Research Article  Open Access
Shijie Wei, Hang Li, GuiLu Long, "A Full Quantum Eigensolver for Quantum Chemistry Simulations", Research, vol. 2020, Article ID 1486935, 11 pages, 2020. https://doi.org/10.34133/2020/1486935
A Full Quantum Eigensolver for Quantum Chemistry Simulations
Abstract
Quantum simulation of quantum chemistry is one of the most compelling applications of quantum computing. It is of particular importance in areas ranging from materials science, biochemistry, and condensed matter physics. Here, we propose a full quantum eigensolver (FQE) algorithm to calculate the molecular ground energies and electronic structures using quantum gradient descent. Compared to existing classicalquantum hybrid methods such as variational quantum eigensolver (VQE), our method removes the classical optimizer and performs all the calculations on a quantum computer with faster convergence. The gradient descent iteration depth has a favorable complexity that is logarithmically dependent on the system size and inverse of the precision. Moreover, the FQE can be further simplified by exploiting a perturbation theory for the calculations of intermediate matrix elements and obtaining results with a precision that satisfies the requirement of chemistry application. The full quantum eigensolver can be implemented on a nearterm quantum computer. With the rapid development of quantum computing hardware, the FQE provides an efficient and powerful tool to solve quantum chemistry problems.
1. Introduction
Quantum chemistry studies chemical systems using quantum mechanics. One primary focus of quantum chemistry is the calculation of molecular energies and electronic structures of a chemical system which determine its chemical properties. Molecular energies and electronic structures are calculated by solving the Schrodinger equation within chemical precision. However, the computational resources needed scale exponentially with the system size on a classical computer, making the calculations in quantum chemistry intractable in high dimension.
Quantum computers, originally envisioned by Benioff, Manin, and Feynman [1–3], have emerged as promising tools for tackling this challenge with polynomial overhead of computational resources. Efficient quantum simulations of chemistry systems promise breakthroughs in our knowledge for basic chemistry and revolutionize research in new materials, pharmaceuticals, and industrial catalysts.
The universal quantum simulation method [4] and the first quantum algorithm for simulating fermions [5] have laid down the fundamental block of quantum chemistry simulation. Based on these techniques and quantum phase estimation algorithm [6], AspuruGuzik et al. presented a quantum algorithm for preparing ground states undergoing an adiabatic evolution [7], and many theoretical and experimental works [8–24] have been developed since then. In 2002, Somma et al. proposed a scalable quantum algorithm for the simulation of molecular electron dynamics via JordanWigner transformation [25]. The JordanWigner transformation directly maps the fermionic occupation state of a particular atomic orbital to a state of qubits, which enables the quantum simulation of chemical systems on a quantum computer. Then, the BravyiKitaev transformation [26–30] encodes both locality of occupation and parity information onto the qubits, which is more efficient in operation complexity. In 2014, Peruzzo et al. developed the variational quantum eigensolver (VQE) [18, 31], which finds a good variational approximation to the ground state of a given Hamiltonian for a particular choice of ansatz. Compared to quantum phase estimation and Trotterization of the molecular Hamiltonian, the VQE requires a lower number of controlled operations and shorter coherence time. However, the VQE is a classical and quantum hybrid algorithm; the optimizer is performed on a classical machine.
Meanwhile, implementations of quantum chemistry simulation have been developing steadily. Studies in presentday quantum computing hardware have been carried out, such as nuclear magnetic resonance system [32, 33], photonic system [34–36], nitrogenvacancy center system [37], trapped ion [38, 39], and superconducting system [40–42]. Rapid development in quantum computer hardware with even the claims of quantum supremacy greatly stimulates the expectation of its real applications. Quantum chemistry simulation is considered a real application in Noisy IntermediateScale Quantum (NISQ) computers [22, 43, 44]. The FQE is an effort on this background. In the FQE, not only calculation of Hamiltonian matrix part is done on quantum computer, but also the optimization by gradient descent is performed on quantum computer. FQE can be used in nearterm NISQ computers and in future faulttolerant large quantum computers.
2. Method
2.1. Preparing the Hamiltonian for Quantum Chemistry Simulation
A molecular system contains a collection of nuclear charges and electrons. The fundamental task of quantum chemistry is to solve the eigenvalue problem of the molecular Hamiltonian. The eigenstates of the manybody Hamiltonian determine the dynamics of the electrons as well as the properties of the molecule. The corresponding Hamiltonian of the system includes kinetic energies of nuclei and electrons and the Coulomb potentials of nucleielectron, nucleinuclei, and electronelectron, and it can be expressed in first quantization as
in atomic units , where and are the positions, charges, and masses of the nuclei and the positions of the electrons, respectively. Under the BornOppenheimer approximation which assumes the nuclei as a fixed classical point, this Hamiltonian is usually rewritten in the particle number representation in a chosen basis: where ⋯ denotes higherorder interactions and and are the creation and annihilation operators of particle in orbital and , respectively. The parameters and are the onebody and twobody integrations in the chosen basis functions . In Galerkin formulation, the scalar coefficients in Equation (2) can be calculated by
In order to perform calculations on a quantum computer, we need to map fermionic operators to qubit operators. We choose the JordanWigner transformation to achieve this task due to its straightforward expression.
The JordanWigner transformation maps Equation (2) into a qubit Hamiltonian form: where Roman indices and denote the qubit on which the operator acts and Greek indices and refer to the type of Pauli operators; i.e., means Pauli matrix acting on a qubit at site . Apparently, in Equation (2) is a linear combination of unitary Pauli matrices. The methods used in this paper finding the molecular ground state and its energy are all based on it.
In this work, we present the FQE to find the molecular groundstate energy by gradient descent iterations. Gradient descent is one of the most fundamental ways for optimization that looks for the target energy value along the direction of the steepest descent. Here, it is performed in a quantum computer with the help of linear combination of unitary operators. We analyze the relationships between the gradient descent iteration depth and the precision of the groundstate energy. The explicit quantum circuit to implement the algorithm is constructed. As illustrative examples, the groundstate energies and electronic structures of four molecules, H_{2}, LiH, H_{2}O, and NH_{3} are presented. Taking and as examples, a comparison between the FQE and VQE, a representative hybrid method, is given. FQE can be accelerated further by harnessing a perturbation theory in chemical precision. Finally, we analyze the computation complexity of FQE and summarize the results.
2.2. Quantum Gradient Descent Iteration
The classical gradient descent algorithm is usually employed to obtain the minimum of an target function . One starts from an initial point , then moves to the next point along the direction of the gradient of the target function, namely, where is a positive learning rate that determines the step size of the iteration. In searching the minimum energy of a Hamiltonian, the target function can be expressed as a quadratic optimization problem in the form . At point , the gradient operator of the objective function can be expressed as
Then, the gradient descent iteration can be regarded as an evolution of under operator , where is redefined as . In the quantum gradient descent, vector is replaced by quantum state , where is the th elements of the vector, is the dimensional computational basis, and is the modulus of vector . Denote and it can be expressed as where is the number of Pauli product terms in . Then, the gradient descent process can be rewritten as where is a linear combination of unitary operators (LCU) which was proposed in [45] in designing quantum algorithms and studied extensively [46–52]. This nonunitary evolution can be implemented in a unitary quantum circuit by adding ancillary qubits that transform it into unitary evolution in a larger space [53]. The realization of LCU can be viewed as a quantum computer wave function passing through slits and operated by a unitary operation in each slit, and then, the wave functions are combined and the result of the calculation is read out by a measurement [49]. As shown in Figure 1, we perform the evolution described by Equation (9) with the following four steps.
Wave division: the register is a composite system which contains a work system and an ancillary register. Firstly, the initial point is efficiently mapped as an initial state of the work system. In quantum chemistry, HartreeFock (HF) product state is usually used as an initial state. And the ancillary register is initialized from , where , to a specific superposition state , where is a normalization constant and is the computational basis. This is equivalent to let the state pass through slits. is a factor describing the properties of the slit, which is determined by the forms of the Hamiltonian in Equation (8). This can be done by the initialization algorithm in [54]. Moreover, the quantum random access memory (qRAM) approach can be used to prepare and , which consume and basic steps or gates, respectively, after qRAM cell is established. We denote the whole state of the composite system as .
Entanglement: then, a series of ancillary systemcontrolled operations are implemented on the work qubits. The work qubits and the ancilla register are now entangled, and the state is transformed into
The corresponding physical picture is that different unitary operations are implemented simultaneously in different subspaces, corresponding to different slits.
Wave combination: we perform Hadamad gates on ancillary register to combine all the wave functions from the different subspaces. We merely focus on the component in a subspace where the ancillary system is in state . The state of the whole system in this subspace is
Measurement: then, we measure the ancillary register. If we obtain , our algorithm succeeds and we obtain the state , where the work system is in . And then, this will be used as input for the next iteration in the quantum gradient descent process. The probability of obtaining for the state is
The successful probability after measurements is , which is an exponential function of . The number of measurements is . The measurement complexity will grow exponentially with respect to the number of iteration steps [55]. Alternatively, one can use the oblivious amplitude amplification [51] to amplify the amplitude of the desired term (ancillary qubits in state ) up to a deterministic order with repetitions before the measurement. Then, the measurement complexity will be the product of iteration depth and , linearly dependent on the number of iteration steps. After obtaining , we can continue the gradient descent process by repeating the above four steps, with replaced by in a wavedivision step. We can preset a threshold defined as as criterion for stopping the iteration. Thus, we judge if the iterated state satisfies the criterion by measuring the expectation value of Hamiltonian around the expected number of iteration, which is easier than constructing the tomography. If the next iterative state does not hit our preset threshold, this output will be regarded as the new input state and run the next iteration. Otherwise, the iteration can be terminated and the state is the final result , as one good approximation of the ground state. The groundstate energy can be calculated by .
Measuring the expectation values during the iteration procedure will destroy the state of the work system, stopping the quantum gradient descent process. So, determining the iteration depth in advance is essential. After times iterations, the approximation error is limited to (ignoring constants) where and are the two largest absolute values of the eigenvalues of Hamiltonian (see Supplemental Material (available here) for proof). The iteration depth
is logarithmically dependent on the system size and the inverse of precision. The algorithm may be terminated at a point with a preset precision . It can be seen that the choise of has little impact on converge rate when is large. This makes this algorithm very robust to this parameter. The rate of convergence primarily depends upon the ratio of and . The gap between the iterative result and the ground state depends on the choice of initial point. If we choose an ansatz state with a large overlap with the exact ground state, the iterative process will converge to the ground state in fewer iterations. Usually, the meanfield state which represents a good classical approximation to the ground state of Hamiltonian H, such as a HartreeFock (HF) product state, is chosen as an initial state. Compared to the VQE, the FQE does not need to make measurements of the expectation values of Hamiltonian during each iteration procedure and this substantially reduces the computation resources.
2.3. Perturbation Theory
The FQE involves multitime iterations to obtain an accurate result, which is difficult to implement in the presentday quantum computer hardware. Here, we present an approximate method to find the ground state and its energy by using the gradient descent algorithm and perturbation theory. The perturbation theory is widely used and plays an important role in describing real quantum systems, because it is impossible to find exact solutions to the Schrodinger equation for Hamiltonians even with moderate complexity. The Hamiltonian described by Equation (4) can be divided into two classes, and . consists of a set of Pauli terms containing only and the identity matrices, and Pauli terms belong to . is a diagonal matrix with exact solutions that can be regarded as a simple system. usually is smaller compared to and is treated as a “perturbing” Hamiltonian. The energy levels and eigenstates associated with the perturbed system can be expressed as “corrections” to those of the unperturbed system. We begin with the timeindependent Schrodinger equation: where and are the th energy and eigenstate, respectively. Unperturbed Hamiltonian satisfies the timeindependent Schrodinger equation: . Our goal is to express and in terms of and . Denote the expectation value of as , and it is easy to see that is zero because only contains Pauli terms . In the firstorder approximation, the energies and eigenstates are expressed as
In the secondorder approximation, they are
The matrix elements in the first and secondorder approximations can be obtained by one iteration of the quantum circuit in Figure 1. Here, we let be equal to . Explicitly, the firstorder approximation only involves , a series of transition probabilities of the state after implemented on state , and they can be obtained by performing the quantum circuit of Figure 1 directly. For the secondorder approximation, matrix elements such as value and can be calculated by . Then, the approximate ground energy and ground state up to the second order are obtained. We will show the performance of the FQE and perturbation theory in the next section.
3. Results
3.1. Calculations of Four Molecules
To demonstrate the feasibility of this FQE with gradient descent iteration, we carried out calculations on the groundstate energy of H_{2}, LiH diatomic molecules, and two relatively complex molecules H_{2}O and NH_{3}. We used a common molecular basis set, the minimal STO3G basis. Via the JordanWigner transformation, the qubit Hamiltonians of these molecules are obtained. The Hamiltonians of H_{2}, LiH, H_{2}O, and NH_{3} contain 15, 118, 252, and 3382 Pauli matrix product terms, respectively. The dimensions of the Hamiltonians of H_{2}, LiH, H_{2}O, and NH_{3} are 16, 64, 4096, and 16384, respectively, which correspond to 4, 6, 12, and 14 numbers of qubits, respectively. In all four simulations, the work system was initialized to the HF state and the learning rate is chosen as . As shown in Figure 2, after about 120 iterations, the molecular energy of H_{2}O converges to 74.94 a.u., only 0.0013346% discrepancy with respect to the exact value of 74.93 a.u. obtained via Hamiltonian diagonalization. The NH_{3} calculation yields (55.525 a.u.) after 80 iterations matched very well with the diagonalization (55.526 a.u.). For the study of atomic molecular structures and chemical reactions, these results are sufficiently accurate. For more complex basis set STO6G, the results are about the same (see Figure 3), and the details are given in Supplemental Material. The converge rates of the four molecules depend on the system size and the ratio of the two largest absolute eigenvalues of the Hamiltonian , which are consistent with the theoretical analysis above.
We also studied the influence of noises which is also shown in Figure 2. The noise term is chosen in the form of added to the Hamiltonian to simulate decoherence. Then, we add a term on the iterative state to simulate measurement error and renormalize the iterative state as . We set a random noise (amplitude 0.01) and a Gaussian noise () for H_{2} and LiH. For H_{2}O and NH_{3}, we choose a random noise (amplitude 0.02) and a Gaussian noise (). The results still converge to the exact values in chemical precision (). This indicates that our method is robust to a certain type of noise, which is important in the implementation of quantum simulation on nearterm quantum devices. For more noisy situations, see Supplemental Material for details, where the parameters of noise are 10 times the above values (see Figure 4). The convergence deteriorates and some oscillations occur as the number of iterations increases.
In Figure 5, a comparison with the VQE is shown for H_{2}O and NH_{3}. In VQE calculation, the initial state is mapped to an ansatz state by a parameterized unitary operation . The VQE solves for the parameter vector with a classical optimization routine. Here, we adopt the standard gradient descent method as the classical optimizer in the VQE. The parameter is updated by . We performed numerical simulations of the VQE for the two molecules. When the learning rate , the VQE does not converge to the ground state. So, in order to compare with each other, we choose the proper learning rates in two methods separately. In both cases, the initial ansatz state is prepared as the HF product state. In H_{2}O and NH_{3}, the VQE converges the fastest with the learning rate . The FQE converges faster with larger and larger learning rate until a fixed speed is reached. As shown in Figure 5, the FQE generally converges faster than the VQE and the advantage will be more obvious in complex molecules.
The above examples are calculated in fixed interatomic distance of the molecules. If we want to calculate the interatomic distance corresponding to the most stable structure, the variation of interatomic distances is necessary. In Figure 6, four examples are given to illustrate the performance of the perturbation theory. To obtain the potentialenergy surfaces for H_{2}, LiH, H_{2}O, and NH_{3} molecules, we studied the dependence of groundstate energy of their molecules on the variating interatomic distances, between the two atoms in H_{2} and LiH, the distance between the oxygen atom and one hydrogen atom (the two hydrogen atoms are symmetric with respect to the oxygen atom) in H_{2}O, and the distance between the nitrogen atom and the plane formed by the three hydrogen atoms in NH_{3}. The lowest energy in potentialenergy surfaces corresponds to the most stable structure of the molecules. As shown in the picture, the groundstate energy of each molecule calculated under the secondorder approximation is already quite close to its exact value, which is obtained from Hamiltonian diagonalizations. The energy values up to the secondorder correction are compared with their exact values at the most stable interatomic distance corresponding to the lowest energy in Table 1. It can be seen that the secondorder approximation has already given results in chemical precision.

3.2. Analysis of Computational Complexity
Here, we analyze the complexity of our algorithm. Usually, a quantum algorithm complexity involves two aspects: qubit resources and gate complexity. For qubit resources, the number of ancilla qubits is , where is the number of Pauli terms in qubit from the Hamiltonian. For gate complexity, the “Wave division” part needs basic steps for state preparation. The dominant factor is the number of controlled operations in the “Entanglement” part in Figure 1. Controlled can be decomposed into basic gates [56, 57]. The “Wave combination” part just comprises Hadamard gates. Totally, the FQE requires in each iteration about basic gates for implementation. If the wave function is expressed by Gaussian orbitals, fermion Hamiltonians contain secondquantized terms; consequently, the qubit Hamiltonians have Pauli terms. The qubit resource and gate complexity can be reduced to and , respectively. In some applications, the perturbation theory only requires one iteration, and an approximate result in chemical precision can be obtained.
4. Summary
An efficient quantum algorithm, full quantum eigensolver (FQE), for calculating the groundstate wave function and the ground energy using gradient descent (FQE) was proposed, and numerical simulations are performed for four molecules. In the FQE, the complexity of basic gate operations is polylogarithmical to the number of singleelectron atomic orbitals. It achieves an exponential speedup compared with its classical counterparts. It has been shown that the FQE is robust against noises of reasonable strengths. For very noisy situations that do not allow many iterations, the FQE can be combined with the perturbation theory that gives the ground state and its energy in chemical precision with onetime iteration. The FQE is exceptionally useful in quantum chemistry simulation, especially for the nearterm NISQ applications. The FQE is a full quantum algorithm, not only applicable for NISQ computers but also directly applicable for future largescale faulttolerant quantum computers.
Data Availability
The data that support the findings of this study are available from the corresponding authors on reasonable request.
Conflicts of Interest
The authors declare no competing interests.
Authors’ Contributions
S.J.W. conceived the algorithm. H.L. performed classical simulations. G.L.L. initialized LCU scheme. All authors contributed to the discussion of results and writing of the manuscript.
Acknowledgments
This research was supported by the National Basic Research Program of China. We gratefully acknowledges support from the National Natural Science Foundation of China under Grant No. 11974205 and No. 11774197, the National Key Research and Development Program of China (2017YFA0303700), the Key Research and Development Program of Guangdong Province (2018B030325002); and Beijing Advanced Innovation Center for Future Chip (ICFC).
Supplementary Materials
Section A: error estimation and iteration complexity of FQE. Section B: FQE with STO6G basis as input. Section C: performance of FQE with large noise. Figure S1: the gradient descent iteration process for convergence of groundstate energy of H_{2}, LiH, H_{2}O, and NH_{3}, respectively, with STO6G basis. Figure S2: influence of large noise on FQE in H_{2}, LiH, H_{2}O, and NH_{3} molecules, respectively [58]. (Supplementary Materials)
References
 P. Benioff, “The computer as a physical system: a microscopic quantum mechanical Hamiltonian model of computers as represented by Turing machines,” Journal of Statistical Physics, vol. 22, no. 5, pp. 563–591, 1980. View at: Publisher Site  Google Scholar
 I. Manin, Vychislimoe i nevychislimoe, Sov. Radio, 1980.
 R. P. Feynman, “Simulating physics with computers,” International Journal of Theoretical Physics, vol. 21, no. 67, pp. 467–488, 1982. View at: Publisher Site  Google Scholar
 S. Lloyd, “Universal quantum simulators,” Science, vol. 273, no. 5278, pp. 1073–1078, 1996. View at: Publisher Site  Google Scholar
 D. S. Abrams and S. Lloyd, “Simulation of manybody Fermi systems on a universal quantum computer,” Physical Review Letters, vol. 79, no. 13, pp. 2586–2589, 1997. View at: Publisher Site  Google Scholar
 A. Y. Kitaev, “Quantum measurements and the abelian stabilizer problem,” 1995, http://arxiv.org/abs/9511026. View at: Google Scholar
 A. AspuruGuzik, A. D. Dutoi, P. J. Love, and M. HeadGordon, “Simulated quantum computation of molecular energies,” Science, vol. 309, no. 5741, pp. 1704–1707, 2005. View at: Publisher Site  Google Scholar
 R. Babbush, P. J. Love, and A. AspuruGuzik, “Adiabatic quantum simulation of quantum chemistry,” Scientific Reports, vol. 4, no. 1, article 6603, 2015. View at: Publisher Site  Google Scholar
 G.R. Feng, Y. Lu, L. Hao, F.H. Zhang, and G. L. Long, “Experimental simulation of quantum tunneling in small systems,” Scientific Reports, vol. 3, no. 1, article 2232, 2013. View at: Publisher Site  Google Scholar
 Y. Lu, G.R. Feng, Y.S. Li, and G.L. Long, “Experimental digital quantum simulation of temporalspatial dynamics of interacting fermion system,” Science Bulletin, vol. 60, no. 2, pp. 241–248, 2015. View at: Publisher Site  Google Scholar
 R. Babbush, J. McClean, D. Wecker, A. AspuruGuzik, and N. Wiebe, “Chemical basis of TrotterSuzuki errors in quantum chemistry simulation,” Physical Review A, vol. 91, no. 2, article 022311, 2015. View at: Publisher Site  Google Scholar
 S.J. Wei, D. Ruan, and G.L. Long, “Duality quantum algorithm efficiently simulates open quantum systems,” Scientific Reports, vol. 6, no. 1, article 30727, 2016. View at: Publisher Site  Google Scholar
 R. Babbush, D. W. Berry, I. D. Kivlichan, A. Y. Wei, P. J. Love, and A. AspuruGuzik, “Exponentially more precise quantum simulation of fermions in second quantization,” New Journal of Physics, vol. 18, no. 3, article 033032, 2016. View at: Publisher Site  Google Scholar
 R. Babbush, D. W. Berry, Y. R. Sanders et al., “Exponentially more precise quantum simulation of fermions in the configuration interaction representation,” Quantum Science and Technology, vol. 3, no. 1, article 015006, 2018. View at: Publisher Site  Google Scholar
 I. Kassal, S. P. Jordan, P. J. Love, M. Mohseni, and A. AspuruGuzik, “Polynomialtime quantum algorithm for the simulation of chemical dynamics,” Proceedings of the National Academy of Sciences of the United States of America, vol. 105, no. 48, pp. 18681–18686, 2008. View at: Publisher Site  Google Scholar
 I. D. Kivlichan, N. Wiebe, R. Babbush, and A. AspuruGuzik, “Bounding the costs of quantum simulation of manybody physics in real space,” Journal of Physics A: Mathematical and Theoretical, vol. 50, no. 30, p. 305301, 2017. View at: Publisher Site  Google Scholar
 B. Toloui and P. J. Love, “Quantum algorithms for quantum chemistry based on the sparsity of the CImatrix,” 2013, http://arxiv.org/abs/1312.2579. View at: Google Scholar
 A. Peruzzo, J. McClean, P. Shadbolt et al., “A variational eigenvalue solver on a photonic quantum processor,” Nature Communications, vol. 5, no. 1, article 4213, 2014. View at: Publisher Site  Google Scholar
 J. R. McClean, J. Romero, R. Babbush, and A. AspuruGuzik, “The theory of variational hybrid quantumclassical algorithms,” New Journal of Physics, vol. 18, no. 2, article 023023, 2016. View at: Publisher Site  Google Scholar
 J. R. McClean, R. Babbush, P. J. Love, and A. AspuruGuzik, “Exploiting locality in quantum computation for quantum chemistry,” The Journal of Physical Chemistry Letters, vol. 5, no. 24, pp. 4368–4380, 2014. View at: Publisher Site  Google Scholar
 J. D. Whitfield, J. Biamonte, and A. AspuruGuzik, “Simulation of electronic structure hamiltonians using quantum computers,” Molecular Physics, vol. 109, no. 5, pp. 735–750, 2011. View at: Publisher Site  Google Scholar
 D. Wecker, M. B. Hastings, and M. Troyer, “Progress towards practical quantum variational algorithms,” Physical Review A, vol. 92, no. 4, article 042303, 2015. View at: Publisher Site  Google Scholar
 M. B. Hastings, D. Wecker, B. Bauer, and M. Troyer, “Improving quantum algorithms for quantum chemistry,” Quantum Information & Computation, vol. 15, no. 12, pp. 1–21, 2015. View at: Google Scholar
 O. Kyriienko, “Quantum inverse iteration algorithm for nearterm quantum devices,” 2019, http://arxiv.org/abs/1901.09988. View at: Google Scholar
 P. Jordan and E. Wigner, “Über das Paulische Äquivalenzverbot,” Zeitschrift für Physik, vol. 47, no. 910, pp. 631–651, 1928. View at: Publisher Site  Google Scholar
 S. B. Bravyi and A. Y. Kitaev, “Fermionic quantum computation,” Annals of Physics, vol. 298, no. 1, pp. 210–226, 2002. View at: Publisher Site  Google Scholar
 J. T. Seeley, M. J. Richard, and P. J. Love, “The BravyiKitaev transformation for quantum computation of electronic structure,” The Journal of Chemical Physics, vol. 137, no. 22, p. 224109, 2012. View at: Publisher Site  Google Scholar
 A. Tranter, S. Sofia, J. Seeley et al., “The BravyiKitaev transformation: properties and applications,” International Journal of Quantum Chemistry, vol. 115, no. 19, pp. 1431–1441, 2015. View at: Publisher Site  Google Scholar
 S. Bravyi, J. M. Gambetta, A. Mezzacapo, and K. Temme, “Tapering off qubits to simulate fermionic Hamiltonians,” 2017, http://arxiv.org/abs/1701.08213. View at: Google Scholar
 R. Babbush, N. Wiebe, J. McClean, J. McClain, H. Neven, and G. K.L. Chan, “Lowdepth quantum simulation of materials,” Physical Review X, vol. 8, no. 1, article 011044, 2018. View at: Publisher Site  Google Scholar
 M.H. Yung, J. Casanova, A. Mezzacapo et al., “From transistor to trappedion computers for quantum chemistry,” Scientific Reports, vol. 4, no. 1, article 3589, 2015. View at: Publisher Site  Google Scholar
 J. Du, N. Xu, X. Peng, P. Wang, S. Wu, and D. Lu, “Nmr implementation of a molecular hydrogen quantum simulation with adiabatic state preparation,” Physical Review Letters, vol. 104, no. 3, article 030502, 2010. View at: Publisher Site  Google Scholar
 Z. Li, X. Liu, H. Wang et al., “Quantum simulation of resonant transitions for solving the eigenproblem of an effective water Hamiltonian,” Physical Review Letters, vol. 122, no. 9, article 090504, 2019. View at: Publisher Site  Google Scholar
 P. Roushan, C. Neill, A. Megrant et al., “Chiral groundstate currents of interacting photons in a synthetic magnetic field,” Nature Physics, vol. 13, no. 2, pp. 146–151, 2017. View at: Publisher Site  Google Scholar
 B. P. Lanyon, J. D. Whitfield, G. G. Gillett et al., “Towards quantum chemistry on a quantum computer,” Nature Chemistry, vol. 2, no. 2, pp. 106–111, 2010. View at: Publisher Site  Google Scholar
 S. Paesani, A. A. Gentile, R. Santagati et al., “Experimental Bayesian quantum phase estimation on a silicon photonic chip,” Physical Review Letters, vol. 118, no. 10, article 100503, 2017. View at: Publisher Site  Google Scholar
 Y. Wang, F. Dolde, J. Biamonte et al., “Quantum simulation of helium hydride cation in a solidstate spin register,” ACS Nano, vol. 9, no. 8, pp. 7769–7774, 2015. View at: Publisher Site  Google Scholar
 Y. Shen, X. Zhang, S. Zhang, J.N. Zhang, M.H. Yung, and K. Kim, “Quantum implementation of the unitary coupled cluster for simulating molecular electronic structure,” Physical Review A, vol. 95, no. 2, article 020501, 2017. View at: Publisher Site  Google Scholar
 C. Hempel, C. Maier, J. Romero et al., “Quantum chemistry calculations on a trappedion quantum simulator,” Physical Review X, vol. 8, no. 3, article 031022, 2018. View at: Publisher Site  Google Scholar
 P. J. J. O’Malley, R. Babbush, I. D. Kivlichan et al., “Scalable quantum simulation of molecular energies,” Physical Review X, vol. 6, no. 3, article 031007, 2016. View at: Publisher Site  Google Scholar
 A. Kandala, A. Mezzacapo, K. Temme et al., “Hardwareefficient variational quantum eigensolver for small molecules and quantum magnets,” Nature, vol. 549, no. 7671, pp. 242–246, 2017. View at: Publisher Site  Google Scholar
 M. Ganzhorn, D. J. Egger, P. Barkoutsos et al., “Gateefficient simulation of molecular eigenstates on a quantum computer,” Physical Review Applied, vol. 11, no. 4, article 044092, 2019. View at: Publisher Site  Google Scholar
 M. Mohseni, P. Read, H. Neven et al., “Commercialize quantum technologies in five years,” Nature, vol. 543, no. 7644, pp. 171–174, 2017. View at: Publisher Site  Google Scholar
 L. Mueck, “Quantum reform,” Nature Chemistry, vol. 7, no. 5, pp. 361–363, 2015. View at: Publisher Site  Google Scholar
 L. GuiLu, “General quantum interference principle and duality computer,” Communications in Theoretical Physics, vol. 45, no. 5, pp. 825–844, 2006. View at: Publisher Site  Google Scholar
 S. Gudder, “Mathematical theory of duality quantum computers,” Quantum Information Processing, vol. 6, no. 1, pp. 37–48, 2007. View at: Publisher Site  Google Scholar
 L. GuiLu and L. Yang, “Duality computing in quantum computers,” Communications in Theoretical Physics, vol. 50, no. 6, pp. 1303–1306, 2008. View at: Publisher Site  Google Scholar
 L. GuiLu, L. Yang, and W. Chuan, “Allowable generalized quantum gates,” Communications in Theoretical Physics, vol. 51, no. 1, pp. 65–67, 2009. View at: Publisher Site  Google Scholar
 G. L. Long, “Duality quantum computing and duality quantum information processing,” International Journal of Theoretical Physics, vol. 50, no. 4, pp. 1305–1318, 2011. View at: Publisher Site  Google Scholar
 A. M. Childs and N. Wiebe, “Hamiltonian simulation using linear combinations of unitary operations,” 2012, http://arxiv.org/abs/1202.5822. View at: Google Scholar
 D. W. Berry, A. M. Childs, R. Cleve, R. Kothari, and R. D. Somma, “Simulating Hamiltonian dynamics with a truncated Taylor series,” Physical Review Letters, vol. 114, no. 9, article 090502, 2015. View at: Publisher Site  Google Scholar
 S.J. Wei and G.L. Long, “Duality quantum computer and the efficient quantum simulations,” Quantum Information Processing, vol. 15, no. 3, pp. 1189–1212, 2016. View at: Publisher Site  Google Scholar
 H. X. Cao, L. Li, Z. L. Chen, Y. Zhang, and Z. H. Guo, “Restricted allowable generalized quantum gates,” Chinese Science Bulletin, vol. 55, no. 20, pp. 2122–2125, 2010. View at: Publisher Site  Google Scholar
 G.L. Long and Y. Sun, “Efficient scheme for initializing a quantum register with an arbitrary superposed state,” Physical Review A, vol. 64, no. 1, article 014303, 2001. View at: Publisher Site  Google Scholar
 P. Rebentrost, M. Schuld, L. Wossnig, F. Petruccione, and S. Lloyd, “Quantum gradient descent and Newton’s method for constrained polynomial optimization,” New Journal of Physics, vol. 21, no. 7, article 073023, 2019. View at: Publisher Site  Google Scholar
 T. Xin, S.J. Wei, J. S. Pedernales, E. Solano, and G.L. Long, “Quantum simulation of quantum channels in nuclear magnetic resonance,” Physical Review A, vol. 96, no. 6, article 062303, 2017. View at: Publisher Site  Google Scholar
 S.J. Wei, T. Xin, and G.L. Long, “Efficient universal quantum channel simulation in ibm’s cloud quantum computer,” Science China Physics, Mechanics & Astronomy, vol. 61, no. 7, article 70311, 2018. View at: Publisher Site  Google Scholar
 M. Panju, “Iterative methods for computing eigenvalues and eigenvectors,” 2011, http://arxiv.org/abs/1105.1185. View at: Google Scholar
Copyright
Copyright © 2020 Shijie Wei et al. Exclusive Licensee Science and Technology Review Publishing House. Distributed under a Creative Commons Attribution License (CC BY 4.0).