1: %%% LaTeX file for the proceedings: qm2003
2: %%% 21 November 2003
3: %%% Izumi Tsutsui and Tamas Fulop
4:
5: \documentclass{ws-ijqi}
6:
7: \begin{document}
8:
9: \markboth{Izumi Tsutsui}
10: {Physics of Points and Walls in Quantum Mechanics}
11:
12: \catchline{}{}{}{}{}
13:
14: \title{PHYSICS OF SINGULAR POINTS IN QUANTUM MECHANICS}
15:
16: \author{IZUMI TSUTSUI${}^*$\,\, and \,\, TAM\'{A}S F\"{U}L\"{O}P${}^{\dagger}$}
17: \address{Institute of Particle and Nuclear Studies\\
18: KEK, Tsukuba 305-0801, Japan\\
19: ${}^*$izumi.tsutsui@kek.jp, \,${}^{\dagger}$fulopt@post.kek.jp}
20:
21: \maketitle
22:
23: \begin{history}
24: \received{(21 November 2003)}
25: % \revised{(DAY MONTH YEAR)}
26: \end{history}
27:
28: \begin{abstract}
29: Defects or junctions in materials serve as a source of interactions for particles, and in idealized limits they may be treated as
30: singular points yielding contact interactions. In quantum mechanics, these singularities accommodate an unexpectedly rich structure and thereby provide a variety of physical phenomena, especially if their properties are controlled properly. Based on our recent studies, we present a brief review on the physical aspects of such quantum singularities
31: in one dimension. Among the intriguing phenomena that the singularities admit, we mention strong vs weak duality, supersymmetry, quantum
32: anholonomy (Berry phase), and a copying process by anomalous caustics. We also show that a partition wall as a singularity in a potential well can give rise to a quantum force which exhibits an interesting temperature behavior characteristic to the particle statistics.
33:
34: \end{abstract}
35:
36: \keywords{quantum singularity; duality; anholonomy; quantum force.}
37:
38: \section{Introduction}
39:
40: If a system contains an object (or subsystem) different from its surroundings, and if the object is very small compared to the size of the system, then one may regard it as a \lq singularity\rq{}
41: in the system to a first approximation. Such an object may be given by an isolated region forming a dot, or it may consist of a planar region forming a wall in the system. In view of the negligible size of the object, we may study the physical property of the system by considering a point singularity interacting with particles such as electrons by contact interaction. The outcome will furnish a basis for studying the physics of systems with singular objects given by quantum dots or
42: junctions in semiconductors, where the finite size effect may be considered as secondary to the zero size effect. This analysis is also useful for analyzing systems from a long range point of view, where the singular object is reduced to a point (or a plane without thickness) in effect.
43:
44: In classical mechanics, a singular
45: point will
46: have no characteristics and is basically trivial. In
47: quantum mechanics, in contrast, it has been known that there are many, {\it distinct} singular points allowed, and if the system is one dimensional ({\it i.e.}, line) they form a $U(2)$ family~\cite{RS,AGHH}.
48: The distinction among them lies in the connection conditions at the singularity
49: obeyed by the wave functions, which can lead to entirely different physical consequences depending on the choice of the singularity. Our interest then is to see how these singular points can be characterized mathematically, and to find what type of physical phenomena
50: can be expected in systems with particles interacting with the $U(2)$ family of singularities if they are manipulated appropriately.
51: The aim of the present paper is to provide a review of our investigations on these matters,
52: and thereby to point out that, once a controllable singularity is introduced in the otherwise free system on a line, then
53: there appear many physically interesting properties, such as
54: duality, anholonomy (Berry phase), supersymmetry and a copying process by quantum anomalous caustics. We also mention a statistical
55: aspect of systems with a singularity, which is the emergence of a force
56: on a partition wall (regarded as singularity).
57: The force is generated by distinct boundary conditions and exhibits an interesting dependence --- characteristic to the particle statistics (bosons or fermions) --- on the temperature and the particle number of the system.
58: Our consideration is restricted only to one dimensional systems, but the essential features of quantum singularities found here will also persist in higher dimensional systems.
59:
60: This paper is organized as follows. After the Introduction, we briefly recall in section 2 how the $U(2)$ family of singularities appear in quantum mechanics on a line. The physical meaning of the matrix notation used to characterize the family of singularities is then discussed in section 3. Sections 4 and 5 are devoted to the physical phenomena afforded under the presence of the singularities mentioned above. Section 6 discusses the quantum force on a partition wall and its temperature behavior. Finally,
61: we give our discussions in section 7.
62:
63:
64: \section{Quantum Description of a Singular Point on a Line}
65:
66: Suppose that there is a point singularity, say, at
67: $x = 0$ on a line $-\infty < x < \infty$. If the singularity is not accompanied by a potential, then
68: our Hamiltonian is the free one,
69: \begin{equation}
70: H =
71: -{{\hbar^2}\over{2m}}{{d^2}\over{dx^2}} ,
72: \label{eqn:ham}
73: \end{equation}
74:
75: and the singularity will just impose a certain connection condition for wave functions at $x = 0$. The connection condition arises from the requirement of the unitarity
76: of the system,\footnote{Mathematically, this is equivalent to the self-adjointness of the Hamiltonian operator $H$ under the presence of the singularity.}
77: which is ensured if the probability current
78: $
79: j(x) = - {{i\hbar}\over{2m}}\left(
80: (\psi^*)'\psi - \psi^* \psi' \right)(x)
81: $
82: is continuous at the singularity,
83: $j(+0)=j(-0)$.
84: This can be shown to be equivalent to the connection condition~\cite{AG,FT}
85: \begin{equation}
86: (U-I)\Psi+iL_0(U+I)\Psi'=0,
87: \label{eqn:bcon}
88: \end{equation}
89: where $U \in U(2)$ is called {\it characteristic matrix},
90: $L_0 \neq 0$ is a real (arbitrary) constant
91: and
92: \begin{equation}
93: \Psi =
94: \left(
95: \begin{array}{c}
96: {\psi (+0)}\\
97: {\psi (-0)}
98: \end{array}
99: \right),
100: \quad
101: \Psi' =
102: \left(
103: \begin{array}{c}
104: {\psi' (+0)}\\
105: {-\psi' (-0)}
106: \end{array}
107: \right),
108: \label{eqn:bvec}
109: \end{equation}
110: are vectors defined from the boundary values of the wave function
111: $\psi(\pm 0) = \lim_{x \to \pm 0} \psi(x)$ and their derivatives.
112: In other words, imposing the probability conservation as the sole requirement
113: in quantum mechanics, we find the $U(2)$
114: freedom in specifying the connection
115: condition, that is, there exist the $U(2)$ family of
116: singularities possessing different connection conditions on a line. For instance, if we choose $U = \sigma_1$ (where
117: $\sigma_i$ are Pauli matrices), the connection condition (\ref{eqn:bcon}) reduces
118: to
119: $\psi(+0) = \psi(-0)$,
120: $\psi'(+0) = \psi'(-0)$, which represents
121: the \lq free system\rq{} --- no actual
122: singularity at all. On the other hand, the choice $U = -I$ gives the Dirichlet condition
123: $\psi(+0) = \psi(-0) = 0$ while $U = I$ gives the Neumann condition
124: $\psi'(+0) = \psi'(-0) = 0$.
125:
126: The last two choices provide two examples of
127: connection conditions that prohibit the probability
128: flow at the singularity.
129: All of these conditions represent, physically,
130: an \lq infinite\rq{} partition wall with distinct characters.
131: The general singularities that do not allow the probability flow are obtained by requiring $j(+0)=j(-0) = 0$, and the characteristic matrices that meet
132: this requirement are given by
133: diagonal $U \in U(2)$.
134: These form the so-called \lq separated subfamily\rq{}
135: $U(1) \times U(1)$ in the $U(2)$ family,
136: for which the connection condition (\ref{eqn:bcon})
137: reduces to the boundary condition,
138: \begin{equation}
139: \psi(+0)+L(\theta_+)\psi'(+0) = 0,
140: \qquad
141: \psi(-0)+L(\theta_-)\psi'(-0) = 0,
142: \label{eqn:redbc}
143: \end{equation}
144: where
145: \begin{equation}
146: L(\theta_\pm) = L_0 \cot{{\theta_\pm}\over 2},
147: \label{eqn:spcprter}
148: \end{equation}
149: with $\theta_\pm$ being the phase parameters of ${\rm diag}\, U = (e^{i\theta_+}, e^{i\theta_-})$.
150: The Dirichlet and the Neumann conditions are obtained, respectively, by choosing
151: $L(\theta_\pm) = 0$ and $L(\theta_\pm) = \infty$. In passing we note that,
152: if the singularity is a wall (an end point of a positive half line), then the condition is simply $\psi(+0)+L(\theta_+)\psi'(+0) = 0$.
153: An important point to be noted is that neither wave functions nor their derivatives are continuous at the singularity under a generic singularity (dot, partition or wall).
154:
155: The forgoing argument applies almost unchanged to cases where
156: the singularity arises as a divergent point of a potential, such as the Coulomb potential $V(x) = c/|x|$. The only technical modification necessary is that the boundary vectors used in the connection condition (\ref{eqn:bcon}) must be slightly generalized as~\cite{TCF}
157: \begin{equation}
158: \Psi=
159: \left(
160: \begin{array}{c}
161: {W[\psi,\varphi_{1}]_{+0}}\\
162: {W[\psi,\varphi_{1}]_{-0}}
163: \end{array}
164: \right),
165: \quad
166: \Psi'=
167: \left(
168: \begin{array}{c}
169: W[\psi,\varphi_{2}]_{+0}\\
170: - W[\psi,\varphi_{2}]_{-0}
171: \end{array}
172: \right)
173: \label{eqn:gbvec}
174: \end{equation}
175: with the help of reference states $\varphi_{1}$, $\varphi_{2}$ which are arbitrarily chosen to provide the self-dual
176: Wronskians
177: $
178: W[\phi,\psi](x) = \phi(x)\psi'(x)
179: -\psi(x)\phi'(x)
180: $.
181: The modification is required because wave functions (and/or their derivatives), and hence the boundary vectors (\ref{eqn:bcon}), may diverge at the singularity under diverging potentials, while the Wronskians are always well-defined. We note that (\ref{eqn:gbvec}) is a generalization of (\ref{eqn:bvec}), since by a suitable choice of the reference states one can regain (\ref{eqn:bvec}) from (\ref{eqn:gbvec}) when the states $\psi$ are well-defined at the singularity.
182:
183:
184: \section{Characteristic Matrix and the Spectral Space}
185:
186: In order to investigate the physics implied by the singularity, we introduce the parametrization~\cite{TFC,CFT} of the characteristic matrix,
187: \begin{equation}
188: U = V^{-1} D V,
189: \label{eqn:ude}
190: \end{equation}
191: with
192: \begin{equation}
193: D = \left(
194: \begin{array}{cc}
195: e^{i\theta_+} & 0 \\
196: 0 & e^{i\theta_-}
197: \end{array}
198: \right),
199: \qquad
200: V = e^{i{\mu\over 2}\sigma_2} e^{i{\nu\over 2}\sigma_3}
201: \label{eqn:udec}
202: \end{equation}
203: where $\theta_\pm \in [0, 2\pi)$ and
204: $\mu \in [0, \pi]$, $\nu \in [0, 2\pi)$.
205: The convenience of the parametrization may be recognized by the following observations. Notice, first, that any eigenstate $ \psi(x)$ with energy $E$ remains to be an eigenstate after the transformation of the parity ${\cal P}$ or the half-reflection ${\cal R}$ defined by
206: \begin{equation}
207: {\cal P}:
208: \psi \rightarrow
209: ({\cal P}\psi)(x) := \psi( - x),
210: \qquad
211: {\cal R}:
212: \psi \rightarrow
213: ({\cal R}\psi)(x) := [\Theta(x) -
214: \Theta(-x)]\psi(x).
215: \label{eqn:dtrsf}
216: \end{equation}
217: The only effect caused by these can be found in the change of the connection
218: condition that the eigenstate obeys, and these are described by the corresponding change in the characteristic matrix,
219: \begin{equation}
220: U \buildrel {{\cal P} } \over
221: \longrightarrow \sigma_1\, U\, \sigma_1, \qquad
222: U \buildrel {{\cal R} } \over
223: \longrightarrow \sigma_3\, U\, \sigma_3.
224: \label{eqn:utrsf}
225: \end{equation}
226: Obviously, the same can be observed with the transformation generated by the product ${\cal Q} = i{\cal P}{\cal R}$, which implies that this remains so even under any transformation given by a linear combination of the three generators $\{{\cal P} , {\cal Q}, {\cal R}\}$ which form an $su(2)$ algebra, as long as it does not change the norm of the state. These general isospectral transformations induce conjugations to the matrix $U$ by
227: $U \rightarrow W^{-1} U W$ with $W \in SU(2)$.
228:
229:
230: \begin{figure}[t]
231: \begin{center}
232: \includegraphics[width=0.8\linewidth]{f1.eps}
233: \caption{$U(2)$ parameter space as a product of the torus $T^2$ determining the spectrum (the spectral space is its half $T^2/Z_2$ which is a M{\"o}bius strip) and the isospectral sphere $S^2$. }
234: \label{moebius}
235: \end{center}
236: \end{figure}
237:
238:
239: Thus we learn that the decomposition (\ref{eqn:ude}) with (\ref{eqn:udec}) provides a split in the parameters into those that determine the sepctrum and those that do not, that is, $(\theta_+, \theta_-)$ and $(\mu, \nu)$. The spaces of these parameters are, therefore, given by the spectral torus $T^2$ and the isospectral sphere $S^2$ (see Figure 1), respectively.
240: More precisely, we shall see soon that the spectrum is unchanged under the interchange of the parameters $\theta_+ \leftrightarrow \theta_-$, and consequently the actual spectral space is given by $T^2/Z_2$ which is a M{\"o}bius strip with boundary~\cite{Moebius}.
241:
242: The combinations $L(\theta_\pm)$
243: given in (\ref{eqn:spcprter}) set the scale of the system and turn out to be more useful than
244: $\theta_\pm$. {}For instance, for
245: $L(\theta_+) > 0$ and/or $L(\theta_-) > 0$, the singularity can support the bound states
246: $\psi_\pm (x) \propto e^{- \vert x\vert/L(\theta_\pm)}$ where $L(\theta_\pm)$ represent the effective range of the particle trapped around the singularity.
247: The isospectral parameters $(\mu, \nu)$, on the other hand, are related to the phase shift of the wave function at the singularity and the degree
248: of mixture of the limiting values of the wave function at $x = \pm 0$. The
249: characterization of the parameters discussed here hold true even under a symmetric potential, $V(-x) = V(x)$.
250:
251:
252: \section{Duality, Anholonomy and Supersymmetry}
253:
254:
255: Having furnished a formal basis to describe a generic quantum singularity on a line, we now turn to its physics. Due to the nontrivial structure of the parameter space $U(2)$, various interesting phenomena can arise if we manipulate the parameters properly on the $U(2)$ space. Here we mention three of them, duality, anholonomy and supersymmetry, which can be readily realized from what we have already.
256:
257: \subsection{Duality}
258:
259:
260: The invariance of the spectrum under the interchange
261: $\theta_+ \leftrightarrow \theta_-$ implies spectral duality for a
262: pair of systems possessing singularities with the two parameters
263: interchanged.
264: To see how this happens, for simplicity we restrict ourselves to parity
265: invariant singularities. Note that, since the parity transformation ${\cal P}$ induces the change
266: (\ref{eqn:utrsf}), parity invariant singularities are characterized by those $U$ satisfying
267: $\sigma_1\, U\, \sigma_1 = U$. The general solution is given by
268: \begin{equation}
269: U = U(\theta_+, \theta_-)
270: = e^{i(\theta_+P^+_1 + \theta_-P^-_1)},
271: \label{eqn:piu}
272: \end{equation}
273: with
274: $P^\pm_1 = {{1 \pm\sigma_1}\over 2}$,
275: which is obtained by setting the ispospectral parameters
276: $(\mu, \nu) = (\pi/2, 0)$ in (\ref{eqn:udec}).
277:
278:
279: We also note that, to these parity invariant $U$ in (\ref{eqn:piu}) the half reflection
280: ${\cal R}$ induces the exchange $\theta_+ \leftrightarrow
281: \theta_-$ through (\ref{eqn:utrsf}). In view of the fact that the spectrum is preserved under ${\cal R}$, we realize that if the parameters $(\theta_+, \theta_-)$ of the two systems are the opposite of each other, then they share the same energy spectrum. Moreover, since any eigenstate that arises under a parity invariant
282: singularity can either be parity symmetric $\psi(-x) = \psi(x)$ or antisymmetric $\psi(-x) = -\psi(x)$, and since ${\cal R}$ swaps the parity, the corresponding eigenstates with the same energy in the two systems must have the opposite parity. Note that ${\cal R}$ is an identity operation for the special type of singularities defined by $\theta_+ = \theta_-$, which are called \lq self-dual\rq. It follows that in the self-dual case, which is indicated by the loop $S$ in Figure 1, the entire spectrum consists of doubly (or evenly) degenerate levels consisting of pairs of parity symmetric and antisymmetric states (see Figure 2).
283:
284:
285: \begin{figure}[t]
286: \begin{center}
287: \includegraphics[width=0.7\linewidth]{f2.eps}
288: \caption{Spectra of two systems with a singularity placed at the centre of an infinite well, where $k$ is the momentum with the energy $E = \hbar k^2/(2m)$. The spectra,
289: drawn as a function of $x$ on the torus, are
290: identical for $x$ and $-x$, that is, for $(\theta_+, \theta_-)$ and $(\theta_-, \theta_+)$.}
291: \label{duality}
292: \end{center}
293: \end{figure}
294:
295:
296: Now we observe that, in this parity
297: invariant subfamily, the free system $U = \sigma_1$ arises
298: at $(\theta_+, \theta_-) = (0, \pi)$. This suggests
299: that we may consider \lq coupling constants\rq{} measuring the strengths of the
300: interaction at the singularity by
301: \begin{equation}
302: g_+(\theta_+) := \tan {\theta_+\over 2}\ ,
303: \qquad
304: g_-(\theta_-) := \cot {\theta_-\over 2}\ ,
305: \label{eqn:cconst}
306: \end{equation}
307: which vanish, $g_+(0) = g_-(\pi) = 0$, at the free point.
308: In terms of these, we can interpret that the spectral duality holds for two systems with different
309: coupling constants.
310: In particular, if the parameters
311: fulfill
312: $\theta_+ = \theta_- \pm \pi$, then we find the reciprocal behavior
313: \begin{equation}
314: (g_+(\theta_+), g_-(\theta_-))
315: \buildrel {{\cal R} } \over
316: \longrightarrow
317: (- 1/g_+(\theta_+), -1/g_-(\theta_-)).
318: \label{eqn:cduality}
319: \end{equation}
320: In this case, the spectral duality can occur between two systems, one with a strong coupling and the other with a weak coupling. This shows that the quantum singularity furnishes a simple example of {\it strong vs weak coupling duality}~\cite{TFC,CFT} which is normally discussed for more complicated systems such as (supersymmetric) gauge theory.
321:
322:
323: \subsection{Anholonomy}
324:
325: Next we turn to the opposite situation where the spectral
326: parameters $(\theta_+, \theta_-)$ are fixed whereas
327: the isospectral parameters $(\mu, \nu)$ are free to vary.
328: If we choose the free case $(\theta_+, \theta_-) = (0, \pi)$ for the fixed point, the collection of such singularities provides
329: the scale invariant subfamily, that is, they are invariant under the
330: Weyl scale transformation,
331: \begin{equation}
332: {\cal W}_\lambda: \quad
333: \psi(x) \longrightarrow
334: ({\cal W}_\lambda\psi)(x) :=
335: \lambda^{1\over 2} \psi(\lambda x),
336: \label{eqn:weyltr}
337: \end{equation}
338: for real $\lambda$. For our convenience,
339: we consider a singularity of this kind placed
340: at the centre of in an infinite well $[-l, l]$ and
341: impose the Dirichlet boundary condition at the ends $x = \pm l$.
342: Then the energy eigenstates are found to be
343: \begin{equation}
344: \psi_n(x)
345: = c_+(\mu)\, \xi_n^+(x)
346: +c_-(\mu) e^{i\nu}\, \xi_n^-(x),
347: \label{eqn:egenst}
348: \end{equation}
349: where
350: \begin{equation}
351: c_\pm(\mu)
352: = \cos{\mu\over 2} \mp \sin{\mu\over 2},
353: \qquad
354: \xi_n^\pm(x)
355: = \sqrt{1\over l} \sin k_n(x\mp l) \Theta(\pm x) ,
356: \label{eqn:eeigenst}
357: \end{equation}
358: with
359: $k_n
360: = \left(n-{1\over 2}\right){\pi\over {2l}}$ for $n = 1,2,3,\ldots$.
361:
362:
363: Now, suppose that we have some means to control the isospectral parameters $(\mu, \nu)$ adiabatically. Then we can consider a cyclic process of change along a loop $C$ on the
364: isospectral sphere
365: $S^2$ as shown in Figure 1.
366: After completing the cycle, each eigenstate returns to the initial
367: one modulo a phase pertinent to the state, $\psi_n \rightarrow e^{i\gamma(C)}\psi_n$. This is the {\it phase anholonomy} (or the Berry phase) and is evaluated to be
368: \begin{equation}
369: \gamma(C)
370: = \oint_C A, \qquad
371: A
372: = i\langle \psi_n \vert
373: d \psi_n \rangle
374: = -{1\over 2}(1+\sin\mu)\, d\nu,
375: \label{eqn:bcn}
376: \end{equation}
377: where $d$ is the exterior derivative in the parameter space~\cite{CFT}.
378: Note that the curvature $F = dA$ is just the magnetic field of the Dirac monopole,
379: $
380: F = -{1\over 2}\cos\mu\,d\mu\,d\nu
381: $.
382:
383: \begin{figure}[t]
384: \begin{center}
385: \includegraphics[width=0.7\linewidth]{f3.eps}
386: \caption{Spectral change along the cyclic process paralleling the self-dual loop $S$ distanced by $\pi$. Both symmetric and antisymmetric levels turn to be a lower (or higher depending on the direction) level by two after the cycle, even though the whole spectrum is unchanged. }
387: \label{loops}
388: \end{center}
389: \end{figure}
390:
391:
392:
393: A similar adiabatic cyclic process may also be considered on the spectral torus,
394: instead of the isospectral sphere. For instance, we may change the spectral parameters along a loop which winds over the surface of the torus nontrivially (the simplest will be the loop $\Gamma$ in Figure 1). After
395: completing one cycle, we find a different type of anholonomy, where each level does not return to the initial one, even though the entire spectrum as a whole is recovered. The response of the spectral change depends on the cycle one chooses; for example, if the loop is taken to be the one paralleling the self-dual loop $S$ with distance $\pi$ as shown in Figure 3, then
396: the discrete momenta $k$ corresponding to symmetric states are given by $k = k(\theta_+)$ while those corresponding to antisymmetric states are
397: $k = k(\theta_- \pm \pi)$, where the function $k(\theta)$ is determined by
398: \begin{equation}
399: k(\theta)\, L_0 \cot k(\theta)l = \tan {{\theta}\over 2}.
400: \label{eqn:nahsc}
401: \end{equation}
402:
403: The resultant spectral change in Figure 3 shows the {\it level anholonomy}, as they shift by two after one cycle~\cite{CFT}. This double spiral structure of energy levels along the loop may in future be used to implement a specific physical process like the one considered for the holonomic quantum computation in systems exhibiting the Berry phase.
404:
405:
406: \subsection{Supersymmetry}
407:
408: The double degeneracy occurring under self-dual singularities
409: suggests that these systems may accommodate
410: supersymmetry (SUSY). In fact, one can show that for a certain
411: class of systems, including those with a special type of self-dual singularities, it is possible to associate SUSY without necessarily yielding degeneracy in the spectrum.
412:
413:
414: To see this, let us rewrite our system into a
415: set of two systems each of which defined on a half line.
416: There, we employ, instead of the wave functions $\psi(x)$ and the
417: Hamiltonian $H$ in (\ref{eqn:ham}), the two-component wave functions and the corresponding Hamiltonian
418: \begin{equation}
419: \Psi(x) = \left(
420: \begin{array}{c}
421: {\psi_+(x)}\\
422: {\psi_-(x)}
423: \end{array}
424: \right),
425: \qquad
426: H = -{\hbar^2\over {2m}}\frac{d^2}{dx^2}\otimes I,
427: \label{eqn:wfham}
428: \end{equation}
429: where
430: $\psi_\pm(x) := \psi(\pm x)$ for $x > 0$
431: and $I$ is the $2 \times 2$ identity matrix.
432: Our supercharge is assumed to take the form
433: \begin{equation}
434: Q =
435: -i\lambda\frac{d}{dx}
436: \otimes\sigma_{\vec{a}}+
437: {\bf 1} \otimes
438: \sigma_{\vec{b}},
439: \label{eqn:sscharge}
440: \end{equation}
441: where
442: $\lambda = \hbar/2\sqrt{m}$
443: and
444: \begin{equation}
445: \sigma_{\vec{a}} = \sum_{i = 1}^3{a_i \sigma_i},
446: \quad
447: \sigma_{\vec{b}} = \sum_{i = 1}^3{b_i \sigma_i},
448: \quad
449: \vert \vec{a}\vert = 1 ,
450: \quad
451: \vec{a}\cdot \vec{b} = 0,
452: \label{eqn:ssche}
453: \end{equation}
454: with real vectors $\vec{a}$, $\vec{b}$. The conditions (\ref{eqn:ssche})
455: ensure the formal relation $2 Q^2 = H+\vert {\vec b}
456: \vert^2$. Thus, if we absorb the constant $\vert {\vec b} \vert^2$ into
457: the Hamiltonian (which causes only the corresponding constant energy shift),
458: we obtain, for a set of independent supercharges $Q_i$ for $i = 1,\ldots,N$ which are normalized properly, the standard SUSY
459: algebra,
460: \begin{equation}
461: \{Q_i, Q_j\} = H\, \delta_{ij}.
462: \label{eqn:ssalg}
463: \end{equation}
464:
465:
466: The important question we need to address at this point is whether the
467: supercharge $Q$ leaves the given connection condition invariant, at least for
468: energy eigenstates, because otherwise the SUSY transformation acts on states allowed under different singularities and hence the SUSY is not defined within a single system.
469: Thus our demand for SUSY to exist is that, given a singularity specified by $U$, both the state $\Psi(x)$ and $Q \Psi(x)$ fulfill the same connection condition.
470: To answer this question,
471: we first note that, if the state
472: $\Psi(x)$ fulfills the connection condition (\ref{eqn:bcon}), then for any $W \in U(2)$
473: the state $W \Psi(x)$ fulfills the same connection condition with $U$ replaced by $WUW^{-1}$. This implies that,
474: if the pair $(U, Q)$ satisfies the above demand, so does the
475: pair $(WUW^{-1}, WQW^{-1})$. Note also that $WQW^{-1}$ is again in the form (\ref{eqn:sscharge}), and hence by
476: choosing in particular
477: $W = V$ with $V$ appearing in the decomposition (\ref{eqn:ude}), we find that the pair $(D, VQV^{-1})$ also satisfies the demand. For this
478: reason, with no loss of generality, we may assume that
479: $U$ is diagonal. We then find, by a straightforward inspection, that
480: the required condition is fulfilled if $\theta_+ = \theta \neq 0$ and
481: $\theta_- = \pi$ (and vice versa), and further if the supercharge takes the form
482: \begin{equation}
483: Q = V^{-1} \,q(\alpha, c; \theta)\, V,
484: \label{eqn:spcg}
485: \end{equation}
486: with
487: \begin{equation}
488: q(\alpha, c; \theta)
489: ={} -i\lambda\frac{d}{dx}\otimes
490: e^{-i{\alpha\over 2}\sigma_3}\sigma_1 e^{i{\alpha\over 2}\sigma_3}
491: {}+ {\bf 1}
492: \otimes
493: \left[-{\lambda\over{L(\theta)}}
494: e^{-i{\alpha\over 2}\sigma_3}\sigma_2
495: e^{i{\alpha\over 2}\sigma_3}
496: + c\,
497: \sigma_3\right],
498: \label{eqn:spss}
499: \end{equation}
500: where $L(\theta)$ is the scale parameter defined in (\ref{eqn:spcprter}).
501: Since $\alpha$ is arbitrary, there are two independent
502: supercharges, {i.e.,} the system has an $N = 2$ SUSY~\cite{UTone,UTtwo} as long as one of the two eigenvalues of the characteristic matrix $U$ is $-1$ and the other not $-1$.
503:
504:
505: \begin{figure}[t]
506: \begin{center}
507: \includegraphics[width=0.5\linewidth]{f4.eps}
508: \caption{Energy levels of the $N = 1$ SUSY system occurring
509: under the singularity with $U = V^{-1} \sigma_3 V$. The levels, which are dependent on $\mu$ because the boundary conditions break the parity invariance,
510: are not degenerate unless $\mu = 0$ or $\pi$.}
511: \label{susy}
512: \end{center}
513: \end{figure}
514:
515:
516:
517: When we put the singularity in an infinite well, then the SUSY may be found depending on the boundary conditions imposed at the ends. Various combinations, and accordingly various types of SUSY systems arise, and before we proceed we mention one of these. Consider the boundary condition
518: \begin{equation}
519: \psi^\prime_+(l) = 0,
520: \qquad
521: \psi_-(l) = 0,
522: \label{eqn:ndbc}
523: \end{equation}
524: and the singularity specified by $U = V^{-1} \sigma_3 V$ which gives the connection condition
525: \begin{equation}
526: e^{i\nu} \psi_+(+0) - \cot {\mu\over 2}\, \psi_-(+0) = 0,
527: \qquad
528: e^{i\nu} \psi_+^\prime(+0) + \tan {\mu\over 2}\, \psi_-^\prime(+0) = 0.
529: \label{eqn:cbtwo}
530: \end{equation}
531: Then, the system admits an $N = 1$ SUSY with the supercharge $Q$
532: for $\vec{b} = 0$ and the energy eigenstates
533: \begin{equation}
534: \Psi^{(n)}(x) = N^{(n)}
535: \left( \matrix{ - e^{-i\nu} \cos k_n (x-l) \cr
536: \sin k_n (x-l) } \right) ,
537: \qquad
538: k_n = {{n\pi + \mu/2}\over{l}},
539: \label{eqn:eeigen}
540: \end{equation}
541: for $n \in Z$. Each eigenstate is invariant
542: under the SUSY transformation generated by $Q$, and
543: the energy levels $E^{(n)} = \hbar^2 k_n^2/(2m)$
544: are not degenerate unless $\mu = 0$ or $\pi$ (see Figure 4).
545: Supersymmetry, therefore, does not necessarily imply degeneracy in levels.
546:
547:
548:
549: \section{Quantum Tunneling and Copy}
550:
551: When the singularity is accompanied with a potential $V(x)$, we can expect various
552: interesting phenomena by combining the property of the quantum singularity
553: and the property pertinent to the potential. One such example is provided by \begin{equation}
554: V(x) = \frac{m {\omega^2}}{2} {x^2} +
555: g\frac{1}{x^2},
556: \label{eqn:cstpot}
557: \end{equation}
558:
559: which is known to admit \lq caustics\rq~\cite{Schulman}, a phenomena that arises when the classical dynamics of the system exhibits a certain type of singularity in the initial value problem, with the typical example being the periodic recurrence of the harmonic oscillator. In fact, for $g > 0$ the dynamics of the system resembles the harmonic oscillator, and the only essential difference is that the recurrence occurs in each of the half lines because the system splits into two subsystems at
560: the singularity $x = 0$ due to the infinite potential wall there.
561:
562: In quantum mechanics, the situation is quite different.
563: To see this, note first that the general solution
564: for the Schr{\"o}dinger equation
565: $H {\psi_n}(x) = {E_n}{\psi_n}(x)$ (where now $H$ has the potential term $V(x)$)
566: is given by a linear combination of
567: the two independent solutions,
568: \begin{equation}
569: \phi^{(1)}_n(x)
570: :={y^{c_1 - 1/2}}e^{-{y^2}/{2}}
571: F\left(\frac{{c_1}-\lambda_n}2,{c_1};{y^2}\right),
572: \label{eqn:indgensla}
573: \end{equation}
574: and
575:
576: \begin{equation}
577: {\phi^{(2)}_n}(x)
578: :={y^{c_2 - 1/2}}e^{-{y^2}/{2}}
579: F\left(\frac{{c_2}-\lambda_n}2,{c_2};{y^2}\right),
580: \label{eqn:indgenslb}
581: \end{equation}
582: where $F(\alpha,\gamma;z)$ is the confluent hypergeometric
583: function, $\lambda_n = E_n/\hbar\omega$ and
584: \begin{equation}
585: c_1 = 1 + a,
586: \qquad
587: c_2 = 1 - a,
588: \qquad
589: a=
590: {1\over 2}\sqrt{1+\frac{8mg}{\hbar^2}},
591: \qquad
592: y =\sqrt{\frac{m\omega}\hbar} \, x.
593: \label{eqn:solprmt}
594: \end{equation}
595: The point is that, if the coupling constant $g$ is in the range
596: \begin{equation}
597: 0 < g < {{3\hbar^2}\over{8m}},
598: \label{eqn:ccrange}
599: \end{equation}
600: we have ${1\over 2} < a < 1$, and therefore both of the two solutions
601: (\ref{eqn:indgensla}), (\ref{eqn:indgenslb}) are square integrable, even though $\phi^{(2)}_n$
602: may diverge at
603: $x = 0$. The existence of the solution which does not vanish (actually diverges) at $x = 0$ implies that in quantum mechanics the system does {\it not} split there, in contrast to the classical case.
604:
605:
606: The general solution $\psi_n (x)$ is then given by a linear
607: combination of these two solutions with arbitrary coefficients
608: $N_{\rm R}^{(s)}$ and $N_{\rm L}^{(s)}$ for $s = 1$, 2, which can differ on the positive and negative sides,
609: \begin{eqnarray}
610: \psi_n (x)
611: &=&
612: [{N_{\rm R}^{(1)}}\phi^{(1)}_n(|x|)
613: +{N_{\rm R}^{(2)}}\phi^{(2)}_n(|x|)]\Theta(x) \nonumber \\
614: &{}&\quad
615: +[{N_{\rm L}^{(1)}}\phi^{(1)}_n(|x|)+{N_{\rm
616: L}^{(2)}}\phi^{(2)}_n(|x|)]\Theta(-x),
617: \label{eqn:solgen}
618: \end{eqnarray}
619: where
620: $\Theta(x)$ is the Heaviside step function.
621: Let us now choose
622: the reference modes
623: \begin{equation}
624: \varphi_{1}(x) := \sqrt{\frac{\hbar}{m\omega}}\,
625: \phi_{n_0}^{(1)}(\vert x\vert)\left[\Theta(x) -
626: \Theta(-x)\right],
627: \qquad
628: \varphi_{2}(x) :=
629: \frac{1}{c_2 - c_1}\,
630: \phi_{n_0}^{(2)}(\vert x\vert),
631: \label{eqn:rrmode}
632: \end{equation}
633: which are the solutions in (\ref{eqn:indgensla}), (\ref{eqn:indgenslb})
634: with $n = n_0$ for which $\lambda_{n_0} = 0$. Using these
635: in (\ref{eqn:gbvec}) to get the boundary vectors,
636: \begin{equation}
637: \Psi= ({c_1}-{c_2})
638: \pmatrix{
639: N_{\rm R}^{(2)}\cr
640: N_{\rm L}^{(2)}
641: },
642: \qquad
643: \Psi'=\sqrt{\frac{m\omega}\hbar} \,
644: \pmatrix{
645: N_{\rm R}^{(1)}\cr
646: N_{\rm L}^{(1)}
647: },
648: \label{eqn:bvcaust}
649: \end{equation}
650: and then plugging these in the connection condition
651: (\ref{eqn:bcon}),
652: one obtains the spectral condition,
653: \begin{equation}
654: \frac{1}{{c_1}-{c_2}}
655: \sqrt{\frac{m\omega}\hbar}\frac{\Gamma\left(({c_1}-\lambda_n)/2\right)}
656: {\Gamma\left(({c_2}-\lambda_n)/2\right)}
657: \frac{\Gamma(c_2 )}{\Gamma(c_1)} = \frac{1}{L(\theta_\pm)}.
658: \label{eqn:ccra}
659: \end{equation}
660: Thus one finds that, in general, there exist two series of energy levels, one
661: specified by $L(\theta_+)$ and the other by $L(\theta_-)$.
662: {}For instance, if the singularity is free, $U = \sigma_1$, then we have
663: the two series of eigenstates,
664: \begin{equation}
665: \psi_n^{(1)}(x)
666: =
667: N^{(1)}\, \phi^{(1)}_n(\vert x\vert)
668: \left[\Theta(x) - \Theta(-x)\right], \qquad
669: \psi_n^{(2)}(x)
670: =
671: N^{(2)}\, \phi^{(2)}_n(\vert x\vert) ,
672: %\left[\Theta(x) + \Theta(-x)\right]
673: \label{eqn:indgen}
674: \end{equation}
675: with the eigenvalues,
676: \begin{equation}
677: E^{(1)}_n =(2n+{1 + a})\hbar\omega, \qquad
678: E^{(2)}_n =(2n+{1 - a})\hbar\omega,
679: \label{eqn:elevels}
680: \end{equation}
681: for $n = 0, 1, \ldots$. In the limit $g \to 0$ ($a \to 1/2$)
682: these states reduce to the familiar eigenstates of the harmonic oscillator as
683: expected. Such a smooth limit does not exist, however, for other
684: singularities, {\it e.g.,} at the Dirichlet point $U = -I$, one obtains
685: the doubly degenerate energy levels
686: $E_n = (2n+{c_1})\hbar\omega$ which do not reduce to those of the harmonic oscillator. This case $U = -I$ corresponds to the conventional connection condition used to
687: provide the solutions in the Calogero model~\cite{Calogero}.
688:
689: Having solved the quantum system, we now see that the singularity allows quantum tunneling. Indeed, for the free case $U = \sigma_1$, for instance, the generic state
690: \begin{equation}
691: \psi(x) = \sum_n(c_n^{(1)}\psi_n^{(1)}(x) +
692: c_n^{(2)}\psi_n^{(2)}(x))
693: \label{eqn:genstate}
694: \end{equation}
695: has the probability current at the singularity
696: \begin{equation}
697: j(\pm 0) =
698: \frac{ia\hbar}{m}\sum_{n, l}
699: \left\{ (c_n^{(1)})^*c_l^{(2)} - (c_n^{(2)})^*c_l^{(1)}\right\},
700: \label{eqn:pcrnt}
701: \end{equation}
702: which is non-vanishing. The tunneling is seen generically, except for
703: those singularity belonging to the separated subfamily mentioned earlier.
704: Another evidence may be gained from the transition amplitude,
705: \begin{equation}
706: K({x_f},{t_f};{x_i},{t_i})
707: = \langle x_f \vert e^{-{i\over\hbar} H (t_f - t_i)} \vert x_i \rangle,
708: \label{eqn:tramp}
709: \end{equation}
710: which can be evaluated exactly with the help of the solutions obtained above.
711: We then find that, for the transition time $T := t_f - t_i \neq k\pi/\omega$
712: ($k = 0, 1, 2,\ldots$), the amplitude is expressed in terms of the modified Bessel function,
713: and from it we learn that the transition across the singularity is indeed allowed.
714:
715:
716:
717: \begin{figure}[t]
718: \begin{center}
719: \includegraphics[width=0.67\linewidth]{f5.eps}
720: \caption{Process of quantum copy through the caustics anomaly.
721: At every period $T =
722: k\pi/\omega$, a mirror image of the original profile on $x > 0$
723: emerges on the
724: other side $x < 0$. The relative
725: size of the mirror image depends on $a$ and
726: $k$.}
727: \label{qcopy}
728: \end{center}
729: \end{figure}
730:
731:
732:
733: The remarkable point is that, at the periods of oscillation
734: $T=k\pi/\omega$, the amplitude turns out to be
735: \begin{equation}
736: K({x_f},{t_f};{x_i},{t_i})
737: =
738: (-1)^k \cos(ak\pi)\delta({x_f}-{x_i})
739: + i(-1)^k \sin(ak\pi)\delta({x_f}+{x_i}).
740: \label{eqn:trptd}
741: \end{equation}
742: The first term on the r.h.s.~corresponds to the return of
743: the particle to its initial position (which is the classical caustics), while the second term corresponds to the tunneling of the particle which reaches
744: the mirror point of the initial position with respect to the singular wall. This shows that the classical caustics
745: has been modified at the quantum level ({\it i.e., caustics anomaly}), in such a way that we can now have the mirror image of the original
746: profile prepared at the initial time $t = t_i$, with the weight factors being the functions of the parameter $a$ determined from the coupling constant $g$ (and the characteristic matrix $U$ for the general case). In other words, one can \lq copy\rq{} an original
747: profile prepared on the $x > 0$ side to the other $x < 0$ side after
748: the periods, and that this can be done with desirable weight factors if one can control the
749: relevant parameters of the factors freely~\cite{MT}. Note that this copying process is not in conflict
750: with the no-go theorem~\cite{WoottersZurek} of quantum cloning, because the process takes
751: place in one Hilbert space rather than two as presumed in the theorem.
752:
753:
754:
755: \section{Quantum Force on a Partition Wall}
756:
757:
758: Let us finish our discussion with another example to exhibit how
759: remarkably distinct physics arises for distinct possible dots and walls.
760:
761: Consider an interval $[-l, l]$ bordered by Dirichlet
762: reflecting walls, $\psi(\pm l) = 0$. Suppose we insert a separating dot at
763: the centre with $L(\theta_{+}) = \infty$, $L(\theta_{-}) = 0$, in other
764: words, a partition wall that imposes the Neumann condition from the right
765: and the Dirichlet one from the left. Suppose also that we put $N$ identical bosonic
766: particles into each of the two half wells, which we keep at the same
767: temperature, and calculate the quantum statistical average forces (or
768: pressure) acting on the partition from the right and the left. Notably,
769: the only difference between the circumstances on the two half wells is
770: the distinct reflecting property of the separating wall from the two
771: directions. We will see
772: that, due solely to this fact, the net force will be nonvanishing and
773: reaching arbitrarily large values at high enough temperatures~\cite{FMT}.
774:
775: To see this, we recall first that
776: the right and left half wells admit the energy levels
777: $E^{\pm}_n = e^{\pm}_n {\cal E}$, $n = 1$, $2$, $3$ $\ldots$, with
778: \begin{equation}
779: e^{+}_n = \left(n-{1\over 2}\right)^2, \qquad
780: e^{-}_n = n^2, \qquad
781: {\cal E} = {{\hbar^2}\over{2m}}\left({{\pi}\over{l}}\right)^2,
782: \label{unitenergy}\end{equation}
783: where we use hereafter the indices \lq $+$\rq{} and \lq $-$\rq{} to indicate quantities for the right and left half wells, respectively.
784: The particles will distribute among these eigenstates according to the
785: Bose-Einstein statistics,
786: \begin{equation}
787: N_n^\pm = {1\over{e^{\alpha^\pm + e^\pm_n / t} - 1}},
788: \label{popl} \end{equation}
789: where $t$ is the dimensionless temperature parameter $t = k T / {\cal E}$,
790: and the temperature dependent $\alpha^\pm = \alpha^\pm(t)$ are determined by $N =
791: \sum_{n} N_n^\pm$. The forces acting from the right and the left are
792: given by
793: \begin{equation}
794: F^\pm = - \sum_n {{\partial E_n^\pm}\over{\partial l}} N_n^\pm =
795: \frac{2 {\cal E}}{l} \sum_n e^\pm_n N^\pm_n.
796: \label{nfp} \end{equation}
797:
798: For low temperatures, most of the particles are in the ground state,
799: and $N^\pm_n$ decrease exponentially fast for higher $n$. Consequently, the net force
800: will be contributed essentially by the ground and first excited levels (see Figure~\ref{fig:low}),
801: giving
802: \begin{equation}
803: \Delta F(t) \approx \frac{2 {\cal E}}{l}
804: \left[ \frac{3}{4} N + 3 \, e^{-3/t} - 2 \, e^{-2/t} \right]
805: \label{nfres} \end{equation}
806:
807: \begin{figure}[t]
808: \begin{center}
809: \includegraphics[width=0.5\linewidth]{f6.eps}
810: \caption{The dimensionless net force $ \frac{l}{2 {\cal E}} \Delta F(t) $
811: for $N = 100$, in the temperature region $t < 1$, obtained by a numerical
812: computation (solid line), and approximated by (\ref{nfres}) (dashed line).}
813: \label{fig:low}
814: \end{center}
815: \end{figure}
816:
817: As temperature is increased, more and more levels enter, and the
818: net force starts to decrease approximately linearly. This behavior
819: can be accounted for by using a heuristic argument \cite{FMT} which
820: classifies the energy levels into three classes and estimates
821: the contribution of each class in turn. The result,
822: \begin{equation}
823: \Delta F(t) \approx \frac{2 {\cal E}}{l}
824: \left[ {3\over 4} N - {t\over{(e - 1)^2}} \right]
825: \label{nflt} \end{equation}
826: proves to be satisfactory up to $t \approx 2N/3$, where the net
827: force reaches a minimum and starts to increase afterwards (see
828: Figure~\ref{fig:med}).
829:
830: To explain this minimum and the increase following it with an
831: analytic approximation formula, let us replace the infinite sums (\ref{nfp}) and $N = \sum_{n}
832: N_n^\pm$ with corresponding integrals, which is allowed
833: in this temperature region. Assuming $|\alpha^\pm| < 1$
834: as well, one then obtains \cite{FMT}
835: \begin{equation}
836: \Delta F \approx \frac{2 {\cal E}}{l} \left[ \left( N t + \frac{35}{96}
837: \sqrt{\pi} t^{3/2} \right) \left( \alpha^+ - \alpha^- \right)
838: + \left( \sqrt{ e_1^+ } - \sqrt{ e_1^- } \right) t \right] ,
839: \label{qnf} \end{equation}
840: where the $\alpha^\pm$ are to be determined from
841: \begin{eqnarray}
842: N & \approx & \frac{1}{ \alpha^\pm + e^\pm_1 / t }
843: + \frac{1/2}{ \alpha^\pm + e^\pm_2 / t } - \frac{3}{4}
844: - \frac{ \sqrt{( 2 - \alpha^\pm ) t} - \sqrt{e^\pm_2} }{2} \nonumber \\
845: & & + \sqrt{ \frac{t}{|\alpha^\pm|} } \left[ A \left(
846: \sqrt{ \frac{ |\alpha^\pm| t }{e^\pm_2} } \right)
847: - A \left( \sqrt{ \frac{ |\alpha^\pm| }{ 2 - \alpha^\pm } } \right) \right],
848: \label{aat} \end{eqnarray}
849: with $A$ denoting the arctan function for positive $\alpha^\pm$ and the
850: arctanh function for negative $\alpha^\pm$. It is not easy to express the
851: solution $\alpha^\pm$ of (\ref{aat}) directly via an approximate analytic
852: formula, but by calculating the solution numerically and
853: applying it in (\ref{qnf}), one can observe that there indeed occur the minimum of the net force and the increase
854: following it (see Figure~\ref{fig:med}).
855: We point out that, in the dimensionless unit, both the zero temperature limit
856: $\frac{l}{2 {\cal E}} \Delta F(0) = \frac{3}{4} N$ and the temperature $t$ at which
857: the net force takes its minimum are of order $N$.
858:
859:
860: \begin{figure}[t]
861: \begin{center}
862: \hskip -0.ex\includegraphics[width=0.45\linewidth]{f7a.eps}\hskip 7.ex%
863: \includegraphics[width=0.45\linewidth]{f7b.eps}
864: \caption{%
865: Left: The
866: dimensionless net force $ \frac{l}{2 {\cal E}} \Delta F(t) $ for
867: $N = 100$, in the temperature region $0 < t < 160$, obtained by a
868: numerical computation (solid line), and approximated by (\ref{nflt})
869: (dotted line), and by (\ref{qnf}) using (\ref{aat}) (dashed line).
870: Right: The dimensionless net force $ \frac{l}{2 {\cal E}} \Delta
871: F(t) $ for $N = 100$, obtained by a numerical computation (solid
872: line), and approximated for high temperatures with (\ref{aaj})
873: (dashed line). The figure is double logarithmic.}
874: \label{fig:med}
875: \end{center}
876: \end{figure}
877:
878: When we increase the temperature further, we find that the net force keeps
879: increasing, and actually proves to tend to infinity with a
880: square-root-of-temperature asymptotic behavior (see Figure 7).
881: This temperature dependence can be derived as follows.
882: By expanding $N_n^\pm$ in terms of $ q^\pm := e^{- \alpha^\pm} $ as
883: \begin{equation}
884: N_n^\pm = \frac{ q^\pm e^{- e_n^\pm / t} }{ 1 - q^\pm e^{- e_n^\pm / t} }
885: = \sum_{k = 1}^{\infty} (q^\pm)^k e^{- k e^\pm_n / t} ,
886: \label{aad} \end{equation}
887: we have
888: \begin{equation}
889: N = \sum\limits_{n = 1}^{\infty} N_n^\pm = \sum\limits_{k = 1}^{\infty}
890: (q^\pm)^k \sum\limits_{n = 1}^{\infty} e^{- k e_n^\pm / t}
891: = \sum\limits_{k = 1}^{\infty} (q^\pm)^k \left[ - \frac{\sigma^\pm}{2} +
892: \frac{1}{2} \sum\limits_{n = - \infty}^{\infty} e^{- k e^\pm_n / t} \right]
893: \label{aae} \end{equation}
894: with $\sigma^+ = 0$ and $\sigma^- = 1$. Applying now the Poisson
895: summation formula,
896: \begin{equation}
897: \sum_{n = - \infty}^{\infty} y(n) = \sum_{m = - \infty}^{\infty}
898: \int_{-\infty}^{\infty} ds\, y(s) e^{ 2 \pi i m s },
899: \label{poisson}
900: \end{equation}
901: we obtain
902: \begin{equation}
903: N = \sum\limits_{k = 1}^{\infty} (q^\pm)^k \left[ - \frac{\sigma^\pm}{2}
904: + \sqrt{ \frac{\pi t}{4 k} } \sum\limits_{m = - \infty}^{\infty}
905: (\mp 1)^m e^{ - \frac{\pi^2 t}{k} m^2 } \right] .
906: \label{aag} \end{equation}
907: Similarly, for the forces $F^\pm$ [see (\ref{nfp})], we find
908: \begin{equation}
909: F^\pm = \frac{2 {\cal E}}{l} \sum\limits_{k = 1}^{\infty} (q^\pm)^k
910: \sqrt{ \frac{\pi t^3}{16 k^3} } \sum\limits_{m = - \infty}^{\infty}
911: (\mp)^m \left( 1 - \frac{2 \pi^2 t}{k} m^2 \right)
912: e^{ - \frac{\pi^2 t}{k} m^2 } .
913: \label{aah} \end{equation}
914: {}For the high-temperature asymptotic behavior ($N^\pm_1 \to 0
915: \Rightarrow q^\pm \to 0 $), it suffices to consider only the first
916: two terms in the sums over $k$ in these sums, and within each term to keep
917: only the $m = 0$ term in the sums over $m$ (the $m \ne 0$ terms being
918: exponentially suppressed). Thus, from (\ref{aag}) we get
919:
920: \begin{equation}
921: q^\pm = 2 N / (\pi t)^{1/2} + 2 N ( \sigma^\pm - \sqrt{2} N ) / ( \pi t)
922: + {\cal O} ( t^{-3/2} ),
923: \label{aai} \end{equation}
924: and, correspondingly, the net force,
925: \begin{equation}
926: \Delta F = \frac{{\cal E} N}{l} \left( \frac{t}{\pi} \right)^{1/2}
927: + {\cal O} ( t^{0} ) ,
928: \label{aaj} \end{equation}
929: which is the promised square-root-of-temperature asymptotic behavior.
930: We can see in Figure 7 how this asymptotic behavior is
931: reached at high temperatures.
932:
933: The result that the net force does not tend to zero (nor
934: to a nonzero constant) seems unusual when contrasted to the naive
935: expectation that such quantum effects coming from the distinct
936: boundary conditions should vanish at high temperatures where the
937: classical picture would be available. However, this surprising
938: feature can be understood by the fact that, contrary to most
939: quantum systems, one dimensional wells have such energy spectra
940: that the level spacing is not decreasing but increasing for higher
941: energy levels (which is actually valid not only for boxes with
942: Dirichlet and/or Neumann boundary conditions but for all other
943: wells as well~\cite{FTC}). In other words, quantum wells can be
944: distinguished by their high-temperature behavior, too.
945:
946: \begin{figure}[t]
947: \begin{center}
948: \includegraphics[width=0.5\linewidth]{f8.eps}
949: \caption{The dimensionless net force
950: $\frac{l}{2 {\cal E}} \Delta F(t)$
951: in the fermionic case, for $N = 100$, in the low-temperature
952: region.}
953: \label{fig:fer}
954: \end{center}
955: \end{figure}
956:
957: One can replace the bosonic particles with fermions, and
958: consider the same problem as above, too.
959: The net force is found to exhibit a
960: qualitatively similar temperature dependence as in the bosonic case, with
961: two main differences. One of them is that the $t = 0$ value of the net
962: force and the temperature where the net force takes its minimum are
963: proportional to $N^2$ rather than to $N$ as seen in the bosonic case.
964: The other is that one
965: slight ``step" can be observed at low temperatures, where the net force
966: starts to decrease from its $t = 0$ value (see Figure~\ref{fig:fer}).
967: Otherwise the fermionic case is similar to the bosonic one, and, for
968: example, the high-temperature asymptotics proves to be the same
969: square-root-of-temperature one~\cite{FTtwo}.
970:
971:
972: \section{Discussions}
973:
974: We have discussed in this paper some of the interesting physical phenomena that can arise on a line if there is a controllable point singularity either in the form of a dot or a wall. It is remarkable that putting just a singular point on a line
975: allows for such variety of phenomena --- duality, anholonomy, supersymmetry, caustics anomaly and the emergence of pressure --- which are found usually in more involved
976: systems such as gauge field theory. In addition, the scale anomaly ({\it i.e.} the breakdown of the classical scale symmetry at the quantum level) which we did not mention in this paper can also be seen in the system with a generic singularity. This is implied by the presence of
977: the scale parameters $L(\pm \theta)$, which are missing in the classical description.
978: It is, therefore, safe to say
979: that the crucial element for those quantum phenomena to occur is not in the complexity of the system nor
980: in the infinity of the degrees of freedoms of the system as often assumed. Rather, these are allowed because the quantum description of a system requires more information (and hence more parameters to be fixed) than the classical description does, and that once the extra parameters are chosen, some of the properties that hold classically may no longer hold, causing the anomalies at the quantum level. The extra parameters in the case of a singularity on a line are given by the group $U(2)$, whose global structure is then used to yield the anholonomy effects, for example.
981:
982: Putting the singularity in many particle systems offers an interesting possibility when considered in the context of statistical mechanics. We have seen this in the simple, albeit not too realistic, example of the quantum force acting on a partition wall in a square potential well. The force attains a minimum at a certain temperature, which is proportional to the particle number $N$ for the bosonic case or to its square $N^2$ for the fermionic case, before it diverges for $T \to \infty$. For the bosonic case
983: with $N \sim 100$, the minimum will be seen in a room temperature if the size of the system is about a few hundred nanometers, whereas for the fermionic case the same can be seen even with larger systems of the size of one micron. In view of the rapid progress of nano-technology in recent years, it may not be entirely unreasonable to expect that some of these effects described here can be observed in laboratory in the near future.
984:
985:
986: \section*{Acknowledgements}
987:
988: We thank our collaborators,
989: T.~Cheon, H.~Miyazaki and T.~Uchino
990: for their valuable contributions and helpful discussions.
991: This work has been supported in part by
992: the Grant-in-Aid for Scientific
993: Research on Priority Areas (No.~13135206) by
994: the Japanese
995: Ministry of
996: Education, Science, Sports and Culture.
997:
998:
999:
1000: \begin{thebibliography}{99}
1001:
1002:
1003: \bibitem{RS}
1004: M. Reed and B. Simon: \textit{Methods of Modern
1005: Mathematical Physics II, Fourier analysis, self-adjointness}
1006: (Academic Press, New York, 1975).
1007:
1008: \bibitem{AGHH}
1009: S. Albeverio, F. Gesztesy, R. H{\o}egh-Krohn and H. Holden:
1010: \textit{Solvable Models in Quantum Mechanics} (Springer, New York, 1988).
1011:
1012:
1013: \bibitem{AG}
1014: N.I. Akhiezer and I.M. Glazman:
1015: \textit{Theory of Linear Operators in Hilbert Space},
1016: Vol.II
1017: (Pitman Advanced Publishing Program, Boston, 1981).
1018:
1019: \bibitem{FT}
1020: T. F\"{u}l\"{o}p and I. Tsutsui:
1021: {\it Phys.\ Lett.} {\bf A264} (2000) 366.
1022:
1023: \bibitem{TCF}
1024: I. Tsutsui, T. Cheon and T. F\"{u}l\"{o}p:
1025: {\it J. Phys. A}: Math. Gen. {\bf 36} (2003) 275.
1026:
1027: \bibitem{TFC}
1028: I. Tsutsui, T. F\"{u}l\"{o}p and T. Cheon:
1029: {\it J. Phys. Soc. Jpn.} {\bf 69} (2000) 3473.
1030:
1031: \bibitem{CFT}
1032: T. Cheon, T. F\"{u}l\"{o}p and I. Tsutsui:
1033: {\it Ann.\ Phys.} {\bf 294} (2001) 1.
1034:
1035: \bibitem{Moebius}
1036: I. Tsutsui, T. F\"{u}l\"{o}p and T. Cheon:
1037: {\it J.\ Math.\ Phys.} {\bf 42} (2001) 5687.
1038:
1039: \bibitem{UTone}
1040: T. Uchino and I. Tsutsui:
1041: {\it Nucl. Phys.} {\bf B662} (2003) 447.
1042:
1043:
1044: \bibitem{UTtwo}
1045: T. Uchino and I. Tsutsui:
1046: {\it J. Phys. A}: Math. Gen. {\bf 36} (2003) 6821.
1047:
1048: \bibitem{Schulman}
1049: L.S. Schulman: \textit{Techniques and Applications of
1050: Path Integration}
1051: (John Wiley and Sons, New York, 1981).
1052:
1053: \bibitem{Calogero}
1054: F. Calogero: {\it J.\ Math.\ Phys.} {\bf 10} (1969) 2191,
1055: 2197; {\bf 12} (1971) 419.
1056:
1057: \bibitem{MT}
1058: H. Miyazaki and I. Tsutsui:
1059: {\it Ann. Phys.} {\bf 299} (2002) 78.
1060:
1061: \bibitem{WoottersZurek}
1062: W.K. Wootters and W.H. Zurek: {\it Nature} {\bf 299} (1982) 802.
1063:
1064:
1065:
1066: \bibitem{FMT}
1067: T. F\"{u}l\"{o}p, H. Miyazaki and I. Tsutsui:
1068: {\it
1069: Quantum Force due to Distinct Boundary Conditions},
1070: to appear in {\it Mod. Phys. Lett. A}; e-print quant-ph/0309095.
1071:
1072:
1073: \bibitem{FTC}
1074: T. F\"{u}l\"{o}p, I. Tsutsui and T. Cheon:
1075: {\it J. Phys. Soc. Jpn.} {\bf 72} (2003) 2737.
1076:
1077: \bibitem{FTtwo}
1078: T. F\"{u}l\"{o}p and I. Tsutsui:
1079: {\it
1080: Quantum Force generated by Boundary Conditions},
1081: in preparation.
1082:
1083:
1084:
1085: \end{thebibliography}
1086: \end{document}
1087:
1088:
1089: