Monday, April 01, 2013

MoAR: good news, some statistical, and perhaps an end is near..?

it's pretty evident that i spend a lot of time on the internet .. at least, when not at work, and i wonder to what extent it's affecting my life [0]. sure, i pick up news and interesting facts, but as information goes that's not terribly important in my life, so it's probably no better for me than watching television.

my guess is that if you're doing research at a university, ph.d. student or postdoc or faculty, then you already know how to waste time effectively and can therefore already find the same article links that i do. there exist plenty of news feeds that now many the search easy, after all .. so that's why i'm thinking about ending these Monday Article Roundups.

[ to skip ahead to why i might end MoAR, click here. ]

at any rate, here's what i found interesting to read, this week ..


where i learn that not everyone uses the arXiv ..

for me, it's easy to forget: we mathematicians only need paper and pens, a computer, and preferably a chalkboard of some kind. in that sense, our work is cheap to sponsor.

as for other fields, especially laboratory sciences, the cost of running operations is orders-of-magnitude higher [1] ..!
"These approaches suit communities that have a culture of sharing preprints, and that either produce theoretical work or see high scrutiny of their experimental work — so it is effectively peer reviewed before it even gets submitted to a publisher. But they find less support elsewhere — in the highly competitive biomedical fields, for instance, researchers tend not to publish preprints for fear of being scooped and they place more value on formal (journal-based) peer review. “If we have learned anything in the open-access movement, it's that not all scientific communities are created the same: one size doesn't fit all,” says Joseph."

~ from "Open access: The true cost of science publishing" @nature


on why i shouldn't be a statistician ..

the real world just doesn't make any sense to me. the scale of how society works is often too .. overwhelming. [2].

on the other hand, sometimes statistical evidence is non-intuitive, in good ways:
"Since launching in Kenya, GiveDirectly continues to evaluate its approach with randomized control trials. They use a lottery system similar to medical trials and compare developmental outcomes of households who have received funding against those who haven't. Their rigorous data shows that no-strings-attached cash transfers improve health and downstream financial gains. They also use this data to refine their model, and make it available on their website. Recipients, who are often living on less than 65 cents a day, invest in everything from food for starving children to long-term assets, including land, livestock and housing. The data fights conventional wisdom: [m]oney spent on alcohol and cigarettes either decreases, stays constant or increases in the same proportion as total other expenses (approximately 2% to 3%)."

~ from "Want to Help People? Just Give Them Money" @hbr
this is pleasantly undepressing (if not good) news. maybe, on average, people aren't so untrustworthy?

as for this bit of news, i don't know if it amounts to an actual medical phenomenon or just corporate spin .. but it's certainly an interesting explanation.
"In December 2011, shares of BioSante Pharmaceutical Inc. slid 77% in a single session after the company's experimental gel for promoting libido in postmenopausal women failed to perform well against placebo in late-stage trials. The drug companies say these failures are happening not because their drugs are ineffective, but because placebos have recently become more effective in clinical trials."

~ from "New Patents Aim to Reduce Placebo Effect" @assertTrue()
.. and as for something more related to maths:
"“Our study shows that it’s not lack of ability or differences in ability that orients females to pursue non-STEM careers, it’s the greater likelihood that females with high math ability also have high verbal ability,” notes Wang. “Because they’re good at both, they can consider a wide range of occupations.”

Notably, those participants who reported feeling more able and successful at math were more likely to end up in a STEM-related job, and this was particularly true for students who had high math and moderate verbal abilities. Thus, math may play a more integral role in these individuals’ sense of identity, drawing them toward STEM occupations.
"

~ from "More Career Options May Explain Why Fewer Women Pursue Jobs in Science and Math" @psychsci


maths as creative commons?

i doubt that the following bit of news will change very much of the status quo .. for example, i'm sure that the patent for PageRank will stay intact. maybe this is a sign of hope .. that the future will be different?
"Chief Judge Leonard Davis based the ruling on U.S. Supreme Court case law that prohibits the patenting of mathematical algorithms. According to Rackspace, this is the first reported instance in which the Eastern District of Texas has granted an early motion to dismiss finding a patent invalid because it claimed unpatentable subject matter."

~ from "Judge Says Mathematical Algorithms Can’t Be Patented" @techcrunch


lastly: maybe the end of MoAR?

my original motivation for these roundups was to reduce the number of "here's a cool article" posts on this blog. it's come to my attention, however, that there are plenty of news feeds out there with lots of good maths new updates.

so though these other feeds don't exactly interfere with my own interests .. i can't help but feel like these roundups are a bit .. redundant. here are a few items from the newsfeeds at scidaily and sciam, to show you what i mean.
"Fermat's Last Theorem is just about numbers, so it seems like we ought to be able to prove it by just talking about numbers," McLarty said. "I believe that can be done, but it will require many new insights into numbers. It will be very hard. Harvey sees my work as a preliminary step to that, and I agree it is."

~ from "Fermat's Last Theorem and More Can Be Proved More Simply" @scidaily
you see what i mean? these are articles that are actually about maths .. which i find to be a breath of fresh air. most other newsfeeds require quite a bit of searching, in between technology/business news and a flood of physics and biology announcements.

as for something that actually has an even more (mathematically) technical flavor ..
"The MIT researchers' approach is much more straightforward. The first thing they do is find a "spanning tree" for the graph. A tree is a particular kind of graph that has no closed loops. A family tree is a familiar example; there, a loop might mean that someone was both parent and sibling to the same person. A spanning tree of a graph is a tree that touches all of the graph's nodes but dispenses with the edges that create loops. Efficient algorithms for constructing spanning trees are well established.

The spanning tree in hand, the MIT algorithm then adds back just one of the missing edges, creating a loop. A loop means that two nodes are connected by two different paths; on the circuit analogy, the voltage would have to be the same across both paths. So the algorithm sticks in values for current flow that balance the loop. Then it adds back another missing edge and rebalances.

In even a simple graph, values that balance one loop could imbalance another one. But the MIT researchers showed that, remarkably, this simple, repetitive process of adding edges and rebalancing will converge on the solution of the graph Laplacian. Nor did the demonstration of that convergence require sophisticated mathematics: "Once you find the right way of thinking about the problem, everything just falls into place," Kelner explains.
"

~ from "Short Algorithm, Long-Range Consequences" @scidaily
an awesome exposition: somehow, in colloquial terms, this writer can give the rough idea of what graph laplacians are .. and without passing too much to the classical theory of PDEs and numerical solutions, brownian motion, or convergence of spaces.

another cool thing is that this algo exploits the intrinsic geometric properties that graphs have. to explain, most of the time i see graph laplacians, they are meant as simulations for PDEs and the graph approximates a space with some (well-defined) notion of area or volume. here, the 1-dimensionality of graphs is used to great effect: a graph laplacian can be viewed simultaneously as an energy minimiser that approximates the laplacian on a surface, as well as a network flow like that of an electric circuit that obeys kirchhoff's laws. (i suppose that one could see the same thing using divergence or curl for level curves on a surface, but for me it's a less obvious feature.)

this is not to say that physics isn't cool, though ..
"Creating a knot in a fluid bears little resemblance to tying a knot in a shoelace, say Dustin Kleckner and William Irvine, physicists at the University of Chicago in Illinois. The entire three-dimensional (3D) volume of a fluid within a confined region, such as a vortex, must be twisted. Kleckner and Irvine have now created a knotted vortex using a miniature version of an airplane wing built with a 3D printer."


~ from "Physicists Twist Water into Knots" @sciam
i haven't come to a decision about whether to continue MoAR or not; suggestions and comments about it are welcome. (at any rate, there will be another roundup next week.)


[0] .. and yes, i mean affect, not effect, since the meaning suggests an ambiguous or unrealised causality. (in other words, my internet addiction can only "effect" an outcome in my life if i know what the outcome actually is. until then, it only affects my life.) call me a dinosaur .. and i understand that languages evolve all the time .. but i hate it when people misuse "effect" as a verb, which is rather often.

[1] my sister, who is a bio postdoc, once told me that her lab bought a second several thousand dollar automated espresso machine. when i asked how much time they spent thinking about the cost and saving up for it, she shrugged and said, compared to the monthly lab budget, that it was essentially nothing.

[2] to be honest, orders of infinity are easier to work with than very large but finite numbers. i suppose it has to do with the fact that "infinity" is a kind of functional black box .. in that one doesn't actually treat it as a number, but as a property of a given set .. whereas my imagination just isn't that good for explicit numbers that don't fit this reductionism.

No comments: