PERSPECTIVES

PHYSICS

When Infinity Does Not Count

Vadim V. Cheianov

A transformation of variables overcomes fundamental difficulties in formulating a unified theory of one-dimensional quantum fluids.

lassical fluid dynamics is an incredi- perturbations of a one-dimensional quantum gas (an ideal gas not of classical particles but bly flexible mathematical tool that can fluid are captured by the Luttinger liquid theory of fermions) (8). In this case, the exact result describe fluids as diverse as water and (7), which describes how sound waves move does not correspond to what one would expect the dense nuclear matter of neutron stars. through the fluid (for example, their velocity). from a rare-event random process (see the figHowever, when a fluid is cooled to tempera- Each sound wave of frequency f can accommo- ure, panel B). As time passes, the phonon distures close to 0 K and exhibits quantum date an integer multiple of the energy quantum, appears, but then continues to reappear and effects, a more fundamental theory must be E = hf, where h is Plancks constant. Such disappear. This result suggests that the used. Yet, attempts to introduce quantum energy quanta constitute elementary particles description of phonon decay in terms of rare effects as perturbations to the classical theory of the system and are called phonons. events is wrong. The disagreement cannot be have not been successful: The solutions The price to pay for the simplicity of the blamed on the particular model used; if quandiverge and give meaningless, infinitely large Luttinger model is that it neglects all of the tum hydrodynamics is a universal theory, it answers. On page 228 of this issue, Imambekov nonlinear terms in the equations of fluid should be insensitive to the exact microscopic and Glazman (1) consider a quantum fluid dynamics, which leaves out a lot of interesting nature of the system. that moves in one spatial dimension and in- physics. For instance, in a nonlinear system, The lack of agreement with the exact vestigate how quantum effects influence the a phonon could decay into several other phonon decay pattern indicates that more decay of elementary excitations caused by the nonlinearity of fluid A B C D dynamics. The authors apply a universal, nonperturbative transforExact hf1 mation of variables, and the new theoretical formulation generates hf0 finite results. Random process The problem of divergent quanhf2 tum corrections is not unique to Time Time Elementary decay process Time fluid dynamics. Initially encountered in the early 1930s at the interface of quantum mechanics and Following the fate of phonons. One measure of a quantum fluid system is how sound waves, or phonons, decay as Maxwells theory of light, it is they move through it. (A) A diagram illustrating the simple case of a phonon of frequency f0 decaying into two weaker phonons of frequencies f1 and f . (B) The probability of detecting a phonon in a model system, the one-dimensional inherent to quantum field theory. Fermi gas, changes as a function2 of time. The exact result from the model (pink curve) does not match the prediction of Infinities encountered in perturba- the theory that includes only the lowest-order decay process as a rare random event (blue curve). This mismatch lies in tion theory can be cured with a ignoring higher-order decay processes. For example, a cascade of events (C) causes a phonon to decay into four weaker procedure called renormalization phonons. In a bad process (D), phonons created in the decay process exchange a virtual phonon. The contribution of (2). This procedure has been highly this process diverges to infinity. successful in both high-energy and condensed-matter physics, but it fails in some phonons. This decay process illustrates how complex phonon decay patterns must be taken important nonlinear theories. The most notori- divergences enter into the quantum descrip- into consideration; in fact, there are infinitely ous example is Einsteins theory of gravity (3), tion of this fluid. Consider, for example, a many higher-order processes (one example is which has interesting parallels with the quan- phonon decaying into two others (illustrated shown in panel C of the figure). In perturbatum theory of fluids (4). Whether there exists a in the figure, panel A). If the theory is weakly tive field theory, there are well-developed meaningful quantum formulation of nonlinear nonlinearthat is, nonlinearity can be methods for summing up all of these contritheories that do not lend themselves to renor- described as small corrections to the linear butions, but in nonlinear fluid dynamics, this malization is an open question. theorysuch processes would be expected to is where the real trouble starts. Phonons born Some simplification of quantum fluid be random rare events, like radioactive decay. by the decay of the initial phonon may dynamics is achieved by considering the prob- If this is true, then the probability P for detect- exchange a virtual phonon (see the figure, lem in one spatial dimension. A few decades ing a phonon after a time t would decay expo- panel D), and the contribution of each such ago, this exercise would have been regarded as nentially as et/, where is the phonon life process is infinite (9). This infinity is similar a mathematical abstraction with no physical time (see the figure, panel B). to the short-distance infinity encountered in examples. Today, several examples of oneHow can we tell if this is what actually hap- quantum electrodynamics, except that it is dimensional quantum fluids are known, includ- pens? Fortunately, there are models of many- much stronger and cannot be cured by means ing electron liquids in carbon nanotubes (5) and particle systems that have a complete quan- of renormalization. cold atomic gases in elongated traps (6). Small tum solution and whose large-scale physics is Rather than trying to make sense out of governed by fluid dynamics. For example, the meaningless infinities in phonon processes, probability of detecting a phonon can be cal- Imambekov and Glazman make a change of Department of Physics, Lancaster University, Bailrigg, culated exactly for a one-dimensional Fermi variables. The nonlinear quantum transformaLancaster LA1 4YB, UK. E-mail: v.cheianov@lancaster.ac.uk

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

213

Published by AAAS

Downloaded from www.sciencemag.org on September 20, 2009

Detection probability

PERSPECTIVES
tion they use was initially proposed for the analysis of a model describing interacting relativistic particles in two-dimensional spacetime (10). The theory now operates in a different world of particles: Phonons, which are normal modes of the entire system, are replaced by weakly interacting soliton-like particles that are fermions traveling through the system. The new theory cannot be renormalized, which should be expected given that it is the old one in disguise. However, if residual weak interactions between the new soliton-like particles are neglected, the theory can be treated exactly, and the interactions between the prototype particlesthe phononsremain practically unchanged. These interactions are encoded in the nonlinear transformation rules from one theory to the other and allow any hydrodynamic observable to be calculated explicitly. The beauty of Imambekov and Glazmans work is that it provides an example of a meaningful quantum field theory for a problem lacking a consistent perturbative formulation. The mathematics of this theory is not only interestingit describes real systems with rich phenomenology, and should allow for further explorations. For example, it should be possible to develop a theory of quantum wave breaking and quantum shock waves.
References
1. A. Imambekov, L. I. Glazman, Science 323, 228 (2009); published online 27 November 2008 (10.1126/ science.1165403). 2. R. P. Feynman, QED: The Strange Theory of Light and Matter (Princeton Univ. Press, Princeton, NJ, 1985). 3. D. Kleppner, R. Jackiw, Science 289, 893 (2000). 4. G. Volovik, JETP Lett. 82, 319 (2005). 5. S. J. Tans et al., Nature 386, 474 (1997). 6. H. Moritz, T. Stferle, M. Khl, T. Esslinger, Phys. Rev. Lett. 91, 250402 (2003). 7. T. Giamarchi, Quantum Physics in One Dimension (Oxford Univ. Press, Oxford, UK, 2003). 8. M. Pustilnik, M. Khodas, A. Kamenev, L. I. Glazman, Phys. Rev. Lett. 96, 196405 (2006). 9. D. N. Aristov, Phys. Rev. B 76, 085327 (2007). 10. D. Mattis, E. Lieb, J. Math. Phys. 6, 304 (1965). 10.1126/science.1168389

MATERIALS SCIENCE

Unjamming a Polymer Glass

David A. Weitz

Small probes reveal that glass can melt in different ways.

hen a glass is heated, it melts and begins to flow. This transition from an elastic solid to a flowing fluid is a distinguishing feature of the glass transition, one of the most widely studied, yet incompletely understood, phase transitions (1). The application of stress can also make a glass flow; softer glasses, including many polymers, yield when subjected to sufficiently Shear flow. Schematics comparing shear-induced flow in a granular material (left) and a glassy polymer (right). large stresses (2). The equivalence of these two routes to flow is a basic tenet of jam- motion of the probe molecules, becomes at the molecular scale, the glass melts due to ming, a conceptual means of unifying glassy measurable and begins to increase with tem- the induced strain. behavior with that of granular materials such as perature, adopting the characteristic stretched This shear-induced melting is exactly what sand (3). The shear-induced flow of sand, or exponential form of a fluid very close to the is expected within the jamming picture. The other granular materials, is well studied. On glass transition (5). Moreover, the structural basic concepts of jamming can be understood page 231 of this issue, Lee et al. (4) show that relaxation exhibits the strong spatial hetero- from the perspective of a granular material, the nature of shear-induced flow in molecular geneity commonly observed in a fluid very such as a bucket of sand. Normally, the sand in glasses can now be probed. By measuring the close to the glass transition (5, 6). The new the bucket is a solid; it does not flow and it motion of small probe molecules in a polymer experiments investigate the behavior of the supports a stress, as easily proven by stepping glass, they find fluidlike properties when the probe molecules when the polymer is sub- on itthe sand supports your weight. glass is sheared; however, the route to melting jected to an external uniaxial tension; the sam- However, if you tip the bucket, the sand flows, the glass is different from that followed when ple is pulled apart by its two ends. Initially, much like a fluid. Here, gravity provides the it is heated. when the strain is small, the structural relax- shear stress that causes the sand to change Lee et al. use an optical method to measure ation rate increases slightly, and can be from a solid to a fluid. To make the analogy the rate of rotation of small dye molecules described by a theory that incorporates the between granular sand and a glass requires a embedded within thin slabs of lightly cross- effects of the induced strain energy, which second route to fluidizing the sandby linked poly(methyl methcrylate) (PMMA). lowers the activation barriers for the relax- increasing its effective temperature. This can These probe molecules provide a direct, local ation of the probe molecules due to their be accomplished by gently shaking it, or by measure of the fluidity of the polymer. As the shear-induced rotation (7). The shape of the blowing air slowly up through the sand, to temperature is increased and the polymer decay retains the characteristic stretched slightly suspend all the grains in the flow of air glass starts to flow, the structural relaxation exponential form. However, as both the strain (8). In this case, the grains are rapidly moving, rate of the polymer, as determined by the and strain rate increase, the relaxation rate of but are trapped in place by all their neighbors. the probe molecules increases by several This rapid random motion of the grains is akin orders of magnitude, and the shape of the to an increased effective temperature. To comDepartment of Physics and School of Engineering and relaxation becomes nearly exponential. At this plete the jamming picture, there is a third Applied Sciences, Harvard University, Cambridge, MA point, the sample undergoes plastic flow, and means of fluidizing a solid, and that is to 02138, USA. E-mail: weitz@seas.harvard.edu

214

9 JANUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

Published by AAAS

CREDIT: P. HUEY/SCIENCE

Downloaded from www.sciencemag.org on September 20, 2009