Top all-time · #13963
Andrew Tulloch
Personal blog · tullo.ch/articles · on HN as ajtullochprogrammingdatamath
Andrew Tulloch's personal blog at tullo.ch covers programming, data and math. Since 2012, 5 different Hacker News readers have linked it 20 times, most often “Python vs Julia – an example from machine learning”. No feed was found, so read it on the site itself.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
Commenters cite Andrew Tulloch's posts on Python versus Julia performance and a modern Emacs setup, and a Tufte-inspired book template. The comments do not discuss the writing itself.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Python vs Julia – an example from machine learning3 people · 3 mentions · first linked 2014-03-13
- A modern Emacs setup in OS X2 people · 2 mentions · first linked 2014-05-16
What Hacker News says?
> is just plain dog shit slow for certain things compared to SciPy Do you have real examples here or is this just FUD? SciPy is not known for using absolute state of the art algorithms or perfectly optimized implementations. It can be pretty easy to improve on the naively implemented or legacy pieces of SciPy, e.g. http://tullo.ch/articles/python-vs-julia/
Comment linking /articles/python-vs-julia/
on “Grumpy: Go running Python”Submitted Python vs Julia – an example from machine learning as a story · 170 points
HN discussionIndeed. For example: http://tullo.ch/articles/modern-emacs-setup/
Comment linking /articles/modern-emacs-setup/
on “Atom 1.0”Submitted A parallel Boggle solver in Haskell as a story · 6 points
HN discussionIf you're interested in this material, you may also like my LaTeX'ed lecture notes covering the last few years of my mathematics degree - mostly pure mathematics with some statistics and financial mathematics. http://tullo.ch/2011/mathematics-lecture-notes/ for the PDFs, and https://github.com/ajtulloch/SydneyUniversityMathematicsNote... for the LaTeX source.
Comment linking /2011/mathematics-lecture-notes/
on “Math notes to take you from one year of college calculus to grad student level”Submitted A modern Emacs setup in OS X as a story · 2 points
HN discussionI'm not averse to C++11 - I personally think it's a great language for certain tasks, and have used it extensively for performance-critical ML - see [1] for some public work. The value proposition of Julia (and Python with Cython) is that you don't have to choose between the productivity you have from a dynamic language (IPython/IJulia [2] are incredibly productive environments for ML/data science IMO), and you have…
Comment linking /articles/decision-tree-evaluation/
on “Python vs Julia – an example from machine learning”Submitted Python vs Julia – an example from machine learning as a story · 2 points
HN discussionHi, author here. Yeah, the active set implementation in Cython is relatively slow - see [1] for where I discussed how I benchmarked the existing algorithm and demonstrated the PAVA implementation was faster. Then again, until two days ago the Cython active set algorithm was the production implementation of isotonic regression in scikit-learn, so it's not like it was a strawman example that I made up. [1]: http://tul…
Comment linking /articles/speeding-up-isotonic-regression/
on “Python vs Julia – an example from machine learning”Submitted Golfing Stripe CTF 3 as a story · 2 points
HN discussionI wrote http://tullo.ch/articles/python-vs-julia/ a few weeks ago, which compared Cythonized implementations of particular ML algorithms in Python against Julia implementations. There were some useful discussions on HN at https://news.ycombinator.com/item?id=7383121
Comment linking /articles/python-vs-julia/
on “Julia vs. Python: Monte Carlo Simulations of Bitcoin Options”Submitted The performance of decision tree evaluation strategies as a story · 1 points
HN discussionScratched an itch and built LaTeX2Markdown ( http://tullo.ch/projects/LaTeX2Markdown ) - a LaTeX to Markdown converter designed for converting AMS-LaTeX documents to webpages. Offers compatibility with AMS-LaTeX (theorems, lemmas, proofs, etc), and drop-in integration with a Markdown/MathJaX blogging environment. Pandoc can't cope with these environments, so it seemed useful to build something that did. The usage mo…
Comment linking /projects/LaTeX2Markdown
on “Ask HN: What did you build in March?”That's actually not the case - that's one effect, sure, but when you compared the compiled trees to the flattened trees (the strategy just described), the compiled trees are substantially faster across a range of parameters. See http://tullo.ch/articles/decision-tree-evaluation/ for a detailed evaluation.
Comment linking /articles/decision-tree-evaluation/
on “The Unreasonable Effectiveness of Random Forests”This is a nice template, but it’s somewhat missing the point IMO. Tufte’s two-page spreads are designed one by one, based on the content. He painstakingly arranges every figure and every word so that the books will be both clear and beautiful. Using LaTeX is to some extent antithetical to this idea. Instead of arranging everything by hand, you’re letting the typesetting engine make most of the decisions. There’s not…
Comment linking /static/cambridge/AdvancedProbability-LectureNotes.pdf
on “Book design inspired by Edward Tufte”I've been a convert to Tufte-LaTeX for a while - it produces beautifully formatted documents. I wrote a post a while ago on setting it (from a vanilla LaTeX setup) at http://ajtulloch.github.io/2012/getting-started-with-tufte-l... , which has some examples. For some heavy mathematics typesetting with Tufte-Latex, have a look at http://tullo.ch/articles/cambridge-mathematics-notes/ (e.g. http://tullo.ch/static/cambri…
Comment linking /static/cambridge/NonParametricStatistics-LectureNotes.pdf
on “Book design inspired by Edward Tufte”For what it's worth, I don't think it's fair to claim that SVM is O(n^3). From [1], LibLinear trains linear SVMs in O(nd log (1/p)) time, where n is the number of examples, d is the number of features, and p is the optimization tolerance, and Pegasos/FOLOS and other subgradient methods can do online kernel SVM training in O(s / λ p) iterations, where s is the number of non-zero features in each example, p is as abov…
Comment linking /articles/speeding-up-decision-tree-training/
on “An Introduction to Decision Trees with Julia”Looks great! Nice scores. I focused on getting the minimal passing solutions (e.g. Level 3, the distributed grep problem, only took ~5 LoC changes to pass, Level 4 was 162 LoC additions, 69 LoC deletions). I wrote up my solutions at http://tullo.ch/articles/stripe-ctf-golfing/ if anyone was interested.
Comment linking /articles/stripe-ctf-golfing/
on “Stripe-CTF 3 Writeup”A few months ago I learned d3.js and made myself an interactive site logo for myself with it. It's an awesome library. Check out the logo (try clicking it and dragging the letters) and tutorial at http://tullo.ch/2012/making-the-site-logo/ .
Comment linking /2012/making-the-site-logo/
on “Build An Animated Chart In 19 Lines Of Code With d3.js”
