* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Quantum process tomography of two-qubit controlled-Z
Orchestrated objective reduction wikipedia , lookup
Matter wave wikipedia , lookup
Canonical quantization wikipedia , lookup
Tight binding wikipedia , lookup
X-ray fluorescence wikipedia , lookup
Quantum dot cellular automaton wikipedia , lookup
Interpretations of quantum mechanics wikipedia , lookup
Quantum key distribution wikipedia , lookup
Coherent states wikipedia , lookup
Theoretical and experimental justification for the Schrödinger equation wikipedia , lookup
Quantum machine learning wikipedia , lookup
Hidden variable theory wikipedia , lookup
Quantum group wikipedia , lookup
Quantum electrodynamics wikipedia , lookup
Quantum computing wikipedia , lookup
Quantum state wikipedia , lookup
Symmetry in quantum mechanics wikipedia , lookup
Density matrix wikipedia , lookup
Quantum decoherence wikipedia , lookup
Two-dimensional nuclear magnetic resonance spectroscopy wikipedia , lookup
Probability amplitude wikipedia , lookup
Algorithmic cooling wikipedia , lookup
PHYSICAL REVIEW B 82, 184515 共2010兲 Quantum process tomography of two-qubit controlled-Z and controlled-NOT gates using superconducting phase qubits T. Yamamoto,1,2 M. Neeley,1 E. Lucero,1 R. C. Bialczak,1 J. Kelly,1 M. Lenander,1 Matteo Mariantoni,1 A. D. O’Connell,1 D. Sank,1 H. Wang,1 M. Weides,1 J. Wenner,1 Y. Yin,1 A. N. Cleland,1,* and John M. Martinis1,† 1 Department of Physics, University of California, Santa Barbara, California 93106, USA Innovation Research Laboratories, NEC Corporation, Tsukuba, Ibaraki 305-8501, Japan 共Received 19 October 2010; published 9 November 2010兲 2Green We experimentally demonstrate quantum process tomography of controlled-Z and controlled-NOT gates using capacitively coupled superconducting phase qubits. These gates are realized by using the 兩2典 state of the phase qubit. We obtain a process fidelity of 0.70 for the controlled phase and 0.56 for the controlled-NOT gate, with the loss of fidelity mostly due to single-qubit decoherence. The controlled-Z gate is also used to demonstrate a two-qubit Deutsch-Jozsa algorithm with a single function query. DOI: 10.1103/PhysRevB.82.184515 PACS number共s兲: 03.67.Lx, 03.67.Ac, 85.25.Cp I. INTRODUCTION Quantum computation and quantum communication rely on excellent control of the underlying quantum system.1 Reasonable control has been achieved with a variety of quantum systems with superconducting qubits emerging as one of the most promising candidates.2 Recent experiments using superconducting architectures include demonstrations of quantum algorithms using two qubits3 and the entanglement of three qubits.4,5 A key element in these experiments is a twoqubit entangling gate, such as the 冑iSWAP 共Ref. 4兲 and the controlled-Z 共CZ兲 gates.3,5 Because the CZ gate is simple to implement, has high fidelity and can readily generate controlled-NOT 共CNOT兲 logic,6 it likely will be an important component in more complex algorithms such as quantum error correction. At present, however, the CZ gate functionality has only been directly tested for a subset of the possible input states. In this paper, we demonstrate the operation of a CZ gate in superconducting phase qubits and fully characterize this gate as well as a CNOT gate using quantum process tomography 共QPT兲. We additionally use the CZ gate to perform the Deutsch-Jozsa algorithm,3 here with a single-shot evaluation of the function. The use of QPT provides a more complete gate evaluation than, for example, measuring the truth table for the corresponding CNOT gate,7,8 as it verifies that the gate will properly transform any possible input state. QPT for two- or three-qubit gates has been reported in NMR,9 optics,10–12 and in ion traps.13,14 In solid-state systems, QPT has been implemented for the 冑iSWAP gate with the phase qubit.15 the shunting capacitance, unequally spaced quantized energy levels appear in the cubic potential. The two lowest levels are used for the qubit states 兩0典 and 兩1典 with a transition A B frequency f 10 共f 10兲 that can be controlled by an external magA B 共⌽ex 兲 applied to the loop. The third energy netic flux ⌽ex level 兩2典 is used as an auxiliary state to realize the CZ gate, as discussed below. The operation of a similar device has been reported previously.15,16 The state of each qubit is controlled by applying a rectangular-shaped current pulse 共Z pulse兲 or a Gaussian-shaped microwave pulse 共X, Y pulse兲 to its bias coil. For an X or Y pulse, we simultaneously apply the derivative of the pulse to the quadrature 共90° phase shifted兲 drive to reduce both unwanted excitation of the 兩2典 state and phase error due to ac Stark effect;17 the derivative scaling factor is determined from the nonlinearity of each qubit.18 This procedure enables us to use a Gaussian pulse with a full width at half maximum of 10 ns while maintaining accurate qubit control19 in spite of a rather weak qubit nonlinearity 共⬃100 MHz兲. Each qubit state is read out individually in a single-shot manner by injecting a large magnitude Z pulse and then measuring the qubit flux with a superconducting quantum interference device 共SQUID兲. The device was fabricated using a photolithographic process with Al films, AlOx tunnel junctions, and a-Si:H dielectric for the shunt capacitors and wiring crossovers, all on a sapphire substrate. The device was mounted in a superconducting aluminum sample holder and cooled in a dilution refrigerator to ⬃25 mK. SQUID A Cc qubit A qubit B CA II. EXPERIMENT The electrical circuit for the device is shown in Fig. 1, comprising two superconducting phase qubits A and B, coupled by a fixed capacitance Cc. Each qubit is a superconducting loop interrupted by a capacitively shunted Josephson junction. When biased close to the critical current, the junction, and its parallel loop inductance produce a nonlinear potential as a function of the phase difference across the junction. Combined with the kinetic energy originating from 1098-0121/2010/82共18兲/184515共8兲 Bias coil A LA Bias coil B SQUID B CB I0A ΦexA I0B ΦexB LB FIG. 1. Circuit diagram for the experimental device, showing two flux-biased phase qubits coupled by a fixed capacitance Cc. A bias coil and readout SQUID are coupled to each qubit. The design B parameters of the circuit are IA 0 = I0 = 2 A, CA = CB = 1 pF, LA = LB = 720 pH, and Cc = 2 fF. 184515-1 ©2010 The American Physical Society PHYSICAL REVIEW B 82, 184515 共2010兲 YAMAMOTO et al. |01› 1.00 0.895 0.900 B Φex /Φ0 t i0 0.0 ∆i (arb.units) 0.2 Probability 0.4 7.25 t0 (b) 7.15 -0.05 0.00 0.05 0.10 ∆i (arb.units) 0.15 FIG. 2. 共Color online兲 High-power spectroscopy for qubit B. The escape probability 共gray scale兲 is plotted versus microwave frequency and the Z-pulse amplitude ⌬i. The single-photon 兩0典 → 兩1典 and two-photon 兩0典 → 兩2典 transitions are visible, along with two avoided-level crossings. The upper inset shows the calculated states and eigenenergies with the thick 共thin兲 lines representing single 共two兲 photon excitations. The lower inset illustrates the Z-pulse amplitude i0 for the CZ= −共SWAP兲2 operation. In the present experiment, the two qubits were biased so A B = 7.16 GHz and f 10 = 7.36 GHz when no Z pulse was that f 10 applied. The relaxation times 共T1兲 were measured to be 510 ns and 500 ns for qubit A and B, respectively. The dephasing times determined from a Ramsey interference experiment 兲, which showed Gaussian decay proportional to 共TRamsey 2 兲2兴 due to 1 / f flux noise,20 were 200 ns and exp关−共t / TRamsey 2 230 ns, respectively. III. RESULTS AND DISCUSSION A. Spectroscopy Figure 2 shows the high-power spectroscopy for qubit B, which is used to guide formation of the CZ gate. We plot the escape probability of qubit B in gray scale as a function of the amplitude ⌬i of a 2 s long Z pulse 共horizontal axis兲 and the frequency of a microwave X pulse 共vertical axis兲 of the same length. Both pulses were applied simultaneously to qubit B, followed by the Z pulse for the readout. In this way, we can probe the change in the resonance frequency B as a function of detuning ⌬i. In addition to the main f 10 B , somewhat broadened resonance line corresponding to f 10 because of the large amplitude of the microwave pulse B ⬃ 10⌽0兲, a sharper line is observed on the low共⌬⌽ex frequency side of the main resonance; this corresponds to the two-photon excitation from the 兩0典 to the 兩2典 state.21 The vertical distance between the main and two-photon lines is 1/2 the qubit nonlinearity ⌬f = f 10 − f 21, yielding ⌬f = 114 MHz for qubit A 共data not shown兲 and 87 MHz for qubit B. We observe an avoided-level crossing in the main resonance at ⌬i ⯝ 0.13 when the two-qubit frequencies overlap B A = f 10 . Here, the degeneracy of the 兩AB典 = 兩10典 and 兩01典 f 10 states produces a splitting with size 14.2⫾ 0.2 MHz, deter- tcmp i0 0.4 0.06 0.0 0.04 100 ∆t (ns) 2 i0 P|2> (d)1.0 0.8 0.08 0 -SWAP qubit B meas. A icmp π/2 t0 π/2 ∆i qubit B |02› qubit A ∆t π |11› π (c) qubit A 200 Probability 7.35 π (a) 1.02 Frequency Frequency (GHz) 7.45 B icmp meas. π pulse off π pulse on 0.5 0.0 -0.04 0.00 B icmp (arb.units) FIG. 3. 共Color online兲 共a兲 Operation sequence for 共b兲, the state evolution between 兩11典 and 兩02典 states. 共b兲 Plot of 兩2典 state probability of qubit B versus Z-pulse time ⌬t and Z-pulse amplitude ⌬i. The dashed lines correspond to the optimal setting for the CZ gate. 共c兲 Operation sequence for 共d兲, demonstration of the CZ gate. 共d兲 Plot of 兩1典 state probability of qubit B as a function of iBcmp 共iA cmp is fixed as 3 ⫻ 10−4兲. The 共red兲 solid and 共blue兲 dashed curves are for qubit A initialized to the 兩0典 and 兩1典 states, respectively. The vertical dotted-dashed line indicates the value of iBcmp for the CZ gate. mined from a fit to the data, consistent with the designed capacitance Cc. The avoided crossing for the two-photon line at ⌬i ⯝ 0.066 gives a splitting of 9.7⫾ 0.2 MHz, about 冑2 / 2 times as large as the main resonance, as expected from a 兩11典 and 兩02典 interaction. The slope of the resonance between the B , as expected for the 兩11典 state. two crossings is 1/2 that of f 10 Our interpretation of the spectroscopy is validated by a numerical calculation. Using three states for each qubit and the qubit design parameters, we calculate from the resulting 9 ⫻ 9 Hamiltonian22,23 the energies for the coupled eigenA , are plotted verstates. The energy bands, normalized to f 01 sus the flux bias for qubit B in the upper inset of Fig. 2. Here, a band is plotted only when its transition matrix element from the ground state is above a threshold, to simulate the appearance of the transition in the spectroscopic measurement. The details of the calculation are described in Appendix A. The 共red兲 thick lines correspond to the 兩01典 and 兩10典 states, whereas the 共blue兲 thin lines represent half of the excitation energy of the 兩11典 and 兩02典 states. The overall structure agrees well with the experimental data. B. Bring up of controlled-Z gate As proposed theoretically by Strauch et al.,24 the avoided crossing due to the degeneracy of the 兩11典 and 兩02典 states can be used to construct a CZ gate, whose action produces no change in state except for 兩11典 → −兩11典. By applying a nonadiabatic Z pulse, the 兩11典 and 兩02典 states become degenerate 共see lower inset of Fig. 2兲. Initially in the 兩11典 state, the system evolves as an iSWAP interaction, giving 兩⌿共t兲典 = cos共␥⌬t / ប兲兩11典 + i sin共␥⌬t / ប兲兩02典, where 2␥ is the splitting energy of the avoided crossing and ⌬t the duration of the Z pulse. After twice the iSWAP time ⌬t = h / 2␥, the system returns to the initial state 兩11典 but with a minus sign. If the system starts in 兩00典, 兩01典, or 兩10典, the state does not change since it is off-resonance with both avoided-level 184515-2 PHYSICAL REVIEW B 82, 184515 共2010兲 QUANTUM PROCESS TOMOGRAPHY OF TWO-QUBIT… Exp. FIG. 4. 共Color online兲 matrices of CZ and CNOT gates. 共a兲 Left panel: the real part of the experimentally obtained matrix 共p兲 for the CZ gate with Fp = 0.70. Right panel: the real part of the simulated matrix for the CZ gate with Fp = 0.67. 共b兲 Left panel: the real part of p for the CNOT gate with Fp = 0.56. Right panel: the real part of the simulated matrix for the CNOT gate with Fp = 0.52. The open boxes in the figure represent the ideal matrix. Sim. Exp. Re(χ) Re(χ) (b) CNOT Sim. Re(χ) Re(χ) (a) CZ crossings. A similar scheme using an adiabatic Z pulse has been used to successfully demonstrate a quantum algorithm3 and the same 共nonadiabatic兲 scheme has recently been used to create a three-qubit entangled state in transmon qubits.5 To experimentally determine the amplitude and length of the required nonadiabatic Z pulse, we directly measured the coherent oscillation between the 兩11典 and 兩02典 states. This 共iSWAP兲2 operation sequence is shown in Fig. 3共a兲: we first prepare the 兩11典 state with a pulse to both qubits and then apply a Z pulse with amplitude ⌬i and length ⌬t to qubit B. Here only qubit B is probed and we adjust the measurement pulse amplitude so that the qubit is detected only when in the 兩2典 共or higher兲 state.25 In Fig. 3共b兲, we plot the tunneling probability P兩2典 as a function of ⌬t and ⌬i, which shows the expected chevron pattern. The minimum oscillation frequency occurs at a value of ⌬i that agrees with i0 determined in Fig. 2共a兲. The oscillation period t0 = 51.8 ns is also consistent with the splitting size of the avoided crossing. At the intersection of these two dashed lines, the time evolution of the state produces a minus sign, as required for the CZ gate. We stress that no discernable increase in P兩2典 is observed 共⬍1%兲 at this operation point 共⌬t , ⌬i兲 = 共t0 , i0兲, confirming that we return to the 兩11典 state after the CZ operation. Because the qubits themselves also accumulate phase during the CZ pulse, the general unitary evolution from the gate is given by U= 冢 1 0 0 0 0 e iA 0 0 0 0 iB 0 0 0 e 0 − ei共A+B兲 冣 . 共1兲 By adding additional Z pulses to both qubits, we can compensate these phases and even place the minus sign at any diagonal position in the matrix.3 The compensation pulses are shown in Fig. 3共c兲, which consist of a fixed 10 ns pulse of variable amplitude icmp after the CZ pulse. In Fig. 3共d兲, we B plot the tunneling probability of qubit B as a function of icmp A for fixed icmp. The phase of qubit B is measured through a Ramsey fringe experiment. The 共red兲 solid and 共blue兲 dashed curves correspond to qubit A being in the 兩0典 or 兩1典 state. B , but are They both show a sinusoidal dependence on icmp shifted by from each other, confirming the correct operation of the CZ gate. A similar experiment was done for qubit A 共data not shown兲. The phases for the CZ gate are set by taking the values of icmp that give maximum probability when the control qubit is in the 兩0典 state, as indicated by the vertical dashed-dotted line in Fig. 3共d兲. Controlled-NOT 共CNOT兲 gates are constructed by combining the CZ gate with single-qubit rotations UCNOT = 共I 丢 Ry /2兲CZ共I 丢 R−y /2兲, where Ry represents the rotation of a single-qubit state by an angle about the y axis and I is the identity operator. 184515-3 PHYSICAL REVIEW B 82, 184515 共2010兲 YAMAMOTO et al. FIG. 5. 共Color online兲 共a兲 Pulse sequence for the Deutsch-Jozsa algorithm. 关共b兲–共d兲兴 Real part of the density matrix of the final state for four Deutsch-Jozsa functions. Open boxes represent the ideal density matrix. C. Quantum process tomography We evaluate the performance of these gates with QPT in which we determine the matrix with the elements defined as6 16 E共兲 = 兺 EmE†nmn . 共2兲 m,n Here, E共兲 is the density matrix obtained by applying the gate to and Em’s are the operator bases formed by the Kronecker product of Pauli operators 兵I , x , −iy , z其 for each qubit. For QPT, we prepare 16 input states in total, chosen from the set 兵兩0典 , 兩1典 , 兩0典 + 兩1典 , 兩0典 + i兩1典其 for each qubit. After preparing these input states, we determine the density matrix of the output state with quantum state tomography16 共QST兲 in which we measure each qubit along the six directions ⫾x, ⫾y, and ⫾z of the Bloch sphere.26 For each combination of QPT and QST pulses, we repeat the sequence 1800 times to obtain the joint qubit probabilities PAB = P00, P10, P01, and P11. After correcting for small measurement errors,15 we reconstruct the 16⫻ 16 experimental e matrix from the resulting 16 density matrices.6 We assumed that input states are ideally prepared.27 With experimental noise, the e matrix found in this way is not necessarily physical, i.e., completely positive and trace preserving. We thus use convex optimization to obtain the physical matrix p that best approximates e, as used in Ref. 15. The difference between e and p is small as shown in Appendix B. We plot the real part of p for the CZ and CNOT gates in the left panel of Figs. 4共a兲 and 4共b兲. The open boxes represent the ideal matrix. The imaginary parts of p have very small magnitude 共⬍0.04 for CZ and ⬍0.03 for CNOT兲 and are shown in Appendix C. For both gates, we observe elements with large amplitudes at the proper positions. More quantitative evaluation is obtained by calculating the process fidelity Fp, defined by Fp = Tr共ip兲, where i represents an ideal matrix. We obtain Fp = 0.70 for the experimentally measured CZ gate and 0.56 for the CNOT gate. For CZ gates with a minus sign at other positions on the diagonal, the measured Fp’s are 0.68, 0.69, and 0.70 for CZ00, CZ01, and CZ10 共see Appendix C兲. To understand the loss of process fidelity, we performed numerical simulations. We solved the standard master equation, ˙ = −共i / ប兲关H , 兴 + L关兴, where H is a 9 ⫻ 9 Hamiltonian for capacitively coupled phase qubits under rotating wave approximation and L关兴 1 i† i 1 i† i = 兺i=A,B兺 j=1,2LijLi† − L L − L L . Here, for example, j j j j 2 j 2 A † 冑 A 冑 A LA 1 = aA / T1 and L2 = aAaA 2 / T2 describe the relaxation and dephasing for qubit A, respectively.28 The entire sequence, including the QPT and QST pulses, was simulated to construct sim. Experimental Ramsey interference shows Gaussian decay, which is not reproduced by the above master equation. Thus, in order to approximate this situation, we used an effective T2 that depends on the length of the control sequence for a particular experiment tseq. In particular, we 2 / tseq in the simulation in order for both the used T2 = TRamsey 2 Gaussian decay and exponential decay to give the same decay factor at tseq. The real part of sim is shown in Fig. 4. The actual tseq is 101.8 ns in QPT for CZ gates and 141.8 ns for CNOT gate. The simulation reproduces reasonably well the reduction in the expected elements and the appearance of TABLE I. Summary of performance for Deutsch-Jozsa algorithm. Deutsch-Jozsa functions are defined as f 0共x兲 = 0, f 1共x兲 = 1, f 2共x兲 = x, and f 3共x兲 = 1 − x. Deutsch-Jozsa function Constant Element 具00兩兩00典 + 具01兩兩01典 具10兩兩10典 + 具11兩兩11典 Ideal Measured Ideal Measured Balanced f0 f1 f2 f3 0 0.29 1 0.71 0 0.28 1 0.72 1 0.76 0 0.24 1 0.74 0 0.26 184515-4 PHYSICAL REVIEW B 82, 184515 共2010兲 QUANTUM PROCESS TOMOGRAPHY OF TWO-QUBIT… (a) 60 (b) CZ11 50 σ = 0.0020 40 60 # of counts # of counts small unwanted elements. These imperfections are removed as we increase the single-qubit coherence time in the simulation, which suggests that loss of Fp in our system is mostly dominated by single-qubit decoherence.29 We note that it is possible to obtain more information on the decoherence mechanisms by analyzing the magnitude of particular elements in the matrix.30 The simulated matrix of all CZ and CNOT gates including the imaginary parts are shown in Appendix D. 30 20 By using these conditional gates, we can perform the Deutsch-Jozsa algorithm6 using the pulse sequence described in Ref. 3. Figure 5共a兲 shows the pulse sequence for the Deutsch-Jozsa algorithm. The four different two-qubit gates Ui correspond to the four Deutsch-Jozsa functions, which we want to determine by a single-quantum evaluation of the function. They are given by U1 = I 丢 Rx , 0 0 -10 -5 0 5 -3 Diff. in peak height (10 ) FIG. 6. 共Color online兲 Histogram of the differences in the peak height of each of the 256 matrix elements in the real part of matrix. Data is for 共a兲 CZ and 共b兲 CNOT. The solid curves are a Gaussian fit to the data. APPENDIX A: CALCULATION OF THE ENERGY BANDS AND TRANSITION MATRIX ELEMENTS U2 = 共I 丢 Ry /2Rx兲CZ00共I 丢 Ry /2兲, U3 = 共I 20 We calculated the energy band of the capacitively coupled flux-biased phase qubits by diagonalizing the following 9 ⫻ 9 Hamiltonian:22,23 U0 = I 丢 I, −/2 丢 R y Rx 兲CZ11共I 丢 40 10 -4 0 4 -3 Diff. in peak height (10 ) D. Deutsch-Jozsa algorithm CNOT σ = 0.0017 R−y /2兲. H⯝ 共3兲 The sequence is same as that used in Ref. 3 except that Ry /2 pulse was not applied to qubit B before the tomography. This makes the final state a superposition state. Also, in order to shorten the total sequence time, the last Ry /2 on qubit A was applied before the Ui part finishes. The real part of the final density matrices are plotted in Figs. 5共b兲–5共e兲. The experimental probability to obtain the correct answer is summarized in Table I. Because our phase qubit has single-shot readout, we can obtain the correct answer to a single function query more than 70% of the time, greater than the 50% probability for a classical query and guess. We stress that no calibration for the measurement error is applied here. IV. CONCLUSIONS In conclusion, we have demonstrated CZ and CNOT gates in capacitively coupled phase qubits using the higher energy 兩2典 state. Quantum process tomography measures a matrix that is in good accord with predictions, which is a definitive test of proper gate operation for any input state. 冢 0 丢 丢 共A兲 hf 10 共A兲 共A兲 hf 10 + hf 21 冢 冢 冣 0 共B兲 hf 10 1 0 0 冑2 冣冢 0 −1 −g 1 共B兲 共B兲 hf 10 + hf 21 0 −1 丢 I2 + I1 0 冣 − 冑2 , 0 0 0 冑2 0 − 冑2 0 冣 共A1兲 共B兲 where f 共A兲 i,j 共f i,j 兲 is the flux-dependent transition frequency between ith and jth state of the qubit A 共B兲 and g is the coupling energy between the qubits. The last term in the Hamiltonian is based on yy-type coupling of the two qubits. To calculate the transition matrix from the ground state 兩g典 to the excited state 兩e典 in the spectroscopy experiment, we calculated the transition matrix element of 兩具e兩a† + a兩g典兩2 for 具e兩a†+a兩i典具i兩a†+a兩g典 one-photon excitation and 兩兺i Ee−Ei−hf d 兩2 for two-photon excitation,31 where a共a†兲 is an annihilation 共creation兲 operator for the harmonic oscillator, Ei is the energy gap of the state 兩i典 from the ground state, and f d is the frequency of the -wave drive, which is set to be Ee / 2h in the calculation. APPENDIX B: DIFFERENCE BETWEEN p AND e ACKNOWLEDGMENTS The authors would like to thank F. K. Wilhelm and A. N. Korotkov for valuable discussion. They would also like to thank Y. Nakamura for useful comments on the manuscript. Semidefinite programming convex optimization was carried out using the open-source MATLAB packages YALMIP and SeDuMi. This work was supported by IARPA under ARO, Award No. W911NF-04-1-0204. We checked the difference between e 共the experimental matrix兲 and p 共the physical matrix兲 by histogramming the differences in the peak height ⌬ = e − p of each of the 256 matrix elements in the real part.10 We fit it by Gaussian a exp共−⌬2 / 2兲 as shown in Fig. 6. The obtained are 0.0020 for CP11 and 0.0017 for CNOT gate, which implies e and p are close. 184515-5 PHYSICAL REVIEW B 82, 184515 共2010兲 YAMAMOTO et al. FIG. 7. 共Color online兲 p of 共a兲 CZ00, 共b兲 CZ01, 共c兲 CZ10, 共d兲 CZ = CZ11, and 共e兲 CNOT. Process fidelity Fp are 0.68, 0.69, 0.70, 0.70, and 0.56, respectively. 184515-6 PHYSICAL REVIEW B 82, 184515 共2010兲 QUANTUM PROCESS TOMOGRAPHY OF TWO-QUBIT… FIG. 8. 共Color online兲 Simulated matrix of 共a兲 CZ00, 共b兲 CZ01, 共c兲 CZ10, 共d兲 CZ, and 共e兲 CNOT. Process fidelity Fp are 0.67, 0.67, 0.66, 0.67, and 0.52, respectively. 184515-7 PHYSICAL REVIEW B 82, 184515 共2010兲 YAMAMOTO et al. APPENDIX C: MATRIX FOR ALL GATES APPENDIX D: SIMULATED MATRIX FOR ALL GATES In Fig. 7, the physical matrix p is plotted for all the CZ and CNOT gates. In Fig. 8, the simulated matrix is plotted for all CZ and CNOT gates. 16 *[email protected] †[email protected] 1 T. D. Ladd, F. Jelezko, R. Laflamme, Y. Nakamura, C. Monroe, and J. L. O’Brien, Nature 共London兲 464, 45 共2010兲. 2 J. Clarke and F. K. Wilhelm, Nature 共London兲 453, 1031 共2008兲. 3 L. DiCarlo, J. M. Chow, J. M. Gambetta, L. S. Bishop, B. R. Johnson, D. I. Shuster, J. Majer, A. Blais, L. Frunzio, S. M. Girvin, and R. J. Schoelkopf, Nature 共London兲 460, 240 共2009兲. 4 M. Neeley, R. C. Bialczak, M. Lenander, E. Lucero, M. Mariantoni, A. D. O’Connell, D. Sank, H. Wang, M. Weides, J. Wenner, Y. Yin, T. Yamamoto, A. N. Cleland, and J. M. Martinis, Nature 共London兲 467, 570 共2010兲. 5 L. DiCarlo, M. D. Reed, L. Sun, B. R. Johnson, J. M. Chow, J. M. Gambetta, L. Frunzio, S. M. Girvin, M. H. Devoret, and R. J. Schoelkopf, Nature 共London兲 467, 574 共2010兲. 6 M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information 共Cambridge University Press, Cambridge, England, 2000兲. 7 T. Yamamoto, Y. A. Pashkin, O. Astafiev, Y. Nakamura, and J. S. Tsai, Nature 共London兲 425, 941 共2003兲. 8 J. H. Plantenberg, P. C. de Groot, C. J. P. M. Harmans, and J. E. Mooij, Nature 共London兲 447, 836 共2007兲. 9 A. M. Childs, I. L. Chuang, and D. W. Leung, Phys. Rev. A 64, 012314 共2001兲. 10 J. L. O’Brien, G. J. Pryde, A. Gilchrist, D. F. V. James, N. K. Langford, T. C. Ralph, and A. G. White, Phys. Rev. Lett. 93, 080502 共2004兲. 11 N. K. Langford, T. J. Weinhold, R. Prevedel, K. J. Resch, A. Gilchrist, J. L. O’Brien, G. J. Pryde, and A. G. White, Phys. Rev. Lett. 95, 210504 共2005兲. 12 N. Kiesel, C. Schmid, U. Weber, R. Ursin, and H. Weinfurter, Phys. Rev. Lett. 95, 210505 共2005兲. 13 M. Riebe, K. Kim, P. Schindler, T. Monz, P. O. Schmidt, T. K. Körber, W. Hänsel, H. Häffner, C. F. Roos, and R. Blatt, Phys. Rev. Lett. 97, 220407 共2006兲. 14 T. Monz, K. Kim, W. Hänsel, M. Riebe, A. S. Villar, P. Schindler, M. Chwalla, M. Hennrich, and R. Blatt, Phys. Rev. Lett. 102, 040501 共2009兲. 15 R. C. Bialczak, M. Ansmann, M. Hofheinz, E. Lucero, M. Neeley, A. D. O’Connell, D. Sank, H. Wang, J. Wenner, M. Steffen, A. N. Cleland, and J. M. Martinis, Nat. Phys. 6, 409 共2010兲. M. Steffen, M. Ansmann, R. C. Bialczak, N. Katz, E. Lucero, R. McDermott, M. Neeley, E. M. Weig, A. N. Cleland, and J. M. Martinis, Science 313, 1423 共2006兲. 17 F. Motzoi, J. M. Gambetta, P. Rebentrost, and F. K. Wilhelm, Phys. Rev. Lett. 103, 110501 共2009兲. 18 E. Lucero, J. Kelly, R. C. Bialczak, M. Lenander, M. Mariantoni, M. Neeley, A. D. O’Connell, D. Sank, H. Wang, M. Weides, J. Wenner, T. Yamamoto, A. N. Cleland, and J. M. Martinis, Phys. Rev. A 82, 042339 共2010兲. 19 E. Lucero, M. Hofheinz, M. Ansmann, R. C. Bialczak, N. Katz, M. Neeley, A. D. O’Connell, H. Wang, A. N. Cleland, and J. M. Martinis, Phys. Rev. Lett. 100, 247001 共2008兲. 20 R. C. Bialczak, R. McDermott, M. Ansmann, M. Hofheinz, N. Katz, E. Lucero, M. Neeley, A. D. O’Connell, H. Wang, A. N. Cleland, and J. M. Martinis, Phys. Rev. Lett. 99, 187006 共2007兲. 21 P. Bushev, C. Müller, J. Lisenfeld, J. Cole, A. Lukashenko, A. Shnirman, and A. Ustinov, Phys. Rev. B 82, 134530 共2010兲. 22 M. Steffen, J. M. Martinis, and I. L. Chuang, Phys. Rev. B 68, 224518 共2003兲. 23 A. G. Kofman, Q. Zhang, J. M. Martinis, and A. N. Korotkov, Phys. Rev. B 75, 014524 共2007兲. 24 F. W. Strauch, P. R. Johnson, A. J. Dragt, C. J. Lobb, J. R. Anderson, and F. C. Wellstood, Phys. Rev. Lett. 91, 167005 共2003兲. 25 M. Neeley, M. Ansmann, R. C. Bialczak, M. Hofheinz, E. Lucero, A. D. O’Connell, D. Sank, H. Wang, J. Wenner, A. N. Cleland, M. R. Geller, and J. M. Martinis, Science 325, 722 共2009兲. 26 M. Neeley, M. Ansmann, R. C. Bialczak, M. Hofheinz, E. Lucero, A. D. O’Connell, D. Sank, H. Wang, J. Wenner, A. N. Cleland, M. R. Geller, and J. M. Martinis, Nat. Phys. 4, 523 共2008兲. 27 For identity operation, namely, QST right after the preparation pulses, we obtained Fp of 0.91. 28 D. F. Walls and G. J. Milburn, Phys. Rev. A 31, 2403 共1985兲. 29 According to the simulation, roughly half of the fidelity loss is caused by pure dephasing 共T2兲 both in CZ and CNOT gates. 30 A. G. Kofman and A. N. Korotkov, Phys. Rev. A 80, 042103 共2009兲. 31 C. Cohen-Tannoudji, J. Dupont-Roc, and G. Grynberg, AtomPhoton Interaction 共Wiley, New York, 1992兲 Chap. IIIC. 184515-8