1: %000817-001124 Vienna - web
2: \documentstyle[epsf]{article}
3: \begin{document}
4:
5: \parskip 2mm plus 1mm \parindent=0pt
6: \def\cl{\centerline}\def\lel{\leftline}\def\rl{\rightline}
7: \def\hs1{\hskip1mm} \def\h10{\hskip10mm} \def\hx{\h10\hbox}
8: \def\vs{\vskip3mm} \def\vup{\vskip-2mm}
9: \def\page{\vfill\eject}
10: \def\<{\langle} \def\>{\rangle} \def\br{\bf\rm} \def\it{\tenit}
11: \def\de{\partial} \def\Tr{{\rm Tr}} \def\dag{^\dagger}
12: \def\half{{\scriptstyle{1\over 2}}}
13: \def\ne{=\hskip-3.3mm /\hskip3.3mm}\def\tr{{\rm tr}}
14: \def\Pr{{\br Pr}}\def\to{\rightarrow}\def\N{{\cal N}}
15: \def\vb{\vskip20mm}\def\vm{\vskip10mm}
16: \def\cite{}
17:
18:
19: \def\be{\begin{equation}}\def\ee{\end{equation}}
20:
21:
22: \vskip10mm
23: \cl{\bf Speakable and unspeakable after John Bell} %- 001124
24: \vs\vs
25: \cl{\it A talk given at the International Erwin Schr\"odinger
26: Institute, Vienna (ESI)}
27: \vs\cl{at the November 2000 Conference in commemoration of John Bell}
28: \vs\vs\vs
29: \cl{2000 Dec 05}
30:
31: \vskip10mm
32: \cl{Ian C. Percival}
33: \vskip7mm
34: \cl{(i.c.percival@qmw.ac.uk)}
35: \vskip15mm
36:
37: {\bf Abstract} `Philosophy' was speakable for John Bell but is not for
38: many physicists. The border between philosophy and physics is here
39: illustrated through Brownian motion and Bell experiments.
40: `Measurement', however, was unspeakable for Bell. His insistence that
41: the physics of quantum measurement should not be confined to the
42: laboratory and that physics is concerned with the big world outside leads
43: us to examples from zoology, meteorology and cosmology.
44:
45: \page
46:
47:
48: {\bf Philosophy and physics}
49:
50: I first met John and Mary Bell at CERN because of our common
51: interest in the applications of nonlinear dynamics, but late
52: in life I became sufficiently fascinated by quantum philosophy
53: to follow John Bell's seminal contributions to that field, his
54: `hobby'.
55:
56: `Philosophy' is an unspeakable word for many physicists, but it was
57: not for John Bell. He turned the question of local reality, or
58: nonlocality, from a philosophical question to one of physics, by
59: proposing an experimental test, the breaking of inequalities between
60: quantities that are measurable in the laboratory.
61:
62: The boundary between philosophy and physics is not always clear. Take
63: the earlier example of Brownian motion: `` Besides, there is a further
64: reason why you should give your mind to these particles that are seen
65: dancing in a sunbeam: their dancing is an actual indication of
66: underlying movements of matter that are hidden from our sight. There
67: you will see many particles under the impact of invisible blows
68: changing their course and driven back upon their tracks, this way and
69: that, in all directions. You must understand that they all derive
70: this restlessness from the atoms. It originates from the atoms, which
71: move of themselves. \dots So the movement mounts up from the atoms and
72: gradually emerges to the level of our senses ...''
73:
74: If you wanted to explain Brownian motion to a poet in the twenty-first
75: century, you couldn't do much better than this. But it was not
76: written in this century. It was written in Latin verse {\it by} the
77: poet and Epicurean philosopher Lucretius, before 55BC. [1]
78:
79: Despite the clarity and accuracy of the description, the poem shows no
80: evidence of physics as we know it since the Renaissance. For Brownian
81: motion, Brown's systematic studies were published in 1828, following
82: earlier observation using microscopes. The comparison of
83: the detailed theory of Einstein and the experiments of Perrin
84: in the early years of the twentieth century finally established
85: for almost all physicists the reality of atoms.
86:
87: Physics has many sides, theoretical and experimental, and one
88: important part of it is asking the right questions. Ancient
89: philosophers like Lucretius did that, and modern
90: physicist-philosophers like Einstein, Bell and Shimony have done it
91: too. They have survived on dangerous ground, and built a bridge
92: between unspeakable philosophy and speakable physics.
93:
94: In physics we are required to put the answers to these questions to
95: the test, particularly experimental test, and here there is a parting
96: of the ways. Lucretius could not distinguish the atomic hypothesis
97: from a continuum hypothesis, or any other. Lucretius's was a
98: philosopher's poem. There was keen observation, but to our knowledge
99: it was not systematic, and there was no attempt to vary the
100: conditions. His explanation was plausible, but he could not check the
101: validity of the atomic hypothesis. That did not happen for another
102: 1950 years.
103:
104: In providing his tests of nonlocality, Bell showed how
105: the implicitly philosophical considerations of Einstein, Podolsky and
106: Rosen [2] might be tested in the laboratory. Like the theoretical
107: physicists who studied Brownian motion, he turned the philosophy into
108: physics.
109:
110: \vs
111: {\bf Black boxes and nonlocality}
112:
113: Einstein [3] and Bell [4] were
114: particularly concerned with the question of nonlocal causality, in
115: which cause and effect are spatially separated in spacetime, so that a
116: signal from cause to effect would have to go faster than the velocity
117: of light.
118:
119: According to classical special relativity, causality can act forwards
120: in time, but it does not act nonlocally over intervals that are
121: spatially separated. An event can affect a future event, in or on its
122: forward light cone, but not a spatially separated event. Consequently
123: signals cannot be sent faster than light.
124:
125: But apparently, according to quantum theory, classical events that are
126: linked by quantum systems are different. For them, there is a sense
127: in which causality might act nonlocally, but without any signalling
128: faster than light. This is Bell's weak nonlocality, which can be
129: formulated in terms of the inputs and outputs of black boxes.
130:
131: John Bell in `Against Measurement' [5], discussed the possibility of
132: an exact formulation of some serious part of quantum mechanics: ``By
133: `serious' I mean that some substantial fraction of physics should be
134: covered.\dots I mean too, by `serious' that `apparatus' should not be
135: separated out from the rest of the world into black boxes, as if it
136: were not made of atoms and not ruled by quantum mechanics'
137: Nevertheless, it helps to analyse an experiment to test Bell
138: inequalities as a black box containing classical parts and a quantum
139: system.
140:
141: An electrical engineer's black box consists of a circuit with input
142: and output terminals. He may not know what circuit is inside, but it
143: is assumed here to be classical. If there is no noise in the circuit,
144: then the black box is deterministic. The outputs $j$ then depend on the
145: inputs $i$ through a unique transfer function $F$, where
146: $$
147: j = F(i)
148: $$
149: and by experimenting
150: with different inputs and looking at the outputs, engineers can find
151: $F$.
152:
153: In practice the resistors in the circuit produce noise, which we
154: assume to be classical noise. The system is then stochastic. The
155: noisy circuit can be represented by a probability distribution
156: $\Pr(F)$ over the transfer functions $F$, in which the unknown values
157: of supposedly classical background variables, like the coordinates of
158: thermal electrons, determine the particular $F$ that operates.
159:
160: A physicist's black box contains an evolving physical system, such
161: as a classical electrical circuit, or an entangled quantum state with
162: classical inputs and outputs. She may not know what physical system
163: is inside, but by experimenting with different inputs and looking at
164: the outputs, she can find out something about it.
165:
166: A Bell experiment is an example of a black box with classical
167: terminals and an entangled quantum system inside. We suppose that the
168: source of entanglement is inside the black box, not an input. For
169: photon polarization the setting of the orientations of the polarizers
170: is an input, and the detection of the directions of polarization is an
171: output. All the inputs and outputs are classical events.
172:
173: If we ignore backward causality, special relativity distinguishes
174: between two types of deterministic system, those with local transfer
175: functions $F$ for which the influence of an input on an output goes at
176: no more than the velocity of light, and those with nonlocal transfer
177: functions, for which the influences can act over spacelike
178: intervals. It is possible to determine whether the transfer function
179: of a system is local or not by experimenting with different values of
180: the inputs, and observing the outputs. There is no need to look
181: inside the black box. All classical systems have local transfer
182: functions, as required by special relativity.
183:
184: %?? I need to look at GHZ systems. They appear to be deterministic.
185: % If they are deterministic and nonlocal they should be mentioned
186: %explicitly.
187:
188: When classical or quantum systems are stochastic, and the inputs are
189: given, the probabilities of the outputs can be obtained from a
190: probability distribution $\Pr(F)$. The stochastic systems are of
191: three main types, with different locality properties.
192:
193: In the first type only local $F$ contribute. It is therefore not
194: possible to send signals faster than the velocity of light.
195: For the second type, which may soon be seen, transition probabilities
196: can only be obtained from $\Pr(F)$ in which at least one nonlocal
197: transfer function has nonzero probability, so there is an element of
198: nonlocality. But nevertheless it is not possible to send signals
199: faster than the velocity of light. The system is then weakly
200: nonlocal, or nonlocal in the sense of Bell. The {\it definition}
201: of weak nonlocality needs no quantum theory.
202: For the third type, which we never expect to see, it is possible to send
203: signals faster than the velocity of light.
204:
205: The stochasticity of classical systems comes from background variables
206: that are not included in the system, but for quantum systems it does
207: not come from any background variables that we can see, so either they
208: are assumed not to exist, as in the Copenhagen interpretation, or they
209: are called hidden variables.
210:
211: In a Bell experiment in which the entangled quantum system is
212: sufficiently close to a pure state, and the measurements sufficiently
213: good, the black box is weakly nonlocal in the sense of Bell. An
214: experimenter who has never seen the apparatus before can then tell by
215: experimenting with the inputs and outputs, and without looking inside,
216: that the black box contains a quantum system. This property of black
217: boxes containing quantum systems comes from weak nonlocality.
218:
219:
220: %So quantum black boxes have a dual role, illustrated by the classic
221: %Bell experiment. This was originally proposed by Bell as an
222: %experiment to test whether local hidden variable theories are
223: %possible. But we see that
224:
225: So these black boxes tell us something about the world: there are
226: correlations between classical events that can only be produced by
227: quantum links. These correlations are important in their own right.
228: They are weakly nonlocal. They also show that the properties of our
229: world cannot be explained using local hidden variables, but that is
230: not their main significance.
231:
232: So weak nonlocality is important for all physicists, whether they are
233: interested in hidden variable theories or not. Weak nonlocality is
234: unique in modern physics: classical dynamics, quantum dynamics and
235: classical general relativity are all local. Nonlocality only occurs
236: in some of those processes for which quantum states influence
237: classical events. Laboratory quantum measurement is such a process,
238: but it is not the only one.
239:
240: Today only some ideal experiments involving quantum measurement are
241: nonlocal in any sense, though there may soon be real experiments.
242:
243: \vs
244: {\bf The nature of Bell experiments}
245:
246: There is a profound distinction between experiments to test weak
247: nonlocality by the violation of Bell inequalities and most other
248: experiments with quantum systems.
249:
250: Consider for example an experiment to determine the spectrum of an
251: atom, or a differential cross section for the scattering of an
252: electron by a molecule, or an experiment to determine the band gaps of
253: a solid, or to find a new particle. The aim of all these experiments
254: is to determine the properties of quantum systems. The classical
255: apparatus used to prepare the system and to make the necessary
256: measurements is essential, but secondary to obtaining these properties.
257:
258: In Bell experiments the converse is true. The aim is to test for
259: violation of the inequalities, which are derived from the
260: (statistical) properties of classical events, such as the setting of
261: the apparatus, which is a classical input, or the detection of a
262: particle by an electron avalanche, which is a classical output. The
263: probabilities of the outputs, given the inputs, are what appear in the
264: Bell inequalities, and it is the location of these events in spacetime
265: that determine the locality or nonlocality.
266:
267: These classical events are connected by an ancillary quantum system,
268: whose function is to produce their unusual statistical properties.
269: The quantum properties, like the entanglement of particles, or the
270: polarization of photons, or the spins of atoms, are essential, but
271: secondary. The primary result is the violation of an inequality.
272: The apparatus as well as the quantum system is essentially involved.
273:
274: %which depends on the statistical properties of the classical outputs,
275: %given the classical inputs.
276:
277: %paper - This property is shared with quantum computation, in which the main purpose is to obtain the solution of a problem from a classical output from a formation of the problem as a classical input.
278:
279: This distinction has implications for the analysis of Bell
280: experiments. Real Bell experiments are designed to approximate ideal
281: experiments. But the classical events in a real experiment are
282: usually different from those in the ideal experiment which it simulates.
283: There are generally more types of possible output.
284: For example, in an ideal experiment, it is usually assumed that the
285: detectors detect every particle, but in real experiments they
286: don't.
287:
288: It follows that the inequalities of the ideal experiment do not
289: always apply directly to the real experiment, and further assumptions
290: are needed to demonstrate weak nonlocality.
291:
292: There are two ways to tackle this problem.
293:
294: One is to perform those special experiments for which the ideal and
295: the real are the same. In these the detectors do not distinguish
296: between some of the particle quantum states and a failure to detect a
297: particle.
298: Experiments to test the Clauser-Horne-Shimony-Holt inequality
299: is of this type, so this inequality has been widely used.
300:
301: Another way is to recognize that every real experiment that simulates
302: an ideal Bell experiment has its own critical inequalities, that apply
303: directly to the probabilities for all the outputs of the real
304: experiment.
305: These can be derived from the condition that all the
306: $\Pr(F)$ are local. The larger the number of detectors, the larger
307: the number of inequalities, and a computer program may be needed to
308: obtain them.
309:
310: %One such program is to be made available on December 1 on the website
311: %http://www.strings.ph.qmw.ac.uk/QI/main.htm
312:
313: If any one of these inequalities is violated, weak nonlocality has
314: been demonstrated, and no further assumptions are needed.
315:
316:
317: \vs
318: {\bf Measurement}
319:
320: was an unspeakable word for John Bell:
321:
322: John Bell: Against `measurement' {\it Physics World} 33-40 Aug 1990,
323: p34:
324: `` When I say that the word `measurement' is worse than the others
325: \dots I do have in mind its use in the fundamental interpretive rules
326: of quantum mechanics. \dots
327: `` The first charge against `measurement', in the fundamental axioms
328: of quantum mechanics, is that it anchors there the shifty split
329: of the world into `system' and `apparatus'. ''
330: I will not be discussing this charge. It does not apply to explicit
331: dynamical models of physical processes of which quantum measurement is
332: an example, and these are discussed by Gian-Carlo Ghirardi at this
333: meeting.
334:
335: `` A second charge is that the word comes loaded with with meaning
336: from everyday life, meaning which is entirely inappropriate
337: in the quantum context. When it is said that something is
338: `measured' it is difficult not to think of the result as referring
339: to some preexisting property of the object in question.
340: This is to disregard Bohr's insistence that in quantum phenomena
341: the apparatus as well as the system is essentially involved. ''
342:
343: This charge was avoided above by treating the quantum system and
344: classical apparatus together as a single system.
345:
346: `` In the beginning natural philosophers tried to understand the
347: world around them.\dots Experimental science was born. But
348: experiment is a tool. The aim remains: to understand the world.
349: To restrict quantum mechanics to be exclusively about piddling
350: laboratory operations is to betray the great enterprise. A serious
351: formulation [of quantum mechanics] will not exclude the big
352: world outside the laboratory. ''
353:
354: I will discuss this third charge now.
355:
356: \vs
357: {\bf The big world}
358:
359: Traditionally quantum measurements take place in the laboratory, but
360: the laboratory is only part of our universe, and all such measurements
361: start out as imitations of natural phenomena.
362:
363: Cloud chambers were based on the physics of clouds, which are natural
364: detectors of charged particles. Spark chambers imitate lightning. We
365: can generalize {\it quantum measurement} to mean any process whereby
366: the state of a quantum system influences the value of a classical
367: variable. This definition then applies to the big world.
368:
369:
370:
371: %Does the influence follow unique dynamical laws, which could be
372: %statistical laws?
373:
374: Laboratory quantum measurements include particle states producing the
375: droplets in cloud chambers, bubbles in bubble chambers and sparks in
376: spark chambers.
377: They include photon states producing silver grains in photographic
378: emulsions, and also electron avalanches in solid state detectors and
379: photomultipliers.
380:
381: Other quantum measurements include photon states sending
382: impulses through the optic nerves of owls, the states of cosmic rays
383: that produced small but very long-lived dislocations in mineral
384: crystals in the Jurassic era, and the quantum fluctuations that are
385: believed to have caused today's anisotropies in the universal
386: background radiation and in galactic clusters.
387:
388: This takeover of the physics of laboratory quantum measurement into
389: the world outside the laboratory is here generalized, and one of the
390: questions we have to ask is how far this generalization can go.
391:
392: \vs
393: {\bf Equilibrium gases}
394:
395: Laboratory systems used for quantum measurement are very complicated
396: physical systems, even stripped down to their bare essentials.
397: They involve amplification in one form or another, and so do the
398: natural systems that they imitate.
399:
400: A gas in equilibrium is much simpler, yet generalized quantum
401: measurement takes place there also.
402: The reason is that the motion of the molecules in the gas is chaotic,
403: and small changes now result in large changes later. In particular
404: changes at the quantum level now produce significant classical
405: fluctuations in the density later. However, unlike earlier examples,
406: we can't use the classical density fluctuations to learn anything
407: specific about these earlier quantum states, because the chaos causes
408: mixing, which effectively obscures the signal.
409:
410: In the nineteenth century, Rayleigh recognized that these
411: classical density fluctuations would scatter light, and
412: that the scattering was strongly dependent on the wavelength of
413: the light. The result is the blue of the sky.
414:
415: The growth of droplets of water around the charged particles
416: produced by cosmic rays in the atmosphere is a quantum measurement.
417: So are the density fluctuations in the atmosphere that cause
418: the sky to be blue where there are no clouds.
419: So if you ever look at the sky, as every physicist sometimes should,
420: whether it is clear or overcast, you are seeing one example or another
421: of quantum measurement.
422:
423: \vs
424: {\bf Theoretical cosmology makes a prediction}
425:
426:
427: %The most successful theory of the early universe is the inflation
428: %theory.
429:
430: %In this model, the universe at about the Planck time of $10^{-43}\rm s$ was unstable, and the observable universe grew in radius exponentially, just as the displacement of a vertical pendulum at the upward vertical position grows exponentially.
431:
432: %Any fluctuations before inflation would be damped so much that
433: %they have nothing to do with anything we see now.
434:
435: % 10.21 Linde p190
436:
437: % 10.22 Linde p180
438:
439:
440: %{\ bf The end of inflation}
441:
442: %Although inflation seemed a panacea for all of cosmology's ills,
443: %there was still an important unresolved issue: how exactly does
444: %inflation end?
445:
446: %Would the uniformity produced by inflation survive the end of
447: %inflation?
448: According to the cosmologists there occurs in the early universe a
449: generalized quantum measurement. Quantum fluctuations produce
450: fluctuations in classical variables and these are then rapidly
451: stretched from subatomic scales to the size of galaxies or even
452: larger. So according to this theory, the inhomogeneities in the
453: universe, such as galactic clusters, galaxies, and the fluctuations in
454: the background radiation are all due to quantum measurement in the
455: early universe. Somewhat later, physicists meeting in Vienna were
456: also produced by fluctuations from the same process of quantum
457: measurement.
458:
459: Recently there have been several detailed observations of the
460: inhomogeneities\hs1 in the universal background radiation,\hs1 for example
461: those of the `Boomerang' project [6]. The intensity of the fluctuations
462: were plotted as a function of the order of the spherical harmonics,
463: as illustrated in the figure. There is a pronounced peak.
464:
465: \begin{figure}[htb]
466: \epsfxsize14.0cm
467: \centerline{\epsfbox{fig.ps} }
468: \label{fig}
469:
470: \caption{Theory (continuous line) and experiment for inhomogeneities
471: in the universal background radiation}
472: \end{figure}
473:
474: The observations are compared with theory, represented by the
475: continuous curve, and based on the assumption of a flat universe.
476: The theory shows the same peak, and a number of smaller peaks
477: which are not (yet) seen. By the standards of cosmology, the
478: agreement between theory and observation is excellent, providing
479: evidence for quantum measurement in the early universe.
480:
481: \vs
482: {\bf Conclusions}
483:
484: One of John Bell's major objections to quantum `measurement' might be
485: overcome by generalizing the definition to include processes in the
486: big world. With this extended definition, quantum measurement is a
487: universal property of physical systems. It is all around us, and we
488: would not be here without it.
489:
490: Consequently the dynamics of quantum measurement has universal
491: significance. So have its properties, like the weak nonlocality
492: of John Bell.
493:
494: \vs
495: {\bf Acknowledgements} My thanks go to the ESI for their hospitality and
496: support, and to Mary Bell, Reinhold Bertlmann, Abner Shimony and Dave
497: Sutherland for amendments and corrections.
498:
499:
500: \vs\vs
501: {\bf References}
502:
503: [1] Lucretius, Nature of the Universe (Trans. R.Latham, Penguin, Harmondsworth 1951) p53-54.
504:
505: [2] A. Einstein, B. Podolsky and N. Rosen. Can quantum-mechanical
506: description of reality be considered complete? {\it Phys. Rev.} {\bf
507: 47}, (1935) 777-780.
508:
509: [3] A. Einstein, Albert Einstein, Philosopher Scientist
510: (ed. Schilpp, Cambridge University Press, Cambridge, UK 1959) p85.
511:
512: [4] J. S. Bell. On the Einstein Podolsky Rosen paradox {\it Physics}
513: {\bf 1} (1964) 195-200.
514:
515: [5] J. S. Bell. Against `Measurement' {\it Physics World}, August
516: (1990) 33-35.
517:
518: [6] De Bernadis et al. A flat universe from high-resolution maps of
519: the cosmic microwave background radiation. {\it Nature}
520: {\bf 404} (2000) 955-959.
521:
522: \end{document}
523:
524: \end
525:
526:
527:
528:
529:
530:
531:
532:
533:
534:
535: