Abstract
We present a general quantum circuit design for finding eigenvalues of non-unitary matrices on quantum computers using the iterative phase estimation algorithm. In addition, we show how the method can be used for the simulation of resonance states for quantum systems.
Similar content being viewed by others
Notes
This matrix is given by Pablo Serra, private communication, 2012 [37].
References
Harrow, A.W., Hassidim, A., Lloyd, S.: Quantum algorithm for linear systems of equations. Phys. Rev. Lett. 103, 150502 (2009). doi:10.1103/PhysRevLett.103.150502
Cao, Y., Daskin, A., Frankel, S., Kais, S.: Quantum circuit design for solving linear systems of equations. Mol. Phys. 110(15–16), 1675 (2012). doi:10.1080/00268976.2012.668289
Deutsch, D.: Quantum theory, the Church-Turing principle and the universal quantum computer. Proc. R. Soc. Lond. A Math. Phys. Sci. 400(1818), 97 (1985). doi:10.1098/rspa.1985.0070
Bennett, C.H., Bernstein, E., Brassard, G., Vazirani, U.: Strengths and weaknesses of quantum computing. SIAM J. Comput. 26(5), 1510 (1997). doi:10.1137/S0097539796300933
Martin-Delgado, M.A.: On quantum effects in a theory of biological evolution. Sci. Rep. 2, 302 (2012)
Abrams, D.S., Lloyd, S.: Quantum algorithm providing exponential speed increase for finding eigenvalues and eigenvectors. Phys. Rev. Lett. 83(24), 5162 (1999). doi:10.1103/PhysRevLett.83.5162
Papageorgiou, A., Zhang, C.: On the efficiency of quantum algorithms for Hamiltonian simulation. Quantum Inf. Process. 11(2), 541 (2012). doi:10.1007/s11128-011-0263-9
Finkelstein, D.. In Gudehus, T., Kaiser, G., Perl- mutter, A. (eds.) Coral Gables Conference on Fundamental Interactions at High Energy. Center of Theoretical Studies. Gordon and Breach, New York (1969)
Feynman, R.: Simulating physics with computers. Int. J. Theor. Phys. 21, 467 (1982). doi:10.1007/BF02650179
Sanders, B.C.: Algorithm for quantum simulation. Appl. Math. Inf. Sci 3(2), 117 (2009)
Raeisi, S., Wiebe, N., Sanders, B.C.: Quantum-circuit design for efficient simulations of many-body quantum dynamics. New J. Phys. 14(10), 103017 (2012)
Kassal, I., Jordan, S.P., Love, P.J., Mohseni, M., Aspuru-Guzik, A.: Polynomial-time quantum algorithm for the simulation of chemical dynamics. Proc. Natl. Academ. Sci. 105(48), 18681 (2008)
Berry, D.W., Ahokas, G., Cleve, R., Sanders, B.C.: Efficient quantum algorithms for simulating sparse Hamiltonians. Commun. Math. Phys. 270(2), 359 (2007)
Lidar, D., Wang, H.: Calculating the thermal rate constant with exponential speedup on a quantum computer. Phys. Rev. E 59, 2429 (1999)
Brown, K.L., Munro, W.J., Kendon, V.M.: Using quantum computers for quantum simulation. Entropy 12, 2268–2307 (2010)
Lu, D., Xu, B., Xu, N., Li, Z., Chen, H., Peng, X., Xu, R., Du, J.: Quantum chemistry simulation on quantum computers: theories and experiments. Phys. Chem. Chem. Phys. 14, 9411 (2012). doi:10.1039/C2CP23700H
Kassal, I., Whitfield, J.D., Perdomo-Ortiz, A., Yung, M.H., Aspuru-Guzik, A.: Annual Review of Physical Chemistry 62(1), 185 (2011). doi:10.1146/annurev-physchem-032210-103512. PMID: 21166541
Young, K.C., Sarovar, M., Aytac, J., Herdman, C., Whaley, K.B.: Finite temperature quantum simulation of stabilizer Hamiltonians. J. Phys. B Atomic Mol. Opt. Phys. 45(15), 154012 (2012)
Kitaev, A.: Quantum measurements and the Abelian stebilizer problem. Electronic Colloquium on Computational Complexity (ECCC) 3(3)(1996)
Aspuru-Guzik, A., Dutoi, A., Love, P., Head-Gordon, M.: Simulated quantum computation of molecular energies. Science 309, 1704 (2005)
Wang, H., Kais, S., Aspuru-Guzik, A., Hoffmann, M.: Quantum algorithm for obtaining the energy. Phys. Chem. Chem. Phys. 10, 5388 (2008)
Veis, L., Pittner, J.: Quantum computing applied to calculations of molecular energies: CH benchmark. J. Chem. Phys. 133, 194106 (2010)
Daskin, A., Kais, S.: Decomposition of unitary matrices for finding quantum circuits: application to molecular Hamiltonians. J. Chem. Phys. 134(14), 144112 (2011). doi:10.1063/1.3575402
Lanyon, B.P., Whitfield, J.D., Gillett, G.G., Goggin, M.E., Almeida, M.P., Kassal, I., Biamonte, J.D., Mohseni, M., Powell, B.J., Barbieri, M., Aspuru-Guzik, A., White, A.G.: Towards quantum chemistry on a quantum computer. Nat. Chem. 2(2), 106 (2010). doi:10.1038/nchem.483
Aspuru-Guzik, A., Walther, P.: Photonic quantum simulators. Nat. Phys. 8(4), 285 (2012). doi:10.1038/nphys2253
Du, J., Xu, N., Peng, X., Wang, P., Wu, S., Lu, D.: NMR implementation of a molecular hydrogen quantum simulation with adiabatic state preparation. Phys. Rev. Lett. 104, 030502 (2010). doi:10.1103/PhysRevLett.104.030502
Moiseyev, N.: Non-Hermitian Quantum Mechanics. Cambridge University Press, Cambridge (2011)
Terashima, H., Ueda, M.: Nonunitary quantum circuit. Int. J. Quantum Inf. 03(04), 633 (2005). doi:10.1142/S0219749905001456
Wang, H., Wu, L., Liu, Y., Nori, F.: Measurement-based quantum phase estimation algorithm for finding eigenvalues of non-unitary matrices. Phys. Rev. A 82, 062303 (2010). doi:10.1103/PhysRevA.82.062303
Terashima, H.: Phys. Rev. A 85, 022124 (2012). doi:10.1103/PhysRevA.85.022124
Daskin, A., Grama, A., Kollias, G., Kais, S.: Universal programmable quantum circuit schemes to emulate an operator. J. Chem. Phys. 137(23), 234112 (2012). doi:10.1063/1.4772185
Dobšíček, M., Johansson, G., Shumeiko, V., Wendin, G.: Arbitrary accuracy iterative quantum phase estimation algorithm using a single ancillary qubit: a two-qubit benchmark. Phys. Rev. A 76(3), 030306 (2007)
Usami, K., Nambu, Y., Tsuda, Y., Matsumoto, K., Nakamura, K.: Accuracy of quantum-state estimation utilizing Akaike’s information criterion. Phys. Rev. A 68, 022314 (2003). doi:10.1103/PhysRevA.68.022314
Bogdanov, Y.I., Brida, G., Genovese, M., Kulik, S.P., Moreva, E.V., Shurupov, A.P.: Statistical estimation of the efficiency of quantum state tomography protocols. Phys. Rev. Lett. 105, 010404 (2010). doi:10.1103/PhysRevLett.105.010404
Nielsen, M.A., Chuang, I.L.: Quantum Computation and Quantum Information. Cambridge University Press, Cambridge (2000)
Golub, G.H., Van Loan, C.F.: Matrix computations, 3rd edn. Johns Hopkins University Press, Baltimore, MD, USA (1996)
Serra, P., Kais, S., Moiseyev, N.: Crossover phenomena and resonances in quantum systems. Phys. Rev. A 64, 062502 (2001). doi:10.1103/PhysRevA.64.062502
Möttönen, M., Vartiainen, J.J., Bergholm, V., Salomaa, M.M.: Quantum circuits for general multiqubit gates. Phys. Rev. Lett. 93, 130502 (2004). doi:10.1103/PhysRevLett.93.130502
Acknowledgments
We thank Pablo Serra for providing the Hamiltonian matrix used as an example in this paper. We also thank the two anonymous referees for their help in improving the clarity of this paper. This work is supported by the NSF Centers for Chemical Innovation: Quantum Information for Quantum Chemistry, CHE-1037992.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1: Verification of the circuit in Fig. 2
The circuit is divided in three blocks: Formation, Combination, and Input Modification as shown in the figure. In the Input Modification block, the initial input is modified to \(|\varphi \rangle \). Then, the matrix \(V\) is constructed in the Formation and Combination blocks.
1.1 Formation block
The second block in Fig. 2 is called Formation Block which consists of uniformly controlled rotation gates for each element of \(U\). For the matrix element \(u_{ij}\), assumed to be real, the rotation gate \(R_{ij}(\theta _{ij})\) is defined as follows:
where \(\theta _{ij}\) is determined from the value of the element in accordance with the equality \(cos(\frac{\theta _{ij}}{2})=u_{ij}\) to get the following:
Based on the linear indices of the elements (row-wise), \(R_{ij}\)s are controlled by the different states of \((2n)\) qubits: e.g., \(u_{12}\) is the second element in the matrix and has a linear index of \(2\). Hence, the control of \(R_{12}\) is set accordingly so that this gate operates when the first \((2n)\) qubits in \(\left| 0\dots 01\right\rangle \) state. Combination of these controlled rotation gates forms a network as the second block in Fig. 2. This block has the following matrix representation in the computational basis:
Here, we have used rotations around the y-axis since the matrix elements are assumed to be real. However, for the complex elements of \(U\), a product of \(R_z\) and \(R_y\) gates is needed for each \(R_{ij}\). For instance, if \(u_{ij} = e^{i\phi }cos(\theta ),\,e^{i\phi }\) is created by \(R_z\), and \(cos(\theta )\) is by \(R_y\).
1.2 Combination block
For a system of \((2n+1)\) qubits, the third block in Fig. 2 is defined as the application of the Hadamard gates to \((n+1)\hbox {th }, (n+2)\hbox {th }, \ldots , (2n-1)\hbox {th}\), and \((2n)\)th qubits from the top of the circuit: i.e. \((I^{\otimes n}\otimes H^{\otimes n}\otimes I)\), where \(H\) and \(I\) are the Hadamard and identity matrices, respectively. The matrix representation of this block is as follows:
Note that \(C_{block}\) also has negative elements, however, they are not located in the first row. Therefore, they shall not affect the predetermined states. The application of the above matrix \(C\) to the matrix \(F\) defined in Eq. (29) forms the matrix \(V\) defined in Eq. (3) where the same row elements of \(U\) are located on the leading rows of each \(V_i\) (\(i\) represents the row index of \(U\)):
Since the negative elements are not in the first row of \(C_{block}\), the resulting leading rows are not affected by these elements. The elements represented by the symbol “\(\bullet \)” are disregarded by modifying the input to the circuit in the modification block explained below.
1.3 Input modification block
The initial input to the circuit in Fig. 2 is defined as \(\tilde{\left| \alpha \right\rangle }=\left| 0..0\right\rangle \otimes \left| \alpha \right\rangle \) where \(\left| 0\ldots 0\right\rangle \) is the input to the ancilla qubits. The first block in Fig. 2 consists of the Hadamard gates on the first \(n\) qubits and sequential swap operations between the \((n+1)\)th and the remaining last \(n\) qubits (We apply swap operations between the ancilla \((n+1)\)th qubit and the main qubits: First we swap the \((n+1)\)th qubit with the \((2n+1)\)th, then the \((n+1)\)th with the \((2n)\)th, then the \((n+1)\)th with the \((2n-1)\)th, and so on. Finally, we swap the \((n+1)\)th with the \((n+2)\)th). This block modifies the input \(\tilde{\left| \alpha \right\rangle }\) in a way that in the application of \(V=CF\) to the input, the output is not affected by the elements represented by “\(\bullet \)” between \(ku_{ij}\) and \(ku_{i(j+1)}\). Hence, the application of this block to the initial input transforms \(\tilde{\left| \alpha \right\rangle }\) to \(\left| \varphi \right\rangle \) in Eq. (3):
where \(\gamma \) is a normalization constant.
Consequently, the circuit in Fig. 2 describes the operation \(V\left| \varphi \right\rangle =CF\left| \varphi \right\rangle \) which simulates any matrix \(U\), having elements less than or equal to 1, on the following normalized set of \(N\) states: \(\{\left| 0\dots 000\right\rangle _a\left| 0\dots 0\right\rangle ,\,\dots ,\,\left| 0\dots 110\right\rangle _a\left| 0\dots 0\right\rangle \}\), where \(\left| \dots \right\rangle _a\) represents the ancilla qubits. This is shown in Eq. (3).
For illustration purposes, we also present the full forms of the operators for the formation, combination and input modification blocks, and the output vector for the simulation of the following \(2 \times 2\) arbitrary matrix [31]:
The full form of the matrix for the formation block is as follows:
The combination matrix \(C\) and the matrix for the input modification \(M\) are defined as:
For the initial input \(\left| \mathbf{0}\right\rangle \left| \alpha \right\rangle =\hat{\left| \alpha \right\rangle }\), we find the following final state:
Clearly, the normalized states \(|000\rangle \) and \(|100\rangle \) simulate the original given system.
Appendix 2: Simulation details
1.1 The decomposition of a multi-controlled network
The circuit in Fig. 2 includes a network of rotation gates in the formation block which dominates the complexity of the circuit. The uniformly controlled networks such as the one in Fig. 5a controlled by \(k\) qubits can be decomposed in terms of \(2^k\) CNOT gates and \(2^k\) single rotation gates [38]. For instance, the circuit as illustrated for \(k=2\) in Fig. 5a can be decomposed as in Fig. 5b.
The angle values in the decomposed circuit are solutions of the system of the linear equation \(M^{k}{\varvec{\theta }}={\varvec{\phi }}\):
where \(k\) is the number of control qubits in the network, and the entries of \(M\) are defined as:
in which the power term is found by taking the dot product of the standard binary code of the index \(i-1,\,b_{i-1}\) and the binary representation of \(j-1\)th Gray-coded integer, \(g_{j-1}\). Since \(M^k\) is a column permuted version of the Hadamard matrix, we see that \(M\) is unitary. Thus, \((M^k)^{-1}=2^{-k}(M^k)^T\) and the new angle values in the decomposed circuit are results of the matrix vector multiplication [38]:
1.2 Simulation details for the example system
In Fig. 4, we have a multi-controlled network composed of 4 gates. This network basically comes from the Formation step of the circuit design method, which has been represented in matrix form as \(F\) in Eq. (29). Since the elements of \(U\) are complex, we need to have rotations around the \(z-\)axis and the \(y-\)axis. Hence, the above matrix is the product of two matrices \(F_z\) for rotation around the z-axis and \(F_y\) for rotations around the y-axis: \(F=F_zF_y\):
To complete this network to a whole uniformly controlled network, we assume that the initial four other gates are identity. Hence, the final decomposition for \(2 \times 2\) matrix, the decomposed circuit includes: 4 Hadamard gates, 16 CNOT and 16 single gates (8 CNOTs and 8 \(R_y\) gates for the \(F_y\); and 8 CNOTs and 8 \(R_z\) gates for the \(F_z\)), and 2 swaps.
The parameters for the last iteration (The operator is \(U^{2^0}\).) of the phase estimation algorithm is shown below, where after scaling the elements of \(U\) by the matrix norm \(||U||_1\) (maximum of the absolute sums of the columns), we find the angle values for \(R_y\) and \(R_z\) gates:
Matrix Elements | Scaled Elements | Angles for \(R_y\)s | Angles for \(R_z\)s |
---|---|---|---|
0.2588 \(+\) 1.1214i | 0.1469 \(+\) 0.6364i | 1.0521 | \(-\)0.9634 |
\(-\)0.4569 \(-\) 0.1109i | \(-\)0.2593 \(-\) 0.0629i | 0.0278 | 0.1837 |
\(-\)0.4569 \(-\) 0.1109i | \(-\)0.2593 \(-\) 0.0629i | \(-\)0.2486 | 1.9401 |
1.0594 \(+\) 0.7394i | 0.6012 \(+\) 0.4196i | 0.0278 | 0.1837 |
\(-\)0.0278 | \(-\)0.1837 | ||
0.2486 | \(-\)1.9401 | ||
\(-\)0.0278 | \(-\)0.1837 | ||
\(-\)1.0521 | 0.9634 |
The angle value for the \(R_z\) gate on the first qubit is \(-\)2.51572849. The output of the phase estimation algorithm on the chosen states and the normalized probability of the phase qubit are shown below.
States | Probabilities on the Chosen States | Normalized Probabilities of the phase qubit |
---|---|---|
0 | 0.0002 | 0.0058 |
1 | 0.001 | 0.9942 |
2 | 0.0393 | |
3 | 0.1765 |
The bit values of the phase is found as \((11100110100)\) which corresponds to 0.9004. The absolute value of the eigenvalue is found from the ratio of the probability values of the phase qubit:
From the above, we find \(|\lambda |=0.8560\times ||U||_1=1.5084\). The eigenvalue of \(U\) is found as \(e^{i2\pi 0.9004}\times 1.5084=1.22255 + 0.88355i\). The eigenvalue of the Hamitlonian matrix is found from \((log(1.22255 + 0.88355i)/i)\) which is \(0.62581 - 0.41105i\).
Rights and permissions
About this article
Cite this article
Daskin, A., Grama, A. & Kais, S. A universal quantum circuit scheme for finding complex eigenvalues. Quantum Inf Process 13, 333–353 (2014). https://doi.org/10.1007/s11128-013-0654-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11128-013-0654-1