Clark School Home UMD

December 2016

Symmetry and reduction in collectives: low-dimensional cyclic pursuit

Kevin S. Galloway, Eric W. Justh, P. S. Krishnaprasad

Image: Planar precession of three particles (Figure 7a of the paper; screenshot from animation created by K. S. Galloway; copyright Royal Society)

“Symmetry and reduction in collectives: low dimensional cyclic pursuit,” published October 26 in the Proceedings of the Royal Society A, investigates the types and stability of motion patterns in controlled small collectives. The journal website includes supplementary material in the form of animations of small collectives. The paper was written by U.S. Naval Academy Assistant Professor Kevin Galloway (ECE Ph.D. 2011), Naval Research Laboratory Electronics Engineer Eric Justh (ECE Ph.D. 1998), and P. S. Krishnaprasad.

They use an abstraction of agents in a collective as self-steering particles interacting according to a strategy known as constant bearing cyclic pursuit, to develop an understanding of spatial motion patterns. In the case of collectives with two or three agents, they predict the presence of periodic orbits in phase space and precession in physical space.

The mathematics underlying the current paper has modern control-theoretic aspects as well as classical roots. In recognizing the presence of time-reversal symmetry in the problem of small collectives, the authors use a theorem of G. D. Birkhoff from his studies in 1915, on three body problems of Newtonian particles bound by gravitation. It is rewarding that one can build intellectual bridges between problems arising from widely separated concerns.

There are other threads to explore, such as how small collectives are embedded as network motifs in larger multi-agent systems, and whether this would lead to insights into the behavior of complex interacting systems in nature and in the engineered world. You can read a story about this paper here.

Proceedings of the Royal Society A, 26 October 2016. DOI: 10.1098/rspa.2016.0465  A PDF of the article can be downloaded here.


Anomalously high alumina atomic layer deposition growth per cycle during trimethylaluminum under-dosing conditions

Hossein Salami, Andrew Poissant and Raymond Adomaitis

Image: Thickness map for two representative films both deposited with 500 ALD cycles under TMA saturating (a) and depletion condition (b) corresponding to samples 59 and 60 on Fig. 5. Precursor gas flows from the top to bottom over the illustrated wafer maps. (Fig. 3 from the paper)

The paper “Anomalously high alumina atomic layer deposition growth per cycle during trimethylaluminum under-dosing conditions” by Hossein Salami, Andrew Poissant and Raymond A. Adomaitis has made the “Most Read List for November 2016” of JVST A. The authors report on unexpectedly high atomic layer deposition (ALD) growth per cycle of spatially uniform alumina films during underdosing of the aluminum precursor – precisely the opposite of what one would expect. They postulate a potential deposition mechanism explaining this unusual growth. Follow-up work on an alternative reactor-specific hypothesis explaining this phenomenon is underway.

J. Vac. Sci. Technol. A 35, 01B101 (2017);   A PDF of the article can be downloaded here.


Google DeepMind’s AlphaGo

Hyeong Soo Chang, Michael Fu, Jiaqiao Hu, Steven Marcus

Image: AlphaGo’s two-deep neural networks (Figure 1 from the article)

The University of Maryland researchers who first explored ideas at the core of Google’s AlphaGo wrote an article for the October issue of OR/MS Today, the main magazine of the professional society INFORMS. AlphaGo is the AI system that defeated Go master Lee Se-Dol in early March. In the article, former ISR postdoc Hyeong Soo Chang, now a professor at Sogang University in Korea; Michael Fu; Jiaqiao Hu, (ECE Ph.D. 2006), now an associate professor at Stony Brook University; and Steve Marcus talk about how their ideas influenced the development of AlphaGo.

They also delve into disconnects between the operations research (OR) and computer science/artificial intelligence (CS/AI) communities that hinder idea sharing between the disciplines. The authors then suggest ways these communities could interact in a mutually beneficial way to build on the research strengths of both. You can read their article here. You can also read our story about the original research here.

OR/MS Today, Oct. 2016; Vol. 43, No. 5.  A PDF of the article can be downloaded here.


Evidence of degraded representation of speech in noise, in the aging midbrain and cortex

Alessandro Presacco, Jonathan Z. Simon, Samira Anderson

Image: Audiogram of the grand averages of both ears of younger (gray) and older (black) adults. All participants have clinically normal hearing. HL, hearing level. (Fig. 1 from the paper)

In an interdisciplinary study published in the Journal of Neurophysiology, Samira Anderson, Jonathan Simon and their former student Alessandro Presacco (NACS Ph.D. 2016), currently a postdoctoral researcher at the University of California, Irvine, found that adults aged 61–73 with normal hearing scored significantly worse on speech understanding in noisy environments than adults aged 18–30 with normal hearing. “Evidence of degraded representation of speech in noise, in the aging midbrain and cortex” is part of ongoing research into the so-called cocktail party problem, or the brain’s ability to focus on and process a particular stream of speech in the middle of a noisy environment. The researchers were interviewed by The Guardian newspaper and BBC4 radio. You can read the ISR story about the research here.

Journal of Neurophysiology, 1 November 2016 Vol. 116 no. 5, 2346-2355 DOI: 10.1152/jn.00372.2016  A PDF of the article can be downloaded here.


Probabilistic information transmission in a network of coupled oscillators reveals speed-accuracy trade-off in responding to threats

Amanda Chicoli and Derek A. Paley

Image: (a) Illustration of an agent (depicted here as a fish) at a random orientation at the first time step (t = 0, blue), and transitioning to startled state (t = 1, red). When startled directly from a threat, agents react by instantaneously reorienting towards a reference direction, with some noise (ω 2). Without loss of generality, the reference direction is set to θs = 0. (b) Illustration of two agents. When startled indirectly by social cues, agents move in the average direction of their startled neighbors. In this illustration, only the difference in orientation between two agents is shown, along with noise (ω 1) that is added to the difference in orientation, which represents the noise in the agent's ability to sense their neighbor's orientation. Credit: American Institute of Physics (AIP)

In November, the American Institute of Physics published a press release about this paper on Eureka Alert. The paper was written by Amanda Chicoli (NACS Ph.D. 2016) and her advisor Derek Paley. Amanda is currently a postdoctoral researcher at the Carnegie Institution for Science. The paper models how information spreads in a group using a susceptible-infected-removed epidemic model. This model is applied to a simulated shoal of fish using the motion dynamics of a coupled oscillator model, to test the biological hypothesis that polarized or aligned shoaling leads to faster and more accurate escape responses. The researchers found that when perfectly aligned individuals in a group are startled, there is a rapid escape by individuals that directly detect the threat, as well as by individuals responding to their neighbors. However, individuals that are not startled do not head away from the threat. In startled groups that are randomly oriented, there is a rapid, accurate response by individuals that directly detect the threat, followed by less accurate responses by individuals responding to neighbor cues. Over the simulation duration, however, even unstartled individuals head away from the threat. This study illustrates a potential speed-accuracy trade-off in the startle response of animal groups, in agreement with several previous experimental studies. Additionally, the model can be applied to a variety of group decision-making processes, including those involving higher-dimensional motion.

American Institute of Physics Chaos: An Interdisciplinary Journal of Nonlinear Science,  A PDF of the article can be downloaded here.


The Generalized Regenerator Location Problem

Si Chen, Ivana Ljubić, S. Raghavan

Image: The RESTRICTED_BFS ((a)–(c)) and PLACE_REGENS ((d)–(f)) Procedure (Fig. 6 of the paper)

“The Generalized Regenerator Location Problem,” a paper by S. Raghavan; his former student Si Chen (BMGT Ph.D. 2007), currently at the Federal Reserve; and Ivana Ljubić of the Essec Business School in Paris was selected for the 2016 INFORMS Telecommunications Section Prize Best Paper Award (for the most outstanding paper applying operation research techniques in the context of telecommunications). It was published in the INFORMS Journal on Computing.

The paper deals with a problem in optical networks. In an optical network a signal can only travel a maximum distance dmax before its quality deteriorates to the point that it must be regenerated by installing regenerators at nodes of the network. As the cost of a regenerator is high, the goal is to deploy as few regenerators as possible in the network, while ensuring all nodes can communicate with each other.

The authors significantly expand the scope of the definition of the problem so that any regenerator location problem can be modeled as the generalized regenerator location problem. They establish a correspondence between the (node-weighted) directed Steiner forest problem and the GRLP. Using this fact, they provide several ways to derive natural and extended integer programming (IP) and mixed-integer programming (MIP) models for the GRLP and compare the strength of these models. They also develop a branch-and-cut approach to solve the problem to optimality. You can learn more about the paper and the award here.

INFORMS Journal on Computing, Vol. 27, Issue 2.  A PDF of the article can be downloaded here.