://github.com/iffsid/ad.js
Here's a link that explains what AD is and does: http://alexey.radul.name/ideas/2013/introduction-to-automati...
This library performs abstract interpretation via operator overloading using sweet.js macros. It implements simple forward and reverse mode AD.
Right now, the code is at a somewhat stable state; although there is a fair amount of duplication to deal with different speed points in running the setup.
Comments and suggestions welcome!