Jacopo.bertolotti

Science prizes: Time-lapsed awards for excellence

Nature 500, 7460 (2013). doi:10.1038/500029c

Authors: Ho Fai Chan & Benno Torgler

The number of scientific prizes has proliferated in the past 20 years (see Nature498, 152–154; 201310.1038/498152a). But once a scientist has published a seminal contribution, how long is it before these glittering symbols of recognition come through?Occasionally,

Nature Photonics 7, 639 (2013). doi:10.1038/nphoton.2013.171

Authors: Christopher Barsi & Jason W. Fleischer

Nature Photonics 7, 579 (2013). doi:10.1038/nphoton.2013.185

Authors: Dirk Jalas, Alexander Petrov, Manfred Eich, Wolfgang Freude, Shanhui Fan, Zongfu Yu, Roel Baets, Miloš Popović, Andrea Melloni, John D. Joannopoulos, Mathias Vanwolleghem, Christopher R. Doerr & Hagen Renner

The quest for on-chip optical isolators has recently spawned many new isolator structures. However, there has been some confusion about the requirement of nonreciprocity. Here, we review the essential characteristics of an isolator.

Nature Photonics 7, 577 (2013). doi:10.1038/nphoton.2013.212

Are inaccurate or misleading descriptions problems in photonics? Community feedback indicates that there is room for improvement in several areas.

Nature Methods 10, 687 (2013). doi:10.1038/nmeth.2571

Authors: Martin Krzywinski & Alberto Cairo

Relate your data to the world around them using the age-old custom of telling a story.

Piled Higher & Deeper by Jorge Cham		www.phdcomics.com
title: "Confession" - originally published 7/29/2013 For the latest news in PHD Comics, CLICK HERE!

A Harvard Gazette article centers on the eternal question “does correlation imply causation?” The article has this headline and opening paragraph:

Coffee drinking tied to lower risk of suicide

Drinking several cups of coffee daily appears to reduce the risk of suicide in men and women by about 50 percent, according to a new study by researchers at the Harvard School of Public Health (HSPH). The study was published online July 2 in The World Journal of Biological Psychiatry….

This is one of the few reports to explicitly state that coffee seems to cause (“Drinking several cups of coffee daily appears to reduce”) a life-or-death effect in people who drink it. In so doing, this report may be a watershed (and/or coffeeshed) in the history of biomedical science, and in the field of biological psychiatry.

BONUS: The report also says, a few paragraphs later, “In spite of the findings, the authors do not recommend that depressed adults increase caffeine consumption…”

(Thanks to investigator Corky White for bringing this to our attention.)

BONUS: (unrelated): The Stephen King of suicidology is now president

BONUS (unrelated): More from Professor Lester

Article

Crackling noise is commonly observed in various physical systems. Schröder et al. demonstrate that crackling noise can be attributed to the concept of fractional percolation, which is found to be applicable to the known Barkhausen effect in ferromagnets.

Nature Communications doi: 10.1038/ncomms3222

Authors: Malte Schröder, S. H. Ebrahimnazhad Rahbari, Jan Nagler

POW!

Author(s): David A. Huse, Rahul Nandkishore, Vadim Oganesyan, Arijeet Pal, and S. L. Sondhi

Closed quantum systems with quenched randomness exhibit many-body localized regimes wherein they do not equilibrate, even though prepared with macroscopic amounts of energy above their ground states. We show that such localized systems can order, in that individual many-body eigenstates can break sy...

[Phys. Rev. B 88, 014206] Published Mon Jul 22, 2013

When the Leap Motion controller was revealed to the world, it brought with it the promise of a new and unique computer user experience. And, ever since we first got to see what the Leap Motion controller could do -- grant folks the ability to interact with a computer by waving their fingers and fists -- we've wanted one of our own to test out. Well, our wish was granted: we've gotten to spend several days with the controller and a suite of apps built to work with it. Does the device really usher in a new age of computing? Is it worth $80 of your hard-earned cash? Patience, dear reader, all will be revealed in our review.

Filed under: Peripherals

Comments

Piled Higher & Deeper by Jorge Cham		www.phdcomics.com
title: "Zeno's Paradoc" - originally published 7/19/2013 For the latest news in PHD Comics, CLICK HERE!

A. D. Boozer
We clarify several issues involving the concepts of time-reversal invariance and time asymmetry in classical electrodynamics. Specifically, we consider three questions: (I) If electrodynamics is time-reversal invariant, why are the radiative processes that occur in nature time asymmetric? (II) Why ... [Am. J. Phys. 81, 585 (2013)] published Thu Aug 01, 2013.

J. Reintjes and Mark Bashkansky
We present an analysis of the transmission of a weak optical pulse, with energy 1ℏω0, through a dense absorbing medium with absorption frequency ω0. We analyze the system by treating the optical pulse classically and the absorbing medium quantum mechanically. We find that the probabilistic back re ... [Am. J. Phys. 81, 610 (2013)] published Thu Aug 01, 2013.

Certain questions keep returning, as if on the edge of a rotating wheel:

“Why the Wheels Won’t Go,” Michael LaBarbera [pictured here, without wheels], American Naturalist, Vol. 121, No. 3, March 1983, pp. 395-408. The author, at the University of Chicago, begins:

“Why don’t animals have wheels? Introductory hiology teachers commonly note the lack of rotating structures in biological systems, usually as a starting point to illustrate the restrictions that structure and physiology place on the forms which may arise via natural selection. The question of why animals do not have wheels is part of the professional folklore of hiology; while rarely addressed in the formal scientific literature, every biologist is familiar with the question and has a favorite set of explanations…”

BONUS: Wikipedia takes the question for a spin

Author(s): M. Treiber, P. M. Ostrovsky, O. M. Yevtushenko, J. von Delft, and I. V. Lerner

We study the quantum corrections to the polarizability of isolated metallic mesoscopic systems using the loop expansion in diffusive propagators. We show that the difference between connected (grand-canonical ensemble) and isolated (canonical ensemble) systems appears only in subleading terms of the...

[Phys. Rev. B 88, 024201] Published Fri Jul 19, 2013

Author(s): Dibyendu Mandal, H. T. Quan, and Christopher Jarzynski

We describe a simple and solvable model of a device that—like the “neat-fingered being” in Maxwell’s famous thought experiment—transfers energy from a cold system to a hot system by rectifying thermal fluctuations. In order to accomplish this task, our device requires a memory register to which it c...

[Phys. Rev. Lett. 111, 030602] Published Fri Jul 19, 2013

thedailywhat:

Time Lapse Study of the Day: 69-Year-Old Pitch Drop Experiment Comes to Fruition

In October 1944, physicists at Trinity College in Dublin began an experiment to measure the flow of tar pitch, a polymeric substance that appears solid at room temperature. Fast forward 69 years to last week, the drop was finally captured on camera for the first time ever.

erano anni che stavo aspettando una cosa del genere (sul serio) (anche: questo spiega molte cose)

Spain's flagship research agency, the Spanish National Research Council, is in deep trouble.

Author: Elisabeth Pain

Vaibhav Gaind, Dergan Lin, Kevin J. Webb
We present a formalism for solving the scalar Bethe–Salpeter equation (BSE) in the nondiffusive regime under the ladder approximation and for an infinite randomly scattering medium having scatterers of size on the order of or larger than the wavelength. We compare the information content in ... [J. Opt. Soc. Am. B 30, 2199-2205 (2013)]

Digital learning: Look, then leap

Nature 499, 7458 (2013). doi:10.1038/499275a

Author: Michael M. Crow

Massive open online courses can make higher education more accessible, immersive and comprehensive — if they are deployed with due caution, says Michael M. Crow.

Online learning: How to make a MOOC

Nature 499, 7458 (2013). doi:10.1038/nj7458-369a

Author: Sarah Kellogg

With forethought and support, science instructors can design effective massive open online courses.

Education online: The virtual lab

Nature 499, 7458 (2013). http://www.nature.com/doifinder/10.1038/499268a

Author: M. Mitchell Waldrop

Confronted with the explosive popularity of online learning, researchers are seeking new ways to teach the practical skills of science.

An analysis of the most highly contested articles on Wikipedia reveals the controversies that appear invariant across languages and cultures

Author(s): Georg Heinze, Christian Hubrich, and Thomas Halfmann

The maximal storage duration is an important benchmark for memories. In quantized media, storage times are typically limited due to stochastic interactions with the environment. Also, optical memories based on electromagnetically induced transparency (EIT) suffer strongly from such decoherent effect...

[Phys. Rev. Lett. 111, 033601] Published Mon Jul 15, 2013

Or: How I Learned to Watch the Best Movies in the Best Way

I remember when I lived across the street from an art movie theater called Le Club, looking at the movie posters on my way back home was often enough to get me in the ticket line. The director or main actors would ring a bell, or a close friend had recommended the title. Sometimes the poster alone would be appealing enough to lure me in. Even today there are still occasions when I make decisions from limited visual information, like when flipping through movie kiosks, TV guides, or a stack of DVDs written in languages I can’t read.

So how can Mathematica help? We’ll take a look at the top 250 movies rated on IMDb. Based on their posters and genres, how can one create a program that suggests which movies to see? What is the best way to see the most popular movies in sequence?

Drawing from methods used in content-based image retrieval, I will first compute visual similarities between movie posters and display them in a graph. The graph will then be enriched with semantic non-visual information like genres, which should then yield some decent viewing recommendations.

To gather the initial dataset, I looked for lists of great movies. There are many of them, including compilations by The New York Times and by the late Roger Ebert. I felt comfortable opting for the top 250 movies rated by users from IMDb, and I started by copying and pasting the table from the site into a Mathematica notebook.

IMDb is constantly updating the content of this top-250 movies table based on new ratings from users; you should expect the current list to differ from the snapshot I took, which happened to be on June 27, 2013.

The function ImportString parsed the pasted input seamlessly and produced a list of titles. I obtained the remaining data from Wolfram|Alpha by programmatically querying for the corresponding posters and genres:

A handful of obviously incorrect images and genres came back, perhaps because the query was ambiguous for some movie titles. I did correct the data by manually massaging those specific queries. The movies without posters in Wolfram|Alpha have been disregarded. After such curation, a list of 240 movies was left:

The first step was to compute similarities between the movie posters. More precisely, I computed a visual dissimilarity for each pair of images. The function ImageDistance features 16 ways to do so. Since movie posters often have a distinct color theme, I selected the EarthMoverDistance method that estimates the dissimilarities between the color histograms of two images. A coffee break later, I had a symmetric matrix containing the visual distances between all pairs of images:

The matrix appears with dark spots corresponding to strong visual similarities between movies:

Distances between images follow a distribution defined on the domain (0, 1), which is unimodal and has positive skewness:

The distances between images provide information about the similarities of movie posters. This information can be expressed in a graph where nodes correspond to movies and edges correspond to strong similarity.

Let’s assume that only the smallest distances constitute edges. With a threshold set so that only 5% of the distances are selected, not allowing self-loops, the movies’ visual similarity graph is:

At this rendering scale, we don’t see much more beyond common sense: a large group of interconnected dark posters as well as a second group of light posters that is connected to the dark posters group through saturated colorful poster offshoots. A few movies are isolated, among which is the top-rated movie, The Shawshank Redemption.

While you could use this graph to select a sequence in which to watch the movies, relying only on visual information wouldn’t be very meaningful. For the next step, I will augment the visual graph with semantic information gathered from the genres.

To do so, I introduce new vertices in the graph, one for each movie genre, as well as edges connecting each movie poster with its associated genres. In other words, I combine two graphs—the visual graph computed from the image distances and the semantic bipartite graph where each genre is connected to a list of movies. To compute the visual weighted adjacency matrix, I simply use either the EarthMoverDistance or zero, if the distance is too high, as weights. I also normalize each row of the matrix so that it sums to 1. This allows us to think of the entries in the matrix as probabilities of going from one vertex to the other:

I do not put weight in the semantic adjacency matrix, since we don’t know much about the reliability of the genres:

Finally, the visual and semantic adjacency matrix is a linear combination of the two matrices. Here, I use the average. I obtain the corresponding graph by calling the function WeightedAdjacencyGraph, making sure to indicate an absence of edge using ∞ instead of 0:

I can use this graph to compute paths between movies. For example, the shortest path between Toy Story 3 and Citizen Kane occurs by watching Toy Story 3, looking for another comedy, then watching Roman Holiday, and finally Citizen Kane:

The function GraphDistanceMatrix allows me to discover the most dissimilar pairs of movies, and it turns out there are two of them, both including Beauty and the Beast:

Another application is to help moviegoers select their next movie. Based on what they just saw, how can I suggest a few relevant movies? I’ll simply select the five most similar vertices in the graph, implicitly assuming that the moviegoers like any movie they see. For example, to someone who just watched Life of Pi, I’d recommend Toy Story 3, The Killing, The Shawshank Redemption, or browsing the lists of adventures and dramas:

Going further, I can actually model the behavior of a movie buff as he or she sees movies in sequence. I must assume that all the movies from the list are available to my movie buff—many are not lucky enough to have such resources at their disposal.

The process of navigating such a weighted graph based on the probabilities of transitioning from one vertex to the other is called a random walk. Instead of just using the visual and semantic graph’s adjacency matrix as a transition matrix, I’m going to modify it so that it can model boredom. Sooner or later, my movie buff will get tired of watching similar movies in sequence, and looking for fresh air, will switch randomly to another movie. To take into account such behavior, I use as transition matrix a linear combination of the graph’s adjacency matrix and a constant matrix of equiprobable transitions—with the exception that the diagonal entries in this constant matrix are set to 0, because no bored viewer is going to see the same movie again right away:

I set the combination factor α to 0.85 because it seems to be a common setting in such systems. The behavior of my movie buff will be modeled thanks to the functions DiscreteMarkovProcess, which represents a random walk process, and RandomFunction, which allows simulation of such processes. For example, here are 10 possible sequences after watching 3 Idiots:

As a final application, what is the best way to see all the movies? We are looking for the optimal sequence that would visit all the nodes corresponding to movies exactly once and the nodes corresponding to genres an arbitrary number of times. This is essentially a modification of the traveling salesman problem. There is no immediate solution for this modified problem in Mathematica. Instead, I am going to start from the visual graph and fill up each of its missing edges. Each pair of movies that is too distinct visually will be assigned a distance estimated by finding the length of the shortest path between the two movies in the visual and semantic graph.

The code below is a bit more complicated than that because I must take care of scaling the visual and semantic distances in a sensible fashion. Edges connecting movies to genres are weighted such that a jump through the semantic graph is given a weight of at least the maximum visual distance.

The optimal sequence is given by the function FindShortestTour:

And here it is, from The Shawshank Redemption to Memento, the optimal sequence in which to see all IMDb’s top 250 rated movies based on visual and semantic similarity:

From an algorithmic perspective, and maybe for the marathon viewer’s pleasure, it would be interesting in future work to include additional constraints so that movie franchises like The Lord of the Rings are seen in the expected order. Download a companion notebook and start your work on these topics right away.

It was very enjoyable experimenting with these data and models. I was also able to discover new application domains through Mathematica‘s documentation, which contains extensive information on the functionality used here.

Nevertheless, looking back in time, such a program is no match compared to the selections made by the art theater across the street or my dear friends’ suggestions.

Download this post as a Computable Document Format (CDF) file.

Read 6 remaining paragraphs | Comments

Piled Higher & Deeper by Jorge Cham		www.phdcomics.com
title: "Everything he can" - originally published 7/10/2013 For the latest news in PHD Comics, CLICK HERE!