C M Y K

SCIENCE KOLKATA WEDNESDAY 6 MARCH 2013 10

Making good connections comes out of the interaction.

Seeking inspiration
from locusts
How groups connect with Peripheral to Any one or more of theses strategies may not jamie lewis reports on how scientists
each other is important in peripheral
Network B be available for a network, either because of are working on preventing car crashes
Network A physical constraints or even behavioral or
the networked world, says Peripheral to cultural impediments. The airline company, in
s ananthanrayanan central the example, for instance, cannot add airport SCIENTISTS hoping to help prevent car crashes have
hubs at will and may even be constrained in the created a computer system that they say was inspired
routes that it flies because of political reasons. In by a swarm of locusts. From the universities of Lincoln
such cases, the recourse would be to strengthen, and Newcastle, they have discovered that the inspiring
IT is by interdependence or interconnection that or even weaken, its internal connectivity to insect has a fast and accurate warning system that
things work – within the natural world, in reverse the effects of the nature of inter-network enables it to detect approaching objects and avoid them
communities, human settlements, the connects that circumstances have imposed. while in flight.
marketplace, roads, airline routes, supply chains, Based on
power distribution, epidemics, the Internet Uses of network strategies these findings,
There have been studies of how these networks Understanding the way networks change when the team
of related agents grow and stabilise and the Central to connections are modified helps remedy and behind the find-
features that make them efficient, responsive or central optimise real-life situations. In the case of a mix ings has creat-
robust. But there has been less work on how Nodes: Na of species living together, changes in the ed a computer
one network may benefit or lose in the way it Links: La Central to environment may be to the advantage of some. system that
connects to another network, or a set of peripheral Nodes: Nb There would then be evolutionary changes they believe
networks. Links: Lb among the less advantaged species, to bring could become
J Aguirre, D Papo and JM Buldú, of the Centro Ways of connecting about changes in the relationships of networks an early prototype for a technology that could prevent
de Astrobiología, Centre for Biomedical Techno- so that the balance could be restored. In the case collisions.
logy, Madrid, and the Complex Systems Group at of the evolution of populations of RNA Professor Shigang Yue, from the University of Lincoln’s
Móstoles, in Spain, report in the journal Nature Dolphins of Doubtful Sound molecules, for instance, there is work in progress School of Computer Science, said, “We were inspired by
their study of the interaction of networks to 0.95 to quantify the competition for the way different the way the locusts’ visual system works when interact-
derive rules, based on the internal features of 0.05 genetic strains express themselves, to better ing with the outside world and the potential to simulate
networks, of how the interaction must proceed understand their evolvability and adaptability. such complex systems in software and hardware for var-
— which would help intervene where networked Circles and squares The analysis of the way networks behave, as ious applications.
systems need to have, or to be kept away from, show actual dolphin carried out it the study of Aguirre and others, “We created a system inspired by the locusts’ motion
the power of other networks. networks, size helps understand these situations and also sensitive interneuron — the lobula giant movement
indicates centrality.
Competition, which promotes efficiency, is Thick lines show allows prescriptions of successful competition detector. This system was then used in a robot to enable
usually between individuals. But the outcome connector links strategies, and an engineering of desirable it to explore paths or interact with objects, effectively
may be affected not only by the competitors but connectivity patterns. Areas of useful application using visual input only.”
also the network of connections of the different 0.71 With connectors could be the distribution of populations, spread Professor Yue added that building a working system
agents involved. Networks usually evolve to that make for of epidemics, progress of rumours or patterns of was a major research challenge. She continued, “Vision
make for the best returns to all participants and 0.29 max centrality Internet navigation. plays a critical role in the interaction of most animal
also the stability of the network. But things of squares. The green An unlikely area where the science has found species and even relatively low-order animals have
change when the networks need to, or have to, shading shows cases application is the Dolphins of Doubtful Sound, a remarkable visual processing capabilities. For example,
contend with other networks, often in where nodes have fiord in the far south-west of New Zealand. insects can respond to approaching predators with
competition for limited resources. This raises a gained in centrality Doubtful Sound, first called Doubtful Harbour by remarkable speed.
question of how one network can best use or Captain Cook who discovered it in 1770, is a “This research demonstrates that modelling biological-
save itself against the advantage that the other stretch of water that consists of two distinct ly plausible, artificial visual neural systems can provide
network could gain from the interaction. An 0.99 With connectors layers — one some two to 10 metres deep, of new solutions for computer vision in dynamic environ-
example may be of an airline that shares one of for max centrality fresh water from surrounding mountains, and ments. For example, it could be used to enable vehicles
its service or stopover facilities with another 0.29 of the circles. The stained red-brown due to tannin, and below this to understand what is happening on the road ahead and
airline. How should the two airlines modify their increase in centrality a warmer and denser salt water body. The tannin take swifter action.”
flight schedules so that each takes optimum by this connecting stain in the upper layer blocks sunlight and the The research was carried out as part of a collaborative
strategy may not
advantage of the arrangement? always be a good thing. lower layer has become home to many deep sea project with the University of Hamburg in Germany and
To answer this question, we note that each For instance, in the case of a disease- species, despite being quite shallow. These Tsinhua University and Xi’an Jiaotong University in China.
network consists of nodes, with links to other spreading process, the circles network would be more vulnerable include the bottlenose dolphin and the fiord Dr Claire Rind, from Newcastle University’s Institute
nodes. The importance of a node arises from the houses an insular community of these animals, for Neuroscience, said, “Developing robot neural net-
number of links it has and how well connected connection links, then the combined network and the factors that arise from the connections which are locked away from the open sea both work programmes, based on the locust brain, has
other nodes are and a measure of the value of has Na+Nb nodes and La+Lb+L liknks. The made. In the case of one of the two networks by the cooler habitat in the fiord and also the allowed us to create a programme that permits a
the network could be the total importance of its centrality, Ca and Cb, of the two networks, being clearly more connected than the other, it fresh water separator. But the community is only mobile robot to detect approaching objects and avoid
nodes. When networks interact, they create however, would depend on the way the nodes turns out that: some 70 strong and presents a major concern for them.
common links and the question of competition are connected and the centrality, Ca+Cb, of the ■ Connecting the peripheral nodes of both the conservation. “It’s not the conventional approach as it avoids using a
reduces to one of how the value of each network combined network, would depend on the networks optimises the centrality of the more Studying the dolphin community has revealed radar or infrared detectors, which require very heavy-
gets affected by the links created. Quantification manner of linking the networks. connected one; sub-communities and evidence that a sex- and duty computer processing. Instead, it is modelled on the
of the importance of nodes is done through a The nodes of a network could be classified ■ But connecting the central nodes of the two age-related tendency to associate with like locust’s eyes and neurones as the basis of a collision
concept of centrality — which is to assign according to how important they are — the well- networks optimises centrality of the weaker individuals plays a role in the formation of avoidance system.”
relative scores to all nodes in the network based connected ones would be “central” and the network; clusters of preferred companionship. Also Dr Rind added, “While some collision-avoidance fea-
on the principle that connections to high-scoring poorly connected ones would be “peripheral”. ■ Increasing the number of connector links identified are brokers who act as links between tures are pricey options on luxury cars, their perfor-
nodes contribute more to the score of the node There could then be four kinds of connector strengthens the weaker network; sub-communities and who appear to be crucial mance is not always as good as it could be — and they
in question than equal connections to low- links between the networks, and these are ■ In general, a network stands to gain if it to the social cohesion of the population as a come at a high cost. This research offers us important
scoring nodes. And in a group of networks, the shown in the picture. Working out the way increases its own number of links and whole. Network analysis will hopefully result in insights into how we can develop a system for the car
importance of a network is the total of the centrality of networks depends on the connections; finding ways to raise populations and save the which could improve performance to such a level that
centralities of all its nodes. parameters of network before they are ■ While the stronger network generally retains species! we could take out the element of human error.”
A simple case of networks A and B is shown. If interconnected, and on the possible connector its advantage, in the case of large and
the nodes and links in A and B are Na, La and links, provides mathematical expressions that comparable networks, the strategy in choosing The writer can be contacted at the independent
Nb,, Lb and the two networks are connected by L include the values of centralities of the networks connector nodes determines the benefit that [email protected]

Jumping beans and free energy

IF you find the concepts of free energy, entropy and Thus, ∆G is first and foremost a measure of the
equilibrium constants difficult to grasp, perhaps a capacity of a system to do work under specified
simple analogy might help. For this, we will need an conditions. You might, in fact, want to think of ∆G
imaginary supply of jumping beans, which are really as free energy in the sense of energy that is free or
seeds of certain Mexican shrubs, with the larvae of the available to do useful work. Moreover, if we
Laspeyresia saltitans moth inside. Whenever the contrive to keep ∆G negative by continuously
larvae inside the seed wiggle about, the seeds wiggle, two chambers are equal at equilibrium, so the be the most important thermodynamic parameter for adding beans to Chamber 1 and removing them
too. The “jumping” action probably serves to get the equilibrium constant for the jumping reaction under tapan kumar maitra our purposes. ∆G is defined so that negative values from Chamber 2, we have a dynamic steady state, a
larvae out of direct sunlight, which could heat them to these conditions is 1.0. presents a simple analogy to correspond to favourable (ie, thermodynamically condition that effectively harnesses the inexorable
lethal temperatures. spontaneous) reaction and positive values represent drive to achieve equilibrium. Work can then be
For purposes of illustration, imagine we have some Enthalpy change (∆ ∆H) explain the concept unfavourable reactions. Thus, ∆G should have the performed continuously by beans that are forever
high-powered jumping beans in two chambers Now suppose the level of Chamber 1 is somewhat same sign as AH (since a negative AH is vorable), but jumping toward equilibrium but never actually
separated by a low partition, as shown. Notice that the higher than that of Chamber 2, as shown in the next the opposite sign from ∆S (since for ∆S, a positive reach it.
chambers have the same floor area and are at the same diagram. Jumping beans placed in Chamber 1 will the equilibrium constant will be greater than 1 in this value is favourable). In terms of real-life
level, although we will want to vary both of these again tend to distribute themselves between case also. This means that the equilibrium position of thermodynamics, the expression for ∆G in terms of
properties shortly. As soon as we place a handful of Chambers 1 and 2, but this time a higher jump is the jumping reaction has been shifted to the right, ∆H and ∆S is: ∆G = ∆H - T∆S. (Notice that the
jumping beans in Chamber 1, they begin jumping required to get from 2 to 1 than from 1 to 2, so the even though there is no change in enthalpy. temperature dependence of DS is the only feature of
about randomly. Although most of the beans jump latter will occur more frequently. As a result, there this relationship that cannot be readily explained by
only to a modest height most of the time, occasionally will be more beans in Chamber 2 than in Chamber 1 our model, unless we assume that the effect of
one of them, in a burst of ambition, gives a more at equilibrium, and the equilibrium constant will changes in room size is somehow greater at higher
energetic leap, surmounting the barrier and falling therefore be greater than 1. temperatures.)
into Chamber 2. We can write this as the jumping
reaction: DG and the capacity to do work
Beans in Chamber 1 D Beans in Chamber 2 1 2 1 2 You should be able to appreciate the difficulty of
We will imagine this to be a completely random event suggesting a physical equivalent for ∆G because it 1
happening at irregular, infrequent intervals. The floor area of the chambers can be thought of represents an algebraic sum of entropy and energy
Occasionally, one of the beans that has reached as a measure of the entropy, or randomness, of the changes, which may either reinforce or partially offset
Chamber 2 will happen to jump back into Chamber 1, 1 1 system and the difference between the two chambers each other. But as long as ∆G is negative, beans will
which is the back reaction. At first, of course, there will can be represented by ∆S. Since Chamber 2 has a continue to jump from Chamber 1 to Chamber 2,
be more beans jumping from Chamber 1 to Chamber 2 greater floor area than Chamber 1, the entropy whether driven primarily by changes in entropy, 2
2 because there are more beans in Chamber 1, but 2 change is positive for the jumping reaction as it pro- internal energy, or both. This means that if some sort
things will eventually even out so that, on average, The relative heights of the two chambers can be ceeds from left to right under these conditions. Note of bean-powered “bean wheel” is placed between the Looking ahead
there will be the same number of beans in both thought of as measures of the enthalpy, or heat that for ∆H, negative values are associated with two chambers, as shown below, To anticipate the transition from the
compartments. The system will then be at equilibrium. content (H), of the chambers, such that Chamber 1 favourable reactions, while for ∆S, favourable thermodynamics of this chapter to the kinetics of
Beans will still continue to jump between the two has a higher H value than Chamber 2, and the reactions are indicated by positive values. the next, begin thinking about the rate at which
chambers, but the numbers jumping in both directions difference between them is represented by ∆H. Since beans actually proceed from Chamber 1 to
will be equal. it is a “downhill” jump from Chamber 1 to Chamber Free energy change (∆G)) Chamber 2. Clearly, ∆G measures how much
2, it makes sense that ∆H has a negative value for the So far, we have encountered two different factors energy will be released if beans do jump, but it
jumping reaction from Chamber 1 to Chamber 2. that affect the distribution of beans: the difference in 1 says nothing at all about the rate. That would
Similarly, it seems reasonable that ∆S for the reverse levels of the two chambers (∆H) and the difference appear to depend critically on how high the barrier
reaction should have a positive value because that in floor area (∆S). Moreover, it should be clear that 2 between the two chambers is. Label this the
jump is “uphill”. neither of these factors by itself is an adequate the movement of beans from one chamber to the activation energy harrier and then contemplate the
indicator of how the beans will be distributed at other can be harnessed to do work until equilibrium means by which you might get the beans to move
Entropy Change (∆ ∆S) equilibrium because a favourable (negative) ∆H is reached, at which point no further work is possible. over the barrier more rapidly.
1 2 1 2 So far, it might seem as if the only thing that can could be more than offset by an unfavourable Furthermore, the greater the difference in free energy One approach might be to heat the chambers;
The equilibrium constant affect the equilibrium distribution of beans between (negative) ∆S, and a favourable (positive) ∆S could between the two chambers (that is, the more highly this would be effective because the larvae inside
Once our system is at equilibrium, we can count the the two chambers is the difference in enthalpy, ∆H. be more than offset by an unfavourable (positive) negative ∆G is), the more work the system can do. the seeds wiggle more vigorously if they are
number of beans in each chamber and express the But that is only because we have kept the floor area ∆H. You should, in fact, be able to design chamber warmed. Cells, on the other hand, have a far more
results as the ratio of the number of beans in Chamber of the two chambers constant. Imagine, instead, the conditions that illustrate both of these situations, as effective and specific means of speeding up
2 to the number in Chamber 1. This is simply the situation shown below, where the two chambers are well as situations in which ∆H and ∆S tend to reactions: they lower the activation barrier by using
equilibrium constant Keq for the jumping reaction: again at the same height, but Chamber 2 now has a reinforce rather than counteract each other. 1 catalysts called enzymes.
number of beans in chamber 2 at equilibrium greater floor area than Chamber 1. The probability of Clearly, what we need is a way of summing up
Keq = ————————————————————————— a bean finding itself in Chamber 2 is, therefore, these two effects algebraically to see what the net The writer is associate professor and
number of beans in chamber 1 at equilibrium correspondingly greater, so there will be more beans tendency will be. The new measure we come up with head, Department of Botany, Ananda
For the specific case, the numbers of beans in the in Chamber 2 than in Chamber 1 at equilibrium, and is called the free energy change, ∆G, which ought to 2 Mohan College, Kolkata

C M Y K