Anticipations of The Geometric Phase: Additional Resources For Physics Today
Anticipations of The Geometric Phase: Additional Resources For Physics Today
Anticipations of The Geometric Phase: Additional Resources For Physics Today
Michael Berry
Downloaded 15 Apr 2013 to 136.159.235.223. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://www.physicstoday.org/about_us/terms
ANTICIPATIONS
OF THE
GEOMETRIC PHASE
The notion that a quantum system's
wovefunction may not return to its original
phase after its parameters cycle slowly around
a circuit had many precursors—in polarized
light, radio waves, molecules, matrices
and curved surfaces.
Michael Berry
In science we like to emphasize the novelty and originality A physical example of this "global change without
of our ideas. This is harmless enough, provided it does not local change" is the Foucault pendulum (figure 1), whose
blind us to the fact that concepts rarely arise out of direction of swing, described by a unit vector e, is slaved to
nowhere. There is always a historical context, in which the local vertical, described by the radial unit vector r.
isolated precursors of the idea have already appeared. The slaving law is parallel transport, which means that
What we call "discovery" sometimes looks, in retrospect, the direction of swing does not rotate about the vertical—
more like emergence into the air from subterranean that is, e has no component of angular velocity along r.
intellectual currents. However, in spite of never being rotated, e does not return
The geometric phase, whose discovery I reported early to its original value when, after a day, r has completed a
in 1983, is no exception to this rule.1 The paper was about circuit C (here a circle of latitude). The anholonomy is the
quantum systems forced round a cycle by a slow circuit of angle between the initial and final swing directions e, and
parameters that govern them; it gave rise to a number of is equal to the solid angle subtended at the Earth's center
applications and several generalizations, documented in a byC.
series of reviews and books.2"5 My purpose here is to look A note about terminology: Although the anholonomy
back at some early studies that with hindsight we see as of parallel transport of a vector on a curved surface was
particular examples of the geometric phase or the central known to Gauss nearly two centuries ago, the word seems
idea underlying it. to have entered the literature through the study of
mechanics in the presence of constraints. A constraint is
Parallel transport holonomic if it can be integrated and thereby can reduce
First I need to explain this central idea. It is the geometric the number of degrees of freedom, as with a rolling
phenomenon of anholonomy resulting from parallel trans- cylinder. Otherwise, it is nonholonomic (or nonholono-
port. This is a type of nonintegrability, arising when a mous, or anholonomic), as with a rolling disk, which can
quantity is slaved to parameters so as to have no local rate sway from side to side. According to the Oxford English
of change when those parameters are altered, but never- Dictionary the word was first used by Hertz in 1894.
theless fails to come back to its original value when the pa- Nowadays the concept of anholonomy is familiar to
rameters return to their original values after being taken geometers, but they often call it "holonomy," a reversal of
round a circuit. usage I consider a barbarism.
The geometric phase can be regarded as anholonomic
for the parallel transport of quantum states. Mathemat-
Michael Berry is a professor of physics at the University of ically, quantum states are represented by unit vectors in
Bristol, in Bristol, England. Hilbert space. Although these unit vectors are complex,
parallel transport can still be defined. A natural way to direction of propagation, which they accomplished by
implement it is by a slow cycle C of parameters in the sending the light along optical fibers that were coiled into
Hamiltonian governing the evolution of the system paths such as helices, and for which the initial and final
according to Schrodinger's equation. The quantum adia- tangent directions r were parallel. An obvious way to
batic theorem guarantees that if the system starts in the observe the geometric phase would be to split a beam of
instantaneous eigenstate labeled n, it will still be in the (say) left circularly polarized light into two coherent
state n at the end of the cycle C. However, the phase of the beams, send them along two oppositely coiled fibers,
state vector need not, and usually does not, return. Part of recombine them and detect the resulting opposite geomet-
this change—the geometric phase—is the manifestation of ric phases by interference. Instead, Chiao, Tomita and Wu
anholonomy. performed the simpler experiment of sending a single
To give an account of the earlier work, I have first to beam of linearly polarized light along a single fiber. The
describe the geometric phase for spinning particles. This initial linear polarization is a particular superposition of
concerns a spinor state corresponding to a definite value s the s = + 1 and s = — 1 states, which acquire opposite
(integer or half-integer) for the component of spin along phases after passage through the fiber and so emerge in a
some direction r. An example is a spin eigenstate different superposition, corresponding again to linear
(s = ± V2) of a neutron in a magnetic field with direction r. polarization, but now in a different direction. (The
If the direction is cycled, that is, taken round a closed "interference" and "superposition" techniques correspond
curve Con the unit r sphere, the state acquires a geometric to two different general methods for detecting the phase,
phase equal to — s times the solid angle subtended by C at employing, respectively, one state and two different
the center of the sphere. As is well known, the spin-1/, case Hamiltonians or two states and one Hamiltonian.)
is isomorphic to the general quantum two-state system, One manifestation of the geometric phase for light is
where the Hamiltonian is a 2x2 complex Hermitian therefore a rotation of the direction of polarized light
matrix. (figure 2) after it has traveled along a coiled optical fiber.
The angle of rotation is equal to the solid angle through
Coiled light which the fiber tangent r has turned, implying that the po-
Raymond Y. Chiao, Akira Tomita and Yong-Shi Wu were larization has been parallel-transported. Chiao and his
quick to apply the spin phase to optics, by regarding a light coworkers themselves pointed out that this appears to be a
beam as a stream of photons with quantization direction r phenomenon of classical optics, which although originat-
along the direction of propagation.2 The two states, ing in the quantum mechanics of spinning photons
s = + 1, correspond to left- and right-handed circularly survives the classical limit ft — 0 up to the level described
polarized light. To cycle r they therefore had to cycle the by Maxwell's equations. They did not, however, show how
Fiber tangent r
M
V i' w
The three typical patterns of curvature lines near an umbilic singularity, where the surface is locally spherical.
The star has index — '/,. The lemon and monstar have index + '/,. Figure 8
mula for the phase, he showed how its existence, for a 2. A. Shapere, F. Wilczek, eds., Geometric Phases m Physics,
succession of circuits that together cover a closed surface, World Scientific, Singapore (1989).
could provide a topological indicator of the presence of a 3. R. Jackiw, Commun. At. Mol. Phys. 21, 71 (1988). J. W.
degeneracy. Zwanziger, M. Koenig, A. Pines, Annu. Rev. Phys. Chem.
Another anticipation, of which I was regrettably (1990), in press.
unaware when writing my original paper,' was the 4. B. Markovski, V. I. Vinitsky, Topological Phases in Quantum
important work by C. Alden Mead and Donald G. Truhlar Theory, World Scientific, Singapore (1989).
in 1979, containing two developments in the theory of 5. M. V. Berry, in Anomalies, Phases, Defects, M. Bregola, G.
general complex Hamiltonians.23 This theory would Marmo, G. Morandi, eds., Bibliopolis, Naples (1990), p. 125.
apply, for example, to systems with magneticfields,which 6. S. M. Rytov, Dokl. Akad. Nauk. USSR 18, 263 (1938); reprint-
do not have time-reversal symmetry. Like Longuet- ed in ref. 4, p. 6. V. V. Vladimirskii, Dokl. Akad. Nauk. USSR
Higgins and his coworkers, Mead and Truhlar were 21, 222 (1941); reprinted in ref. 4, p. 11.
studying molecules in the Born-Oppenheimer approxima- 7. M. V. Berry, Nature 326, 277 (1987); reprinted in ref. 2, p. 197.
tion. The first of the developments was that they not only
realized that the electronic states must acquire a phase 8. M. V. Berry, in Fundamental Aspects of Quantum Theory,
when the nuclear coordinates are cycled, but they also NATO Adv. Stud. Inst. Ser., vol. 144, V. Gorini, A. Frigerio,
gave a general formula for the phase in the case of eds., Plenum, New York (1986), p. 267.
infinitesimal circuits. Second, they discovered another 9. S. Pancharatnam, Proc. Ind. Acad. Sci. A 44, 247 (1956);
role for the expression whose line integral around the reprinted in ref. 2, p. 51. S. Ramaseshan, R. Nityananda,
circuit generates the phase: It is the potential of an Curr. Sci. (India) 55, 1225 (1986).
effective "gauge force" contributing to the dynamics of the 10. M. V. Berry, J. Mod. Opt. 34,1401 (1987); reprinted in ref. 2, p.
nuclei. The effect of this force is to modify the nuclear vi- 67.
bration-rotation spectrum, as in the special case of 11. M.S. Smith, Proc. R. Soc. London, Ser. A 346, 59(1975). K. G.
pseudorotation mentioned earlier. Budden, M. S. Smith, Proc. R. Soc. London, Ser. A 350, 27
As elaborated elsewhere,24 in 1983 I was familiar with (1975).
the 77 phase shifts of Longuet-Higgins and Darboux 12. M. V. Berry, Proc. R. Soc. London (1990), in press.
through studies of the quantum mechanics corresponding 13. H. C. Longuet-Higgins, U. Opik, M. H. L. Pryce, R. A. Sack,
to classical chaos, where degeneracies play a useful part. Proc. R. Soc. London, Ser. A 244, 1 (1958).
In retrospect it now appears natural that the generaliza- 14. G. Herzberg, H. C. Longuet-Higgins, Faraday Soc. Disc. 35, 77
tion to the full geometric "phase that launched a thousand (1963); reprinted in ref. 2, p. 74.
scripts" should have been made in my department at 15. G. Delacretaz, E. R. Grant, R. L. Whetten, L. Woste, J. W.
Bristol. The reason is that in Bristol there had been Zwanziger, Phys. Rev. Lett. 56, 2598 (1986); reprinted in ref. 2,
several discoveries, over the years, of interesting physics p. 240.
associated with quantities that fail to return after being 16. V. I. Arnold, Mathematical Methods of Classical Dynamics,
taken round circuits—that is, anholonomy.21 I have fol- Springer-Verlag, New York (1978).
lowed that intellectual thread elsewhere, and here 17. K. Uhlenbeck, Am. J. Math. 98, 1059 (1976).
simply list some of those contributions: the descriptions by 18. G. Darboux, Lecons sur la Theorie Generale des Surfaces, vol.
F. Charles Frank of crystal dislocations (1951) and liquid 4, Gauthier-Villars, Paris (1896), note VII, p. 448.
crystal disclinations (1958) in terms of anholonomy; the 19. M. V. Berry, J. H. Hannay, J. Phys. A 10, 1809 (1977).
description of the n-phase for molecular electrons (1958) by 20. F. C. Frank, Faraday Soc. Disc. 25, 19 (1958).
Pryce, one of Longuet-Higgins's coauthors and head of the 21. M. V. Berry, in Sir Charles Frank 80th Birthday Festschrift,
Bristol physics department; and the discovery by Yakir R. G. Chambers, J. E. Enderby, A. Keller, A. R. Lang, J. W.
Aharonov and David Bohm of the electron phase shift in a Steeds, eds., Adam Hilger, Bristol (1990), in press.
circuit of a magnetic flux line (1959). 22. A. J. Stone, Proc. R. Soc. London, Ser. A 351, 141 (1976);
reprinted in ref. 2, p. 80.
References 23. C. A. Mead, D. G. Truhlar, J. Chem. Phys. 70, 2284 (1979);
1. M. V. Berry, Proc. R. Soc. London, Ser. A 392, 45 (1984); reprinted in ref. 2, p. 90.
reprinted in ref. 2, p. 124. 24. M. V. Berry in reference 2, p. 26. •