Surface code quantum error correction incorporating accurate error propagation
Abstract
The surface code is a powerful quantum error correcting code that can be defined on a 2-D square lattice of qubits with only nearest neighbor interactions. Syndrome and data qubits form a checkerboard pattern. Information about errors is obtained by repeatedly measuring each syndrome qubit after appropriate interaction with its four nearest neighbor data qubits. Changes in the measurement value indicate the presence of chains of errors in space and time. The standard method of determining operations likely to return the code to its error-free state is to use the minimum weight matching algorithm to connect pairs of measurement changes with chains of corrections such that the minimum total number of corrections is used. Prior work has not taken into account the propagation of errors in space and time by the two-qubit interactions. We show that taking this into account leads to a quadratic improvement of the logical error rate.
I Introduction
The idea of manipulating quantum systems to perform computation was first proposed by Feynman in 1982 Feynman (1982). Serious interest in making this vision a reality followed the invention of Shor’s factoring algorithm in 1994 Shor (1994). Initial concerns that unfeasible physical control was required were tempered by the invention of quantum error correction (QEC) in 1995 Shor (1995) and proof of the threshold theorem in 1996 Knill et al. (1996); Aharonov and Ben-Or (1997). The threshold theorem showed that arbitrarily large quantum computations can be performed arbitrarily reliably provided the error rates of the various components in the quantum computer are all below some fixed threshold error rate .
Early work Aharonov and Ben-Or (1999) put the value of at approximately . The ability to interact pairs of qubits separated by arbitrarily large distances was also demanded. The current record highest of approximately 3% is shared by two quantum computing schemes Knill (2005); Fujii and Yamamoto (2009). Both of these schemes, however, still require the ability to interact pairs of qubits separated by arbitrarily large distances. This is a major barrier to implementation. Studies of the performance of such error correction techniques when they are restricted to 1-D, quasi-1-D and 2-D lattices of qubits with only nearest neighbor interactions Stephens and Evans (2009); Stephens et al. (2008a); Fowler et al. (2007); Helmer et al. (2009); Szkopek et al. (2006); Svore et al. (2004) have found that the threshold error rate drops to the level or below.
Topological approaches to quantum error correction Bravyi and Kitaev (1998); Dennis et al. (2002); Raussendorf et al. (2006); Raussendorf and Harrington (2007); Raussendorf et al. (2007); Bombin and Martin-Delgado (2006a, b, 2007a, 2007b); Bombin and Martin-Delgado (2009, 2008); Katzgraber et al. (2010); Stace et al. (2009); Stace and Barrett (2009); Duclos-Cianci and Poulin (2010); Fowler et al. (2009a); Fowler and Goyal (2009); Fowler et al. (2009b); Wang et al. (2010, 2009); Devitt et al. (2010) can be implemented optimally using only nearest neighbor interactions. Kitaev’s surface code Bravyi and Kitaev (1998), which makes use of a 2-D square lattice of qubits with nearest neighbor interactions, possesses the highest threshold error rate of the known topological schemes at approximately 0.75% Raussendorf and Harrington (2007); Fowler et al. (2009a); Wang et al. (2010). This is not far below the record threshold error rate and above some existing experimental error rates DiCarlo et al. (2009); Passante et al. (2009). In recent years, a number of quantum computer designs based on this form of error correction have been devised Stock and James (2009); Devitt et al. (2007); Stephens et al. (2008b); Devitt et al. (2009); Van Meter et al. (2010); DiVincenzo (2009) and an experimental demonstration using linear optics has been performed Gao et al. (2009). Many technologies are being developed that are in principle capable of implementing a 2-D lattice of qubits with nearest neighbor interactions, including ion traps Blakestad et al. (2009); Leibrandt et al. (2009); Home et al. (2009); Hanneke et al. (2009); Amini et al. (2009), neutral atom chips Fortágh and Zimmermann (2007); Riedel et al. (2010), optical lattices Bocko et al. (2008); Becker et al. (2009), superconductors DiCarlo et al. (2009); Bialczak et al. (2009) quantum dots Taylor et al. (2005), donors in silicon Hollenberg et al. (2006) and electrons on liquid helium Lyon (2006).
A number of open problems remain concerning the best way to perform the classical processing associated with surface code QEC. In this paper, we address a major failing of prior approaches Fowler et al. (2009a); Wang et al. (2010); Duclos-Cianci and Poulin (2010), which did not take into account the propagation of errors as a result of the two-qubit interactions required to detect errors. We show that, for a given lattice size, carefully taking into account error propagation results in a quadratic improvement of the logical error rate.
The discussion is organized as follows. In Section II, the surface code is briefly reviewed. Section III discusses the standard surface code QEC procedure, its limitations, and numerical results explicitly demonstrating these limitations. Section IV describes our modification to the standard procedure, its advantages, and numerical results explicitly demonstrating these advantages. Section V concludes with a summary of our results.
Ii The surface code
In this Section, we briefly describe the surface code, focusing primarily on its stabilizers and the quantum circuits required to measure them. A full description of surface code quantum computing can be found in Fowler et al. (2009a).
A stabilizer of a state is an operator such that . Any error that anticommutes with can be readily detected since . A generic circuit capable of determining the sign of a stabilizer is shown in Fig. 1a. Surface code stabilizers have the form , , or , as shown for the specific case in Fig. 2a. Circuits capable of measuring such stabilizers are shown in Figs. 1a-b. An appropriate sequence of two-qubit gates to use when measuring all stabilizers across the lattice simultaneously is shown in Fig. 2b.
The lattice of Fig. 2a is capable of protecting a single qubit of information from errors. This protected qubit is called a logical qubit and can be read out in the logical basis by measuring all data qubits in the physical basis. After error correction, which we shall describe in the next Section, the logical measurement result is equal to the product of measurement results along paths connecting boundaries as shown in Fig. 2c. The distance of the code is equal to the shortest of these paths. Fig. 2a is an example of a distance code.
Iii Standard surface code QEC
Repeatedly executing the gate sequence shown in Fig. 2b, along with syndrome qubit initialization and measurement, generates data points in space and time where the measurement values change. These are called syndrome changes. Fig. 3a shows an example. The standard method of surface code QEC effectively only considers errors occurring on data qubits at the same time as the syndrome qubits are being measured and errors on the syndrome qubits just before they measured. Such errors are not propagated by the two-qubit gates. A single data qubit error results in two syndrome changes adjacent in space. A single syndrome qubit error results in two syndrome changes adjacent in time. The separation of two syndrome changes occurring at space-time locations , is defined to be . In other words, the standard approach assumes that an error chain containing at least errors must occur to produce the observed syndrome changes. The minimum weight matching algorithm Edmonds (1965a, b) is used to process the list of space-time locations, matching pairs of locations such that the total of all separations is a minimum. An example is shown in Fig. 3b.
The performance of this approach is shown in Fig. 4. The error rates of initialization, measurement, memory and two-qubit gates are all set to equal . The number of rounds of syndrome measurement that can be performed, on average, before a logical error occurs is plotted for a range of lattice sizes and gate error rates . It can be seen that something is seriously wrong. At low physical error rates, a distance lattice has a logical error rate proportional to , meaning it offers no genuine error correction ability. Both and are only capable of reliably correcting a single error.
The reason the error correction performs suboptimally at low error rates is illustrated in Fig. 5. A single error occurring halfway through syndrome measurement can cause a pair of syndrome changes separated by two units of space and one of time. A chain of such errors as shown in Fig. 6 will be matched as shown in dashed lines, resulting in a logical error. A lattice can thus only cope with errors in the worst case using this error correction method.
Iv Improved surface code QEC
A lattice should be able to cope with errors in the worst case. We achieve this by modifying the way in which the separation is calculated between a pair of syndrome changes.
Let us consider a slightly more general syndrome measurement procedure than that shown in Fig. 2b. Let us assume that some of the two-qubit interactions are significantly slower than others. Some syndrome measurements may be performed more frequently than others. We permit the order of interaction to vary dynamically to enable syndrome measurements to be performed as frequently as possible. We imposed the necessary condition that it must be possible to say whether the gates associated with a given syndrome measurement occurred strictly before or after those associated with any other syndrome measurement.
Consider a pair of neighboring stabilizers sharing a single data qubit. Let us imagine that one stabilizer is measured more frequently than the other. errors occurring on the data qubit at different times will be detected by the adjacent stabilizer measurements in a number of different ways as shown in Fig. 7. These are all single-error processes. As such, the pairs of changed syndromes indicated in red should all be connected. These connections, along with vertical connections between same site syndrome measurements and connections of the form shown in Fig. 5 will be used to determine the correct separation of a given pair of syndrome changes. The separation of a given pair of syndrome changes is defined to be the minimum number of connections in any path connecting them.
It may seem that there must be additional connections to cover the case of two-qubit interactions suffering a two-qubit error, however this is not the case. Two-qubit and errors are equivalent to single or errors occurring before the two-qubit gate. Two-qubit errors consisting of both and operators give rise to propagated errors that are handled independently by the two types of error correction.
The separation of any given pair of syndrome changes is now calculated by determining the number of edges in the shortest path connecting them. When all gates take the same time, the ideal execution order remains as shown in Fig. 2b. Nevertheless, as can be seen in Fig. 8, the new method of calculating the separation, a modification of the classical processing only, dramatically changes the performance of surface code QEC. The threshold error rate remains unchanged at approximately 0.75%. A lattice of distance can now reliably correct errors. For large , the new approach provides a quadratic improvement of the logical error rate. Even for modest parameters, such as and , this translates to a logical error rate improvement of over two orders of magnitude.
V Conclusion
By accurately tracking the propagation of errors due to the two-qubit gates use during surface code QEC, we have improved the number of errors a code can reliably correct from to , which is optimal. Explicitly, given a gate error rate and a surface code logical qubit with (a physical qubit lattice), our method leads to over two orders of magnitude improvement in the reliability of the logical qubit. This improvement is purely as a result of better classical processing. In the limit of large , the new approach provides a quadratic improvement of the logical error rate.
Vi Acknowledgements
We acknowledge support from the Australian Research Council, the Australian Government, and the US National Security Agency (NSA) and the Army Research Office (ARO) under contract number W911NF-08-1-0527.
References
- Feynman (1982) R. P. Feynman, Int. J. Theor. Phys. 21, 467 (1982).
- Shor (1994) P. W. Shor, in Proc. 35th Annual Symposium on Foundations of Computer Science (1994), pp. 124–134, quant-ph/9508027.
- Shor (1995) P. W. Shor, Phys. Rev. A 52, R2493 (1995).
- Knill et al. (1996) E. Knill, R. Laflamme, and W. H. Zurek, Tech. Rep. LAUR-96-2199, Los Alamos National Laboratory (1996), quant-ph/9610011.
- Aharonov and Ben-Or (1997) D. Aharonov and M. Ben-Or, Proc. ACM STOC 29, 176 (1997), quant-ph/9611025.
- Aharonov and Ben-Or (1999) D. Aharonov and M. Ben-Or, quant-ph/9906129 (1999).
- Knill (2005) E. Knill, Nature 434, 39 (2005), quant-ph/0410199.
- Fujii and Yamamoto (2009) K. Fujii and K. Yamamoto, arXiv:0912.5150 (2009).
- Stephens and Evans (2009) A. M. Stephens and Z. W. E. Evans, Phys. Rev. A 80, 022313 (2009), arXiv:0902.2658.
- Stephens et al. (2008a) A. M. Stephens, A. G. Fowler, and L. C. L. Hollenberg, Quant. Info. Comp. 8, 330 (2008a), quant-ph/0702201.
- Fowler et al. (2007) A. G. Fowler, W. F. Thompson, Z. Yan, A. M. Stephens, B. L. T. Plourde, and F. K. Wilhelm, Phys. Rev. B 76, 174507 (2007), cond-mat/0702620.
- Helmer et al. (2009) F. Helmer, M. Mariantoni, A. G. Fowler, J. von Delft, E. Solano, and F. Marquardt, Euro. Phys. Lett. 85, 50007 (2009), arXiv:0706.3625.
- Szkopek et al. (2006) T. Szkopek, H. Fan, V. Roychowdhury, E. Yablonovitch, P. O. Boykin, G. Simms, M. Gyure, and B. Fong, IEEE Trans. Nano. 5, 42 (2006), quant-ph/0411111.
- Svore et al. (2004) K. M. Svore, B. M. Terhal, and D. P. DiVincenzo, quant-ph/0410047 (2004).
- Bravyi and Kitaev (1998) S. B. Bravyi and A. Y. Kitaev, quant-ph/9811052 (1998).
- Dennis et al. (2002) E. Dennis, A. Kitaev, A. Landahl, and J. Preskill, J. Math. Phys. 43, 4452 (2002), quant-ph/0110143.
- Raussendorf et al. (2006) R. Raussendorf, J. Harrington, and K. Goyal, Ann. Phys. 321, 2242 (2006), quant-ph/0510135.
- Raussendorf and Harrington (2007) R. Raussendorf and J. Harrington, Phys. Rev. Lett. 98, 190504 (2007), quant-ph/0610082.
- Raussendorf et al. (2007) R. Raussendorf, J. Harrington, and K. Goyal, New J. Phys. 9, 199 (2007), quant-ph/0703143.
- Bombin and Martin-Delgado (2006a) H. Bombin and M. A. Martin-Delgado, Phys. Rev. A 73, 062303 (2006a), quant-ph/0602063.
- Bombin and Martin-Delgado (2006b) H. Bombin and M. A. Martin-Delgado, Phys. Rev. Lett. 97, 180501 (2006b), quant-ph/0605138.
- Bombin and Martin-Delgado (2007a) H. Bombin and M. A. Martin-Delgado, Phys. Rev. Lett. 98, 160502 (2007a), quant-ph/0610024.
- Bombin and Martin-Delgado (2007b) H. Bombin and M. A. Martin-Delgado, Phys. Rev. A 76, 012305 (2007b), quant-ph/0703272.
- Bombin and Martin-Delgado (2009) H. Bombin and M. A. Martin-Delgado, J. Phys. A: Math. Theor. 42, 095302 (2009), arXiv:0704.2540.
- Bombin and Martin-Delgado (2008) H. Bombin and M. A. Martin-Delgado, Phys. Rev. A 77, 042322 (2008), arXiv:0711.0468.
- Katzgraber et al. (2010) H. G. Katzgraber, H. Bombin, R. S. Andrist, and M. A. Martin-Delgado, Phys. Rev. A 81, 012319 (2010), arXiv:0910.0573.
- Stace et al. (2009) T. M. Stace, S. D. Barrett, and A. C. Doherty, Phys. Rev. Lett. 102, 200501 (2009), arXiv:0904.3556.
- Stace and Barrett (2009) T. M. Stace and S. D. Barrett, arXiv:0912.1159 (2009).
- Duclos-Cianci and Poulin (2010) G. Duclos-Cianci and D. Poulin, Phys. Rev. Lett. 104, 050504 (2010), arXiv:0911.0581.
- Fowler et al. (2009a) A. G. Fowler, A. M. Stephens, and P. Groszkowski, Phys. Rev. A 80, 052312 (2009a), arXiv:0803.0272.
- Fowler and Goyal (2009) A. G. Fowler and K. Goyal, Quant. Info. Comput. 9, 721 (2009), arXiv:0805.3202.
- Fowler et al. (2009b) A. G. Fowler, D. S. Wang, T. D. Ladd, R. Van Meter, and L. C. L. Hollenberg, arXiv:0910.4074 (2009b).
- Wang et al. (2010) D. S. Wang, A. G. Fowler, A. M. Stephens, and L. C. L. Hollenberg, Quant. Info. Comp. 10, 456 (2010), arXiv:0905.0531.
- Wang et al. (2009) D. S. Wang, A. G. Fowler, C. D. Hill, and L. C. L. Hollenberg, arXiv:0907.1708 (2009).
- Devitt et al. (2010) S. J. Devitt, A. G. Fowler, T. Tilma, W. J. Munro, and K. Nemoto, International Journal of Quantum Information (2010), to appear; preprint arXiv:0906.0415.
- DiCarlo et al. (2009) L. DiCarlo, J. M. Chow, J. M. Gambetta, L. S. Bishop, B. R. Johnson, D. I. Schuster, J. Majer, A. Blais, L. Frunzio, S. M. Girvin, et al., Nature 460, 240 (2009), arXiv:0903.2030.
- Passante et al. (2009) G. Passante, O. Moussa, C. A. Ryan, and R. Laflamme, Phys. Rev. Lett. 103, 250501 (2009), arXiv:0909.1550.
- Stock and James (2009) R. Stock and D. F. V. James, Phys. Rev. Lett. 102, 170501 (2009), arXiv:0808.1591.
- Devitt et al. (2007) S. J. Devitt, A. D. Greentree, R. Ionicioiu, J. L. O’Brien, W. J. Munro, and L. C. L. Hollenberg, Phys. Rev. A 76, 052312 (2007), arXiv:0706.2226.
- Stephens et al. (2008b) A. M. Stephens, Z. W. E. Evans, S. J. Devitt, A. D. Greentree, A. G. Fowler, W. J. Munro, J. L. O’Brien, K. Nemoto, and L. C. L. Hollenberg, Phys. Rev. A 78, 032318 (2008b), arXiv:0805.3592.
- Devitt et al. (2009) S. J. Devitt, A. G. Fowler, A. M. Stephens, A. D. Greentree, L. C. L. Hollenberg, W. J. Munro, and K. Nemoto, New. J. Phys. 11, 083032 (2009), arXiv:0808.1782.
- Van Meter et al. (2010) R. Van Meter, T. D. Ladd, A. G. Fowler, and Y. Yamamoto, International Journal of Quantum Information (2010), to appear; preprint arXiv:0906.2686.
- DiVincenzo (2009) D. P. DiVincenzo, arXiv:0905.4839 (2009), nobel Symposium on Qubits for Quantum Information.
- Gao et al. (2009) W.-B. Gao, A. G. Fowler, R. Raussendorf, X.-C. Yao, H. Lu, P. Xu, C.-Y. Lu, C.-Z. Peng, Y. Deng, Z.-B. Chen, et al., arXiv:0905.1542 (2009).
- Blakestad et al. (2009) R. B. Blakestad, C. Ospelkaus, A. P. VanDevender, J. M. Amini, J. Britton, D. Leibfried, and D. J. Wineland, Phys. Rev. Lett. 102, 153002 (2009), arXiv:0901.0533.
- Leibrandt et al. (2009) D. Leibrandt, J. Labaziewicz, R. Clark, I. Chuang, R. Epstein, C. Ospelkaus, J. Wesenberg, J. Bollinger, D. Leibfried, D. Wineland, et al., Quant. Info. Comp. 9, 901 (2009), arXiv:0904.2599.
- Home et al. (2009) J. P. Home, D. Hanneke, J. D. Jost, J. M. Amini, D. Leibfried, and D. J. Wineland, Science 325, 1227 (2009), arXiv:0907.1865.
- Hanneke et al. (2009) D. Hanneke, J. P. Home, J. D. Jost, J. M. Amini, D. Leibfried, and D. J. Wineland, Nature Physics 6, 13 (2009), arXiv:0908.3031.
- Amini et al. (2009) J. M. Amini, H. Uys, J. H. Wesenberg, S. Seidelin, J. Britton, J. J. Bollinger, D. Leibfried, C. Ospelkaus, A. P. VanDevender, and D. J. Wineland, arXiv:0909.2464 (2009).
- Fortágh and Zimmermann (2007) J. Fortágh and C. Zimmermann, Rev. Mod. Phys. 79, 235 (2007).
- Riedel et al. (2010) M. F. Riedel, P. B hi, Y. Li, T. W. H nsch, A. Sinatra, and P. Treutlein, arXiv:1003.1651 (2010).
- Bocko et al. (2008) M. F. Bocko, A. M. Herr, and M. J. Feldman, Rev. Mod. Phys. 80, 885 (2008).
- Becker et al. (2009) C. Becker, P. Soltan-Panahi, J. Kronjäger, S. Dörscher, K. Bongs, and K. Sengstock, arXiv:0912.3646 (2009).
- Bialczak et al. (2009) R. C. Bialczak, M. Ansmann, M. Hofheinz, E. Lucero, M. Neeley, A. O’Connell, D. Sank, H. Wang, J. Wenner, M. Steffen, et al., arXiv:0910.1118 (2009).
- Taylor et al. (2005) J. M. Taylor, H.-A. Engel, W. Dür, A. Yacoby, C. M. Marcus, P. Zoller, and M. D. Lukin, Nature Physics 1, 177 (2005).
- Hollenberg et al. (2006) L. C. L. Hollenberg, A. D. Greentree, A. G. Fowler, and C. J. Wellard, Phys. Rev. B 74, 045311 (2006), quant-ph/0506198.
- Lyon (2006) S. A. Lyon, Phys. Rev. A 74, 052338 (2006), cond-mat/0301581.
- Edmonds (1965a) J. Edmonds, Canad. J. Math. 17, 449 (1965a).
- Edmonds (1965b) J. Edmonds, J. Res. Nat. Bur. Standards 69B, 125 (1965b).