1: \documentclass[aps,prb,twocolumn,groupedaddress,showpacs]{revtex4}
2:
3: \usepackage{amssymb}
4:
5: \usepackage{graphicx}
6:
7: \bibliographystyle{apsrev}
8:
9: \begin{document}
10:
11: \title{Using Qubits to Measure Fidelity in Mesoscopic Systems}
12:
13: \author{G.B.\ Lesovik$^{a,b}$, F.\ Hassler$^{b}$, and G.\ Blatter$^{b}$}
14:
15: \affiliation{$^{a}$L.D.\ Landau Institute for Theoretical Physics RAS,
16: 117940 Moscow, Russia}
17:
18: \affiliation{$^{b}$Theoretische Physik, ETH-H\"onggerberg, CH-8093
19: Z\"urich, Switzerland}
20:
21: \date{\today}
22:
23: \begin{abstract}
24: We point out the similarities in the definition of the `fidelity' of a
25: quantum system and the generating function determining the full counting
26: statistics of charge transport through a quantum wire and suggest to use
27: flux- or charge qubits for their measurement. As an application we use the
28: notion of fidelity within a first-quantized formalism in order to derive new
29: results and insights on the generating function of the full counting
30: statistics.
31: \end{abstract}
32:
33: \pacs{73.23.-b, 05.45.Mt, 03.67.-a, 85.25.Cp}
34:
35: \maketitle
36:
37: \section{Introduction}
38:
39: Mesoscopic devices exhibit an extraordinary rich and complex behavior;
40: their proper characterization has sparked numerous ideas. Two such
41: basic mesoscopic characteristics are the stability of a quantum system
42: \cite{peres}, nowadays compiled under the terms `fidelity' or
43: `Loschmidt echo', and the full counting statistics of charge
44: transport, expressed through the generating function for the
45: distribution of charge transmitted across a quantum wire \cite{ll23}.
46: Here, we combine these items with recent efforts aiming at the
47: physical realization of quantum bits, the controllable quantum
48: two-level systems which are the basic elements of a quantum computer
49: \cite{nielsonchuang}; by now, a number of devices have been realized
50: in the laboratory, among them particular solid-state implementations
51: such as flux- \cite{chiorescu} or charge- \cite{vion} qubits which are
52: easy to couple to. The purpose of this letter is three-fold: i) to
53: draw attention to the equivalence between the fidelity of a quantum
54: system and the generating function for the full counting statistics,
55: two quantities which have been conceived to be unrelated so far. This
56: insight generalizes the concept of fidelity to mixed states and
57: many-particle systems. ii) to suggest measuring the fidelity/full
58: counting statistics using quantum bits by exploiting the induced
59: `decoherence' as a signal; this is opposite to the standard setting
60: where the main interest is in the qubit's decoherence due a noisy
61: environment \cite{makhlin} or a measuring device \cite{averin}. Our
62: proposal renders the theoretical concepts of fidelity/full counting
63: statistics amenable to real experimental tests, e.g., using
64: high-quality qubits available today \cite{chiorescu,vion}. In this
65: context, we recognize earlier suggestions to use two-level
66: systems/qubits as measuring devices \cite{schoelkopf}. iii) to use the
67: equivalence between the full counting statistics and the fidelity to
68: obtain further information on the statistics of charge transport.
69: %
70: \begin{figure}[ht]
71: \includegraphics[width=8.0cm]{lo1.eps}
72: \caption[]{(a) Layout: A (flux) qubit (not to scale, placed
73: downstream the scatterer) is used to measure the fidelity or full
74: counting statistics of a quantum point contact (QPC); the
75: measurement current $I_\mathrm{m}$ is switched on only during the
76: readout of the qubit state. For a magnetic coupling $\lambda h
77: \sigma_z$, the direction of the magnetic field defines the
78: $z$-axis. (b) Level scheme for the qubit, which is initially
79: prepared in the ground state $|0\rangle$ at the fully frustrated
80: point O and suddenly switched to M at $t=0$ for the measurement,
81: producing an initial state $|\Psi\rangle =
82: [|+\rangle+|-\rangle]/\sqrt{2}$. The semi-classical states
83: $|\pm\rangle$ carry circulating currents perturbing the wire in
84: the measurement of the fidelity; in the measurement of the full
85: counting statistics the state of the qubit is incrementally
86: changed at each passage of a charge through the wire. }
87: \label{fig:qubit}
88: \end{figure}
89: %
90:
91: In this context, there are two points of view in describing a system
92: (such as a regular or chaotic dot, a quantum wire, etc.) coupled to a
93: qubit: i) the qubit-centered view, where the dot/wire acts as a noisy
94: environment producing decoherence of the qubit state \cite{makhlin}
95: --- this traditional view is pursued in the field of quantum
96: computing; ii) the dot/wire-centered view, where the qubit serves as a
97: measurement device providing information on the system (the dot or
98: wire). This is the new standpoint we take in the present paper where
99: we are interested in two system properties, the fidelity and the full
100: counting statistics.
101:
102: The notion of {\it fidelity} has been introduced in order to quantify
103: the stability of a quantum system (described by the Hamiltonian
104: $H_\mathrm{sys}$) under the action of a small external perturbation
105: $\lambda h$ \cite{peres}: evaluating the evolution of an initial state
106: $\Psi$ under the action of the system's Hamiltonian $H_\mathrm{sys}$,
107: $\Psi(t) = \exp(-iH_\mathrm{sys}t/\hbar) \Psi$, comparison is made
108: with the perturbed evolution $\Psi_\lambda (t) =
109: \exp[-i(H_\mathrm{sys}+\lambda h)t/\hbar]\Psi$ through the matrix
110: element \cite{modsq}
111: %
112: \begin{eqnarray}
113: \chi_\mathrm{fid} (\lambda,t) &=&
114: \langle \Psi_\lambda (t)|\Psi(t)\rangle
115: \label{chi_fid}\\
116: &=&\langle\Psi| e^{i(H_\mathrm{sys}+\lambda h)t/\hbar}
117: e^{-iH_\mathrm{sys}t/\hbar}
118: |\Psi\rangle; \nonumber
119: \end{eqnarray}
120: %
121: a quantum system with a chaotic classical correspondent exhibits a
122: rapid time-decay of the fidelity $\chi_\mathrm{fid}$
123: \cite{peres,jalabert}, while a regular classical analogue leads to its
124: saturation at a finite value \cite{peres}. Above, the fidelity has
125: been defined for a pure state formulated in a first-quantization
126: language; the definition (\ref{chi_fid}) agrees with its quantum
127: information theoretic analogue for two pure state $\Psi$ and
128: $\Psi_\lambda$ \cite{nielsonchuang}.
129:
130: In general, the interference between two wave functions as manifested
131: in (\ref{chi_fid}) is difficult to measure. Here, we put forward the
132: idea to couple the system under investigation to an external device
133: (e.g., a two-level system in the form of a spin or a qubit) which
134: simultaneously acts as a perturbation and as a measuring device for
135: the fidelity, see Fig.\ \ref{fig:qubit}(a): the coupling $\lambda h
136: \sigma_z$ entangles the system with the measurement device, thus
137: transferring information of the system's evolution which then can be
138: measured along with the quantum state of the device. A similar idea
139: has been introduced in the context of the {\it full counting
140: statistics} (FCS) which is characterizing the charge transport
141: through a quantum wire. The task then is to count the number of
142: transferred charges which corresponds to the measurement of the
143: integrated current $Q(t)=\int_0^t dt'\, I(t')$. The straightforward
144: (classical) Ansatz \cite{ll1} for the generating function
145: $\chi(\lambda,t) = \langle \exp[i\lambda Q(t)] \rangle$ is problematic
146: as no prescription for the time-ordering of current operators
147: appearing at different instances of time is given. The idea to couple
148: the system to a measurement device resolves these problems: following
149: the work of Levitov and Lesovik \cite{ll23}, the complete statistical
150: information can be obtained by coupling the wire to a spin degree of
151: freedom serving as the measuring device; the Fourier coefficients
152: $\int d \lambda \exp(-i\,n\lambda) \chi_{\rm\scriptscriptstyle
153: FCS}(\lambda,t)$ of the generating function
154: %
155: \begin{equation}
156: \chi_{\rm\scriptscriptstyle FCS}(\lambda,t) =
157: \mathrm{Tr}_\mathrm{w}
158: [\rho^\mathrm{w}(0)\, e^{i(H_\mathrm{w}+\lambda h)t/\hbar}
159: e^{-i(H_\mathrm{w}-\lambda h)t/\hbar}]
160: \label{chi_fcs}
161: \end{equation}
162: %
163: provide the probabilities $P_n(t)$ for the passage of $n$ charges
164: during the time $t$. Here, $H_\mathrm{w}$ ($\rho^\mathrm{w} (0)$)
165: denotes the wire's Hamiltonian (density matrix) and $\lambda h$
166: describes its coupling to the spin; the trace is taken over the wire's
167: degrees of freedom, while the off-diagonal matrix element has been
168: evaluated in spin space, see below.
169:
170: \section{Equivalence between fidelity and FCS}
171:
172: The full counting statistics refers to a many-body system cast in a
173: second-quantized formalism. Still, comparing (\ref{chi_fid}) and
174: (\ref{chi_fcs}), we immediately note the similarity of the two
175: expressions: indeed, replacing the arbitrary perturbation in
176: (\ref{chi_fid}) by the coupling to a spin (or two-level system)
177: serving at the same time as a perturbation {\it and} as a measurement
178: device, we arrive at the form in (\ref{chi_fcs}) with the trace over
179: the wire's degrees of freedom replaced by the quantum average over the
180: initial state $\Psi$. In previous discussions \cite{peres,jalabert}
181: the fidelity of a quantum system has been related to the
182: chaotic/regular nature of the system; probing such a system with a
183: measurement device as described above provides the identical
184: information. The notion of fidelity introduced here generalizes this
185: concept to quantum systems without classical analogue as well as mixed
186: state- and many-particle systems; the quantum information theoretic
187: definition for density matrices \cite{nielsonchuang} is different,
188: however.
189:
190: The fidelity/generating function $\chi(\lambda,t)$ is obtained by
191: coupling a spin (in zero external magnetic field) to the system via a
192: Hamiltonian of the form $H_\mathrm{int}=\lambda h\sigma_z$ (the
193: absence of terms $\propto \sigma_{x,y}$ is crucial). The evolution of
194: the spin then is described by the reduced density matrix
195: %
196: \begin{equation}
197: \rho^\mathrm{s}(t) = \mathrm{Tr}
198: \bigl[ e^{-i (H_\mathrm{sys} + H_\mathrm{int})t/\hbar} \rho(0) \, e^{i
199: (H_\mathrm{sys} + H_\mathrm{int})t/\hbar}\bigr]
200: \label{rho_qubit}
201: \end{equation}
202: %
203: with the initial separable density matrix $\rho(0) =
204: \rho^\mathrm{sys}(0) \otimes \rho^\mathrm{s}(0)$ and the trace is
205: taken over the system degrees of freedom without the spin; for the
206: fidelity (\ref{chi_fid}) we have to replace $\rho^\mathrm{sys}(0) \to
207: \rho^\Psi(0) = |\Psi\rangle \langle\Psi|$. Evaluating the spin part
208: first, we subsequently exploit the cyclic property of the trace and
209: obtain
210: %
211: \begin{eqnarray}
212: &&\rho^\mathrm{s}_{\sigma_z',\sigma_z}(t)
213: = \rho^\mathrm{s}_{\sigma_z',\sigma_z}(0)
214: \label{rho_qubit_ev}\\
215: &&\qquad \times \mathrm{Tr} \bigl[\rho^\mathrm{sys}(0) \,
216: e^{i(H_\mathrm{sys}+\lambda h\sigma_z)t/\hbar}
217: e^{-i(H_\mathrm{sys}+\lambda h\sigma_z')t/\hbar}\bigr].
218: \nonumber
219: \end{eqnarray}
220: %
221: The measurement of the off-diagonal entry $\rho^\mathrm{s}_{-1,1}
222: (t)$ provides us with the sought-after quantity $\chi(\lambda,t)$.
223:
224: Below, we will replace the spin degree of freedom by the more
225: versatile qubit. Thereby, we transfer the measurement of the fidelity
226: and of the full counting statistics from the realm of a `Gedanken'
227: experiment to a practical proposal realizable with today's qubit
228: technology. In this context, we recognize previous steps taken in this
229: direction in measuring the fidelity of a quantum kicked rotator
230: \cite{qkr} and in the measurement of higher order correlators
231: \cite{prober}, see also Ref.\ \cite{lesovik_94} for alternative
232: theoretical proposals.
233:
234:
235: \section{Measurement with qubits}
236:
237: On a technical level, our problem is described by the Hamiltonian (we
238: formulate the problem for a wire; the extension to other systems is
239: straightforward)
240: %
241: \begin{equation}
242: H = H_\mathrm{w}+H_\mathrm{q}+H_\mathrm{int}
243: \label{ham}
244: \end{equation}
245: %
246: with $H_\mathrm{w}$ the wire's Hamiltonian of which the fidelity
247: and/or full counting statistics shall be determined,
248: %
249: \begin{equation}
250: H_\mathrm{q} = ({\epsilon}/{2})\sigma_z
251: -({\Delta}/{2})\sigma_x
252: \label{hamqub}
253: \end{equation}
254: %
255: is the qubit Hamiltonian written in the semi-classical basis
256: $|\pm\rangle$ (see Fig.\ \ref{fig:qubit}) and
257: %
258: \begin{equation}
259: H_\mathrm{int} = \lambda h \sigma_z \label{hamint}
260: \end{equation}
261: %
262: is the interaction Hamiltonian coupling the qubit to the wire. For a
263: magnetic (transverse) coupling, we have the standard form
264: %
265: \begin{equation}
266: H_\mathrm{int} = \frac{1}{c} \int_\mathrm{w} dx I_\mathrm{w}
267: ({\bf x}) A_x({\bf x}) \sigma_z \label{hamint_t}
268: \end{equation}
269: %
270: with $I_\mathrm{w}({\bf x})$ the current flowing in the wire and
271: $A_x({\bf x}) = \int_\mathrm{q} d{\bf l}\cdot\hat{\bf x} \,
272: |I_\mathrm{q}|/c |{\bf x} - {\bf l}|$ the gauge potential generated by
273: the qubit current $I_\mathrm{q}$; the coupling constant $\lambda$ is
274: defined such as to add a phase $\lambda/2 = 2\pi \int_\mathrm{w} dx
275: A_x(x) /\Phi_0$ to every electron passing the qubit (here, $\Phi_0 =
276: hc/e$ denotes the unit of flux). For an electric (longitudinal)
277: coupling \cite{pilgrambuttiker}, we have the corresponding expression
278: %
279: \begin{equation}
280: H_\mathrm{int} = \int_\mathrm{w} dx \rho_\mathrm{w}({\bf x})
281: \varphi({\bf x}) \sigma_z \label{hamint_l}
282: \end{equation}
283: %
284: with $\rho_\mathrm{w}({\bf x})$ the 1D charge density of the wire and
285: $\varphi({\bf x})= \int_\mathrm{q} d^2 R \, |\rho_\mathrm{q}|/
286: \varepsilon |{\bf x} - {\bf R}|$ the electric potential generated by
287: the qubit charge density $\rho_\mathrm{q}({\bf R})$ ($\varepsilon$ is
288: the dielectric constant); again, we choose a splitting into $\lambda$
289: and $h$ such that each electron passing the qubit acquires an
290: additional phase $\lambda/2 = (e/\hbar v) \int_\mathrm{w} dx
291: \varphi(x)$ ($e>0$; here, $v$ denotes the (typical) electron velocity
292: and we assume a slowly varying potential, see below for details).
293:
294: Typical qubits we have in mind are the flux qubit as implemented by
295: Chiorescu {\it et al.} \cite{chiorescu} or the charge qubit built by
296: Vion {\it et al.} \cite{vion}. The generic level diagram for these
297: devices is shown in Fig.\ \ref{fig:qubit}(b): The semi-classical
298: states $|+\rangle$ and $|-\rangle$ refer to current- (charge-) states
299: which are the energy eigenstates away from the optimally frustrated
300: state. Frustration produces mixing with new eigenstates $|0\rangle =
301: [|+\rangle +|-\rangle]/\sqrt{2}$ and $|1\rangle = [|+\rangle
302: -|-\rangle]/\sqrt{2}$ at the optimal point O where no currents/charges
303: appear on the qubit. In order to measure the fidelity/full counting
304: statistics, the qubit is prepared in the ground state $|0\rangle$ at
305: optimal frustration (point O in Fig.\ \ref{fig:qubit}(b) with $\Delta
306: = E_\mathrm{O}$ and $\epsilon = 0$; this corresponds to the spin-state
307: polarized along the $x$-axis) and then is suddenly switched (at time
308: $t =0$) to the measuring point M which has to be chosen sufficiently
309: far away from O to avoid mixing of the semi-classical states (i.e.,
310: $\Delta = 0$ and $\epsilon = E_\mathrm{M}$, hence the Hamiltonian
311: (\ref{hamqub}) describes a spin in a magnetic field directed along the
312: $z$-axis). On the other hand, the point M should be chosen not too
313: distant away from O in order to avoid the mixing with other levels. At
314: the end of the signal accumulation (i.e., at time $t$) the state of
315: the qubit has to be measured in the following manner: rotation of the
316: qubit state by $-\pi/2$ ($\pi/2$) around the $y$- ($x$-) axis and
317: subsequent measurement along the $z$-axis provides us with the matrix
318: elements $\langle \sigma_x\rangle = \mathrm{Tr_q}[\rho^\mathrm{q} (t)
319: \sigma_x]$ ($\langle \sigma_y\rangle = \mathrm{Tr_q}[\rho^\mathrm{q}
320: (t) \sigma_y]$) from which the {\it final result} $\chi (\lambda,t) =
321: \exp (-i E_\mathrm{M} t / \hbar) [\langle \sigma_x\rangle+i\langle
322: \sigma_y \rangle]$ follows (the phase $\exp (-i E_\mathrm{M} t /
323: \hbar)$ compensates the trivial time evolution of the qubit in the
324: finite residual field).
325:
326: In order to extract the full counting statistics from the generating
327: function $\chi_{\rm\scriptscriptstyle FCS}(\lambda,t)$, a tunable
328: coupling between the wire and the qubit is needed. The tetrahedral
329: superconducting qubit proposed recently \cite{feigel} lends itself as
330: a particularly useful measurement device: its doubly degenerate ground
331: state emulates a spin in zero magnetic field, hence $H_\mathrm{q} =
332: 0$, while a symmetric charge bias $\delta^{\scriptscriptstyle Q}$
333: produces an interaction Hamiltonian $H_\mathrm{int} \propto
334: \delta^{\scriptscriptstyle Q} \delta^{\scriptscriptstyle \Phi}
335: \sigma_z$ which is linear in the magnetic flux
336: $\delta^{\scriptscriptstyle \Phi}$ (produced by the wire's current)
337: threading the qubit and easily tunable with $\lambda\propto
338: \delta^{\scriptscriptstyle Q}$ (note that imposing a flux
339: $\delta^{\scriptscriptstyle \Phi}$ the tetrahedral qubit also serves
340: as a {\it tunable} charge detector). Alternatively, flux-
341: \cite{chiorescu} or charge- \cite{vion} qubits can be used with a flux
342: tunable third junction or an electrically tunable capacitance.
343:
344: Finally, we have to make sure that the individual electrons passing
345: through the wire are sufficiently coupled to the qubit. This is
346: trivially the case for the electric coupling, where a simple
347: calculation leads to the estimate $\lambda \approx (4 e^2/ \hbar v_F
348: \varepsilon) \ln(L/d)$ with $v_F$ the Fermi velocity, $L$ the wire's
349: length, and $d$ its distance from the qubit; with $c/v_F \varepsilon$
350: of order $10^2$ or larger a $\lambda$-value beyond unity is easily
351: realized. The situation is less favorable for the case of magnetic
352: coupling: associating the qubit with a magnetic dipole $m_\mathrm{q} =
353: I_\mathrm{q} S/c$ ($S$ denotes the qubit's area), we obtain a coupling
354: $\lambda \approx 2 \pi \alpha I_\mathrm{q} S/ced$, $\alpha = e^2/\hbar
355: c$ the finestructure constant. With $I_\mathrm{q}$ of order $1~\mu$A
356: and $S/d \sim 1~\mu$m, we find a coupling $\lambda \sim 10^{-2}$.
357: Similar findings apply to the tetrahedral qubit: applying a
358: homogeneous flux $\delta^\Phi$, the qubit can be used as a tunable
359: charge detector with large coupling $\lambda \approx (4 E_J C/\hbar
360: v_F) (\delta^\Phi/\Phi_0) \ln(L/d)$, $C$ denoting the
361: capacitive coupling between the wire and the qubit. On the other
362: hand, applying a symmetric charge bias $\delta^Q$ we find a magnetic
363: coupling of order $\lambda \approx \alpha (\delta^Q /2e) I_\mathrm{q}
364: S/2ced$ (we have chosen a typical parameter $E_J/E_C = 10^2$). As may
365: be expected, the system is easily coupled to the qubit via electric
366: interaction, while its magnetic coupling is generically weak and has
367: to be suitably enhanced, e.g., with the help of a flux transformer.
368:
369: The {\it magnetic} coupling of the flux qubit allows for the
370: measurement of both the full counting statistics and the system's
371: fidelity. Going over to the interaction representation (with the
372: unperturbed Hamiltonian given by the system Hamiltonian), the
373: generating function assumes the form (cf.\ (\ref{rho_qubit_ev}))
374: %
375: \begin{equation}
376: \chi_{\rm\scriptscriptstyle FCS}(\lambda,t) \!=\! \mathrm{Tr}
377: [\rho^\mathrm{sys}(0) \, \widetilde{\cal T} e^{i\frac{\lambda}{\hbar}\!
378: \int_0^t \!dt' h(t')} \,
379: {\cal T} e^{i\frac{\lambda}{\hbar}\!
380: \int_0^t\! dt' h(t')}],
381: \label{chi_int_rep}
382: \end{equation}
383: %
384: where ${\cal T}$ ($\widetilde{\cal T}$) denote the (reverse) time
385: ordering operators; as desired, the magnetic coupling then is
386: proportional to the integrated current (or transferred charge) $Q(t) =
387: \int_0^t dt' I_\mathrm{w}(t')$; the form (\ref{chi_int_rep}) reveals
388: the role of $\chi_{\rm\scriptscriptstyle FCS}(\lambda,t)$ as the
389: generating function for the cumulants of transferred charge.
390: Alternatively, the qubit can be viewed as a system perturbation and
391: $\chi_{\rm\scriptscriptstyle FCS}$ assumes the role of a fidelity
392: $\chi_\mathrm{fid}$. The {\it electric} coupling to the charge qubit
393: generates a quantity $\chi_\mathrm{fid}(\lambda,t)$ whose meaning is
394: predominantly that of a fidelity; on the other hand, for a uniform and
395: unidirectional charge motion with velocity $v_F$, the time-integrated
396: charge $\int_0^t dt' \rho_\mathrm{w}(t')$ can be related to the
397: transferred charge $Q(t)$ via $Q(t)= v_F\int_0^t dt'
398: \rho_\mathrm{w}(t')$ and thus provides an approximate access to the
399: full counting statistics at finite voltage for which the electrons
400: move in a specific direction along the wire.
401:
402: \section{FCS with wave functions}
403:
404: Inspired by the equivalence between the fidelity and the generating
405: function for the counting statistics, we proceed with the calculation
406: of $\chi(\lambda,t)$ within a first-quantized formalism. In
407: particular, we choose a system in the geometry of a point contact and
408: study the fidelity for the case where a wave packet is incident on the
409: scatterer. This scheme allows for a more elaborate discussion of the
410: various parametric dependencies of the fidelity/FCS but suffers from
411: the restriction to a single particle. In order to ameliorate this
412: limitation, we extend the discussion to a sequence of two incident
413: wave packets, from which the extrapolation to the many body case can
414: be performed.
415:
416: Consider a wave packet (for $t \to -\infty$ and traveling to the
417: right)
418: %
419: \begin{equation}
420: \Psi_\mathrm{in}(x,t) \equiv \Psi_f(x,t)
421: = \int \frac{d k}{2\pi} \, f(k) \, e^{i(kx-\omega_k t)},
422: \label{Psi_in}
423: \end{equation}
424: %
425: centered around $k_0 >0$ with $\omega_k= \hbar k^2 / 2m$ and
426: normalization $\int (dk/2\pi) |f(k)|^2 = 1$, incident on a scatterer
427: characterized by transmission and reflection amplitudes $t_k$ and
428: $r_k$. We place the qubit behind the scatterer to have it interact
429: with the transmitted part of the wave function. The transmitted wave
430: packet then acquires an additional phase due to the interaction with
431: the qubit: for a magnetic interaction the extra phase accumulated up
432: to the position $x$ amounts to $\delta \phi_A(x) = 2 \pi \int^x dx'\,
433: A_x(x')/\Phi_0$, independent of $k$; as $x \to \infty$ this adds up to
434: a total phase $\lambda/2= 2\pi \int_{-\infty}^{\infty} dx\,
435: A_x(x)/\Phi_0$, cf.\ (\ref{hamint_t}). For an electric interaction,
436: cf.\ (\ref{hamint_l}), the situation is slightly more involved: the
437: extra phase can be easily determined for a slowly varying
438: (quasi-classical) potential of small magnitude, i.e., $e|\varphi| \ll
439: \hbar^2 k_0^2/2m$. Expanding the quasi-classical phase $\int^x dx' \,
440: p(x')/ \hbar$ with $p(x) = \sqrt{2m(E+e\varphi(x))}$ to first order in
441: the potential $\varphi(x)$ yields the phase
442: $\delta\phi_\varphi(x)=(e/\hbar v)\int^x dx'\, \varphi(x')$ which
443: asymptotically accumulates to the value $\lambda/2$; its
444: $v$-dependence is due to the particle's acceleration in the scalar
445: potential and will be discussed in more detail below. Moreover, note
446: that $\phi_A$ changes sign for a particle moving in the opposite
447: direction ($k \to - k$) (i.e., under time reversal) whereas
448: $\phi_\varphi$ does not. For a qubit placed behind the scatterer both
449: magnetic and electric couplings produce equivalent phase shifts.
450: Depending on the state $|\pm\rangle$ of the qubit, the outgoing wave
451: (for $t \to \infty$)
452: %
453: \begin{eqnarray}
454: && \Psi_\mathrm{out}^\pm (x,t)=
455: \int \frac{d k}{2\pi} f(k) e^{-i \, \omega_k t}
456: \bigl[r_k e^{-i k x} \Theta(-x) \nonumber \\
457: && \qquad\qquad\qquad\qquad\qquad
458: + e^{\pm i \lambda/2} t_k e^{i \, k x}\Theta(x)\bigr]
459: \nonumber
460: \end{eqnarray}
461: %
462: acquires a different asymptotic phase on its transmitted part. The
463: fidelity is given by the overlap of the two outgoing waves,
464: %
465: \begin{eqnarray}
466: && \chi(\lambda,t) = \int dx {\Psi_\mathrm{out}^-}^*(x,t)
467: \Psi_\mathrm{out}^+(x,t) \label{fid_behind}\\
468: && \qquad = \int \frac{d k}{2\pi} (R_k + e^{i \lambda} T_k)
469: |f(k)|^2 \equiv \langle R\rangle_f + e^{i \lambda} \langle T\rangle_f,
470: \nonumber
471: \end{eqnarray}
472: %
473: where $R_k= |r_k|^2$ and $T_k= |t_k|^2$ denote the probabilities for
474: reflection and transmission, respectively, and we have neglected
475: exponentially small off-diagonal terms involving products $\int dk
476: f(-k)^* f(k)$. The result (\ref{fid_behind}) applies to both magnetic
477: and electric couplings; its interpretation as the generating function
478: of the charge counting statistics provides us with the two non-zero
479: Fourier coefficients $P_0 = \langle R\rangle_f$ and $P_1 = \langle
480: T\rangle_f$ which are simply the probabilities for reflection and
481: transmission of the particle. This result agrees with the usual
482: notion of `counting' those particles which have passed the qubit
483: behind the scatterer. When, instead, the interest is in the system's
484: sensitivity, we observe that the fidelity $\chi(\lambda,t)$ lies on
485: the unit circle only for the `trivial' cases of zero or full
486: transmission $T=0,~1$, i.e., in the absence of partitioning, or for
487: $\lambda=2\pi \mathbb{Z}$; the latter condition corresponds to no
488: counting or the periodic vanishing of decoherence in the qubit. On the
489: contrary, in the case of maximal partitioning with $\langle R\rangle_f
490: = \langle T \rangle_f = 1/2$, a simple phase shift by $\lambda = \pi$
491: makes the fidelity vanish altogether. Hence, partitioning has to be
492: considered as a (purely quantum) source of sensitivity towards small
493: changes, as chaoticity generates sensitivity in a quantum system with
494: a classical analogue.
495:
496: The result (\ref{fid_behind}) also applies for a qubit placed in front
497: of the scattering region provided the coupling is of magnetic nature
498: (for the reflected wave, the additional phases picked up in the
499: interaction region cancel, while the phase in the transmitted part
500: remains uncompensated). However, an electric coupling behaves
501: differently under time reversal and the fidelity acquires the new form
502: %
503: \begin{equation}
504: \chi(\lambda,t) = e^{i \lambda }
505: \bigl(e^{i \lambda} \langle R \rangle_f
506: + \langle T \rangle_f\bigr).\label{fid_before}
507: \end{equation}
508: %
509:
510: Next, we comment on the (velocity) dispersion in the electric coupling
511: $\lambda$: the different components in the wave packet then acquire
512: different phases. To make this point more explicit consider a Gaussian
513: wave packet centered around $k_0$ with a small spreading $\delta k \ll
514: k_0$ and denote with $\lambda_0$ the phase associated with the $k_0$
515: mode. The spreading $\delta k$ in $k$ generates a corresponding
516: spreading in $\delta \lambda \approx \lambda_0 (\delta k/k_0)$ which
517: leads to a reduced fidelity
518: %
519: \begin{equation}
520: \chi(\lambda,t)= \langle R \rangle_f + \langle e^{i \lambda} T
521: \rangle_f \approx R_{k_0} + e^{i \lambda_0 -
522: \frac{\lambda_0^2 (\delta k)^2}{2 k_0^2}} T_{k_0},
523: \label{fid_disp}
524: \end{equation}
525: %
526: where we have assumed a smooth dependence of $T_k$ over $\delta k$ in
527: the last equation. The reduced fidelity for $T = 1$ is due to the
528: acceleration and deceleration produced by the two states of the qubit
529: \cite{ac}. The wave packets passing the qubit then acquire a different
530: time delay depending on the qubit's state. As a result, the wave
531: packets become separated in space with an exponentially small residual
532: overlap for the Gaussian shaped packets.
533:
534: Next, we consider the case with two wave packets incident on the
535: scatterer, $\Psi_\mathrm{in} \propto [\Psi_{f_1}(x_1) \Psi_{f_2}
536: (x_2)\pm (x_1 \leftrightarrow x_2)]\chi_{s/t}(s_1,s_2)$, where
537: $f_{1,2}$ denote different wave packet amplitudes, cf.\
538: (\ref{Psi_in}), and $\chi_{s/t}$ are the singlet/triplet spin
539: functions. Placing the qubit behind the scatterer, we obtain the
540: fidelity
541: %
542: \begin{eqnarray}
543: &&\!\!\chi_2(\lambda,t) = \frac{1}{N_\pm} \Bigl[{\prod_{m=1,2}}
544: \int\! \frac{d k_m}{2\pi} |f_m(k_m)|^2 (R_{k_m} + e^{i \lambda} T_{k_m})
545: \nonumber \\
546: &&\pm \! {\prod_{m=1,2}} \int \! \frac{d k_m}{2\pi}
547: f^*_m (k_m) f_{n \neq m} (k_m) (R_{k_m} + e^{i \lambda} T_{k_m})\Bigr],
548: \label{fid2_behind}
549: \end{eqnarray}
550: %
551: with the normalization $N_\pm=1\pm |S|^2$, $S=\int (dk/2\pi)$
552: $f^*_1(k) f_2 (k)$, and we have made use of the fact that all
553: components of the wave packet propagate in specific directions such
554: that $\int d k f^*_n (k) f_m (-k) = 0$. The fidelity
555: (\ref{fid2_behind}) involves two terms, a direct term independent of
556: the spin/orbital symmetry and an exchange term that can be neglected
557: if the two wave packets are well separated either in momentum space
558: (with $k_{0n}$ sufficiently different) or in real space (with $k_0
559: \delta x \gg 1$). In both of these cases the overlap $S$ vanishes and
560: we arrive at the final result
561: %
562: \begin{equation}
563: \chi_2 (\lambda,t) = [\langle R\rangle_{f_1}
564: + e^{i \lambda} \langle T\rangle_{f_1}]
565: [\langle R\rangle_{f_2} + e^{i \lambda} \langle T\rangle_{f_2}],
566: \label{fid2_final}
567: \end{equation}
568: %
569: i.e., the fidelity of the two-particle system is just the product of
570: the single particle fidelities. The result can be trivially
571: generalized to the many particle case provided that exchange terms can
572: be neglected, as is the case for a sequence of ($N$) properly
573: separated wave packets. The result $\chi_N (\lambda,t) = \Pi_{n=1}^N
574: [\langle R\rangle_{f_n} + e^{i \lambda} \langle T\rangle_{f_n}]$ then
575: has to be compared with the previous finding $\chi_V (\lambda,t) = [R +
576: e^{i \lambda} T]^{e|V|t/h}$ \cite{lc,ll23} calculated at constant
577: voltage $V$ within a many-body formalism. The validity of the latter
578: result is restricted to non-dispersive scattering coefficients $T$ and
579: $R$. Identifying $e|V|t/h$ with the number $N$ of transmitted
580: particles the two results agree. However, we note that the present
581: derivation corresponds to a voltage driven many-body setup where
582: distinct voltage pulses with unit flux $c \int dt \, |V(t)| = \Phi_0$,
583: each transferring one particle, are applied to the system \cite{ll23}.
584: The detailed comparison of these two cases, pulsed versus constant
585: voltage, and the absence of interference terms in the latter case is
586: an interesting problem for further investigation. The extension of the
587: fidelity to the many-body case puts additional emphasis on the {\it
588: relation between sensitivity and partitioning} (see also Ref.\
589: \cite{agam_00}) as any finite partitioning with $T < 1$ now generates a
590: fidelity which is vanishing exponentially in time; the quantity
591: $-\ln[1-4RT\sin^2(\lambda/2)]/\Delta t$ with $\Delta t$ the time
592: separation between two wave packets/voltage pulses then plays the role
593: of the Lyapunov exponent in systems with a classical chaotic
594: correspondent.
595:
596: In summary, we have unified the themes of fidelity and full counting
597: statistics and have shown how to exploit qubits for their measurement.
598: As an application of these ideas, we have recalculated the generating
599: function for the full counting statistics within a wave packet
600: approach. This allows for a more in-depth analysis of the counting
601: problem and helps to shed light on the ongoing discussion \cite{lc}
602: relating different expressions for high-order correlators to
603: variations in the experimental setup.
604:
605: We thank Lev Ioffe for discussions and acknowledge financial support
606: by the CTS-ETHZ, the MaNEP program of the Swiss National Foundation
607: and the Russian Science Support Foundation.
608:
609: \begin{thebibliography}{99}
610:
611: \bibitem{peres} A.\ Peres,
612: Phys.\ Rev.\ A {\bf 30}, 1610 (1984).
613:
614: \bibitem{ll23} L.S.\ Levitov and G.B.\ Lesovik,
615: JETP Lett.\ {\bf 58}, 230 (1993) and cond-mat/9401004 (1994);
616: L.S.\ Levitov, H.\ Lee, and G.B.\ Lesovik,
617: J.\ Math.\ Phys.\ {\bf 37}, 4845 (1996).
618:
619: \bibitem{nielsonchuang} M.A.\ Nielsen and I.L.\ Chuang,
620: {\it Quantum Computation and Quantum Information}
621: (Cambridge University Press, Cambridge, 2000).
622:
623: \bibitem{chiorescu} I.\ Chiorescu, Y.\ Nakamura, C.P.M.\ Harmans,
624: and J.E.\ Mooij, Science {\bf 299}, 1869 (2003).
625:
626: \bibitem{vion} D.\ Vion, A.\ Aassime, A.\ Cottet, P.\ Joyez, H.\
627: Pothier, C.\ Urbina, D.\ Esteve, and M.\ Devoret,
628: Science {\bf 296}, 886 (2002).
629:
630: \bibitem{makhlin} Y.\ Makhlin, G.\ Sch\"on, and A.\ Shnirman,
631: Rev.\ Mod.\ Phys.\ {\bf 73}, 357 (2001).
632:
633: \bibitem{averin} D.V.\ Averin and E.V.\ Sukhorukov,
634: Phys.\ Rev.\ Lett.\ {\bf 95}, 126803 (2005).
635:
636: \bibitem{schoelkopf} R.\ Aguado and L.P.\ Kouwenhoven,
637: Phys.\ Rev.\ Lett.\ {\bf 84}, 1986 (2000);
638: R.J.\ Schoelkopf, A.A.\ Clerk, S.M.\ Girvin, K.W.\ Lehnert,
639: and M.H.\ Devoret, in {\it Quantum Noise in Mesoscopic Physics},
640: ed.\ Y.\ Nazarov (Kluver, Amsterdam, 2003).
641:
642: \bibitem{modsq} The usual definition \cite{peres} involves
643: the square modulus of this quantity.
644:
645: \bibitem{jalabert} R.A.\ Jalabert and H.M.\ Pastawski,
646: Phys.\ Rev.\ Lett.\ {\bf 86}, 2490 (2001).
647:
648: \bibitem{ll1} L.S.\ Levitov and G.B.\ Lesovik,
649: JETP Lett.\ {\bf 55}, 555 (1992).
650:
651: \bibitem{qkr} S.A.\ Gardiner, J.I.\ Cirac, and P.\ Zoller,
652: Phys. Rev. Lett. 80, 2968 (1998); S.\ Montangero,
653: A.\ Romito, G.\ Benenti, and R.\ Fazio,
654: Europhys.\ Lett.\ {\bf 71}, 893 (2005).
655:
656: \bibitem{prober} B.\ Reulet, J.\ Senzier, and D.E.\ Prober,
657: Phys.\ Rev.\ Lett.\ {\bf 91}, 196601 (2003);
658: R.K.\ Lindell, J.\ Delahaye, M.A.\ Sillanpaa, T.T.\
659: Heikkila, E.B.\ Sonin, and P.J.\ Hakonen,
660: Phys.\ Rev.\ Lett.\ {\bf 93}, 197002 (2004);
661: Yu.\ Bomze, G.\ Gershon, D.\ Shovkun, L.S.\ Levitov,
662: and M.\ Reznikov, cond-mat/0504382.
663:
664: \bibitem{lesovik_94} G.B.\ Lesovik,
665: JETP.\ Lett.\ {\bf 60}, 820 (1994);
666: J.P.\ Pekola,
667: Phys.\ Rev.\ Lett.\ {\bf 93}, 206601 (2004);
668: E.B.\ Sonin, Phys.\ Rev. B {\bf 70}, 140506(R) (2004)
669: and cond-mat/0505424.
670:
671: \bibitem{pilgrambuttiker} S.\ Pilgram and M.\ B\"uttiker,
672: Phys.\ Rev.\ B {\bf 67}, 235308 (2003).
673:
674: \bibitem{feigel} M.V.\ Feigel'man, L.B.\ Ioffe, V.B.\ Geshkenbein,
675: P.\ Dayal, and G.\ Blatter,
676: Phys.\ Rev.\ Lett.\ {\bf 92}, 098301 (2004).
677:
678: \bibitem{ac} The fidelity involves the matrix element of wave
679: functions perturbed by opposite states of the qubit.
680:
681: \bibitem{lc} G.B.\ Lesovik and N.M.\ Chtchelkatchev,
682: JETP Lett.\ {\bf 77}, 393 (2003).
683:
684: \bibitem{agam_00} O.\ Agam, I.\ Aleiner, and A.\ Larkin,
685: Phys.\ Rev.\ Lett.\ {\bf 85}, 3153 (2000).
686:
687: \end{thebibliography}
688:
689: \end{document}
690: