CROSSTALK

CROSS REFERENCE TO RELATED APPLICATION

[0001] This Application claims priority to U.S. Provisional Patent Application No.: 62/906,658, entitled ”Systems and Methods for Cancellation of Crosstalk,” hied September 26, 2019, which is hereby incorporated by reference in its entirety.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR

DEVELOPMENT

[0002] This invention was made with government support under W911NF-14-1-0103 awarded by the U.S. Army Research Office. The government has certain rights in the invention.

TECHNICAL FIELD

[0003] The present disclosures relate generally to scalable systems and methods for opti- mizing a set of gates of a quantum device.

BACKGROUND

[0004] Achieving the advantages of quantum computing and realizing practical quantum in- formation processing is a challenge due to an innate sensitivity to errors and noise. An aspect of this challenge is the dichotomy apparent in the implementation of high-hdelity simulta- neous one-qubit and high-hdelity two-qubit gates. Fast two-qubit gates require spatially or spectrally nearby qubits, which inherently reduces the addressability of the constituent subsystems. See, Gershenfeld et al., 1997, ”Bulk Spin-Resonance Quantum Computation,” Science Mag., 275, pg. 350; Piori et al., 2008, ”Electrically driven single-electron spin res- onance in a slanting Zeeman held,” Nature Physics, 4, pg. 776; each of which is hereby incorporated by reference in its entirety. A resonant pulse intended for one qubit can induce rotations on the others. Proposed architectures have typically dealt with this crosstalk by maximizing the gap between qubits or by executing local operations asynchronously. See, Vandersypen et al., 2005, ”NMR techniques for quantum control and computation,” Review of Modern Physics, 76, pg. 1037; Schutjens et al., 2013, ”Single-qubit gates in frequency- crowded transmon systems,” Physical Review A, 88, pg. 052330; Piltz et al., 2014, ”A trapped-ion-based quantum byte with 10-5 next-neighbour cross-talk,” Nature Communica- tions, 5, pg. 4679; Boixio et al., 2018, ”Characterizing Quantum Supremacy in Near-term devices,” Nature Physics, 14(6), pg. 595, each of which is hereby incorporated by reference in its entirety. The former solution requires the ability to tune couplings or extra engineer- ing, but the added complexity can adversely impact coherence times and requires additional control wires. In the latter approach, depending on the extent to which the control helds affect neighboring subsystems, the time overhead can be signihcant, and parallel operations should be employed.

[0005] In theory, only a small number of distinct gates are needed to perform fault-tolerant quantum computation. In practice, single-qubit operations have considerably higher fideli- ties than their multi-qubit counterpart. Thus, it is desirable to be able to apply any element of , thereby minimizing a number of two-qubit gates. Moreover, a high-hdelity implementation of the group is naturally suited for randomized compiling, which tailors coherent and spatially correlated errors on two-qubit gates into stochastic Pauli noise. See, Wallman et al., 2016, ”Noise tailoring for scalable quantum computation via randomized compiling,” Physical Review A, 94, pg. 1-10, which is hereby incorporated by reference in its entirety. In near-term applications, the much higher fault-tolerance thresholds and noise reduction afforded by the method will be valuable for fault-tolerant quantum information processing.

[0006] Furthermore, Crosstalk describes a broad range of effects that violate one of two assumptions: spatial locality and independence of operations. See Rudiger et al., 2019,”Probing Context-Dependent Errors in Quantum Processors,” Phys. Rev., X(9), pg. 021045; Sarovar et al., 2019, ”Detecting Crosstalk Errors in Quantum Information Processors,” arXiv preprint, arXiv: 1908.09855, print; Abrams et al., 2019,” Methods for Measuring Magnetic Flux Crosstalk between Tunable Transmons,” Phys. Rev., 12(6), pg. 064022, each of which is hereby incorporated by reference in their entirety. Gates and other operations are supposed to act on disjoint subsets of qubits. However, unintended interactions can couple the qubits, producing nonlocal correlated noise. Even if an operation has a well-defined action on a particular subset of qubits, the effective noise might depend on its context what operations affect other qubits.

[0007] Prior work has often approached the problem of implementing several operations on a collection of qubits by breaking the problem into a temporally-disjoint sequence of gates. See, Gershenfeld et al; Vandersypen et al; Piltz et al In contrast, other methods analyzed the problem by simultaneously driving two spins with a homogeneous held in the setting of NMR. See, Steffen et al., 2000, ”Simultaneous Soft Pulses Applied at Nearby Frequencies,” Journal of Magnetic Resonance, 146, pg. 369, which is hereby incorporated by reference in its entirety. However, application of theses methods to many-qubit or multilevel systems, such as transmons or trapped ions, is not clear. Additional methods have studied how to drive two transmons coupled to the same cavity suffering from spectral crowding with simultaneous X- or Y-gates with rotation angles p and p/2. See, Theis et al., 2016, ”Simultaneous gates in frequency-crowded multilevel systems using fast, robust, analytic control shapes,” Physical Review A, 93, pg. 012324, which is hereby incorporated by reference in its entirety. In either case, these methods do not directly apply to many-qubit systems, nor do they handle non-local correlations. Nevertheless, without modification, these algorithms would he in EXPSPACE.

[0008] The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.

SUMMARY

[0009] The present disclosure provides systems and methods for optimizing a set of gates on a quantum device.

[0010] One aspect of the present disclosure provides a digital computer system for esti- mating one or more error processes, including crosstalk, occurring in a quantum computer system. The quantum computer system comprises a plurality of qudits, and each qudit in the plurality of qudits has a dimension of N, where N is a positive integer of 2 or greater. The digital computer system comprises one or more digital processors, a memory, and one or more programs. The one or more programs are stored in the memory and are configured to be executed by the one or more digital processors. The one or more programs include, for each respective quantum operation in a set of at least one quantum operation, instructions for (i) identifying a plurality of interacting systems within the plurality of qudits during the respective quantum operation, wherein each respective interacting system in the plurality of interacting systems comprises a corresponding subset of the plurality of qudits, (ii) iden tifying a corresponding set of components from the plurality of interacting systems, (iii) for each respective component in the corresponding set of components, evolving the respective component along with the respective components in the corresponding set of components that interact with the respective component in the respective quantum operation, thereby forming a plurality of maps for the respective set of components, (iv) calculating, for each respective component in the corresponding set of components, a corresponding marginal distribution using the corresponding map for the respective component, thereby comput ing a plurality of marginal distributions for the corresponding set of components and (v) constructing a respective estimate of all or a portion of an entire Pauli error distribution for the respective quantum operation from the plurality of marginal distributions for the corresponding set of components.

[0011] In some embodiments, the respective quantum operation is a single-qubit gate or an entangling gate. For example, in some embodiments, the entangling gate is a cross-resonance gate or an iSWAP gate.

[0012] In some embodiments, the set of at least one quantum operation is a plurality of gates implemented in a temporal order on the quantum computer system.

[0013] In some embodiments, each component in the corresponding set of components is a set of qudits, in the plurality of qudits, that are coupled together in the respective quantum operation.

[0014] In some embodiments, the quantum system further comprises a plurality of drive helds, where each respective drive held in the plurality of drive helds is applied to a cor responding qudit in the plurality of qudits, and the one or more programs further include instructions for using each respective estimate of the portion of the entire Pauli error dis tribution or the entire Pauli error distribution for each respective quantum operation in the set of at least one quantum operation to optimize a plurality of parameters associated with the plurality of drive helds. In some embodiments, the plurality of drive helds includes a different drive held for each qudit in the plurality of qudits. In some embodiments, the plurality of parameters includes, for each drive held in the plurality of drive helds, an overall magnitude of a resonant quadrature, an off-resonant quadrature, a carrier signal phase, or a quadrature's phase.

[0015] In some embodiments, the method further comrprises using the respective estimate of an entire Pauli error distribution for each respective quantum operation to adjust the plurality of drive helds.

[0016] In some embodiments, a gate in the plurality of gates is an element of SU(2) or SU (4) . In some embodiments each gate in the plurality of gates is an element of the unitary group.

[0017] In some embodiments, the plurality of qudits is two or more physical qudits, four or more physical qudits, or 25 or more physical qudits.

[0018] In some embodiments, the set of at least one quantum operation is a single quantum operation. In some embodiments, the set of at least one quantum operation is a plurality of quantum operations. In some embodiments, the set of at least one quantum operation is 2, 3, 4, 5, or more than 5 quantum operations.

[0019] In some embodiments, the identifying the plurality of interacting systems (i) com prises constructing a corresponding graph G comprising a corresponding plurality of nodes and a corresponding plurality of edges, each respective node in the corresponding plurality of nodes represents a corresponding interacting system in the plurality of interacting systems during the respective quantum operation, and each edge in the corresponding plurality of edges denotes crosstalk that couples a hrst interacting system associated with a hrst node and a second interacting system associated with a second node in the corresponding graph. In some embodiments, the corresponding set of components includes each component, in each node in each subgraph of the corresponding graph, that has an order that is less than or equal to a predetermined maximum component order in the corresponding graph. In some embodiments, the respective components in the corresponding set of components that inter act with the respective component are within a predetermined environment distance within the quantum computer system to the respective component. In some embodiments, each edge in the corresponding plurality of edges denotes nonlocal crosstalk that couples a hrst interacting system associated with a hrst node and a second interacting system associated with a second node in the corresponding graph. In some embodiments, the predetermined maximum component order in the corresponding graph is a positive integer between 2 and 8.

[0020] Another aspect of the present disclosure provides a method for estimating one or more error processes, including crosstalk, occurring in a quantum computer system. In this aspect of the present disclosure, the quantum computer system comprises a plurality of qu- dits. Each qudit in the plurality of qudits has a dimension of N, where N is a positive integer of 2 or greater. The method comprises, at a digital computer system comprising one or more digital processors and a memory, the memory comprising non-transitory instructions conhg- ured to perform a procedure for each respective quantum operation in a set of at least one quantum operation. The procedure comprises instructions for (i) identifying a plurality of interacting systems within the plurality of qudits during the respective quantum operation, wherein each respective interacting system in the plurality of interacting systems comprises a corresponding subset of the plurality of qudits, (ii) identifying a corresponding set of com ponents from the plurality of interacting systems; (iii) for each respective component in the corresponding set of components, evolving the respective component along with the respec tive components in the corresponding set of components that interact with the respective component in the respective quantum operation, thereby forming a plurality of maps for the respective set of components; (iv) calculating, for each respective component in the corre sponding set of components, a corresponding marginal distribution using the corresponding map for the respective component, thereby computing a plurality of marginal distributions for the corresponding set of components; and (v) constructing a respective estimate of all or a portion of an entire Pauli error distribution for the respective quantum operation from the plurality of marginal distributions for the corresponding set of components.

[0021] Another aspect of the present disclosure provides a non-transitory computer read able storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a digital computer system with one or more digital processors, cause the digital computer system to perform a method of estimating one or more error processes, including crosstalk, occurring in a quantum computer system, the quantum computer system comprising a plurality of qudits. Each qudit in the plurality of qudits has a dimension of N, where N is a positive integer of 2 or greater. The method comprises, for each respective quantum operation in a set of at least one quantum operation, instructions for: (i) identifying a plurality of interacting systems within the plurality of qudits during the respective quantum operation, wherein each respective interacting system in the plurality of interacting systems comprises a corresponding subset of the plurality of qudits; (ii) iden tifying a corresponding set of components from the plurality of interacting systems; (iii) for each respective component in the corresponding set of components, evolving the respective component along with the respective components in the corresponding set of components that interact with the respective component in the respective quantum operation, thereby forming a plurality of maps for the respective set of components; (iv) calculating, for each respective component in the corresponding set of components, a corresponding marginal distribution using the corresponding map for the respective component, thereby comput- ing a plurality of marginal distributions for the corresponding set of components; and (v) constructing a respective estimate of all or a portion of an entire Pauli error distribution for the respective quantum operation from the plurality of marginal distributions for the corresponding set of components.

[0022] Another aspect of the present disclosure provides a digital computer system for esti- mating one or more error processes, including crosstalk, occurring in a quantum computer system. The quantum computer system comprises a plurality of qudits. Each qudit in the plurality of qudits has a dimension of N, where N is a positive integer of 2 or greater. The digital computer system comprises one or more digital processors, a memory, and one or more programs. The one or more programs are stored in the memory and are configured to be executed by the one or more digital processors. The one or more programs include, for each respective quantum operation in a set of at least one quantum operation, instruc- tions for (i) constructing a corresponding graph G comprising a corresponding plurality of nodes and a corresponding plurality of edges, where each respective node in the correspond- ing plurality of nodes represents a corresponding interacting system during the respective quantum operation, where the corresponding interacting system comprises a corresponding subset of the plurality of qudits, and where each edge in the corresponding plurality of edges denotes crosstalk that couples a hrst interacting system associated with a hrst node and a second interacting system associated with a second node in the corresponding graph, (ii) identifying a corresponding set of components from the corresponding graph G, where the corresponding set of components includes each component, in each node in each subgraph of the corresponding graph, that has an order that is less than or equal to a predetermined maximum component order in the corresponding graph, (iii) for each respective component in the corresponding set of components, evolving the respective component along with the components at each of the nodes within a predetermined environment distance within the quantum computer system to the respective component in the corresponding graph, thereby forming a plurality of maps for the respective set of components, (iv) calculating, for each respective component in the corresponding set of components, a corresponding marginal distribution using the corresponding map for the respective component, thereby comput ing a plurality of marginal distributions for the corresponding set of components, and (v) constructing a respective estimate of all or a portion of an entire Pauli error distribution for the respective quantum operation from the plurality of marginal distributions for the corresponding set of components.

BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figures 1A, 1B, 1C, and 1D collectively illustrate a schematic depicting expansions with order (1, 1) and (1, 2) of a quinquepartite system with nearest-neighbor (NN) coupling, in accordance with an exemplary embodiment of the present disclosure;

[0024] Figure 2 illustrates a plot of an average process inhdelity of a 3-quibit logical subspace for a system of three superconducting qutrits, in accordance with an exemplary embodiment of the present disclosure;

[0025] Figure 3 illustrates a graphical model of a system of twenty qubits on a grid with nearest-neighbor non-local crosstalk, in accordance with an exemplary embodiment of the present disclosure;

[0026] Figure 4 illustrates a plot of an average single-qubit process indidelity as a function of time, in accordance with an exemplary embodiment of the present disclosure; and [0027] Figure 5 illustrates an average two-qubit process inhdelity for a square array of transmons, in accordance with an exemplary embodiment of the present disclosure.

[0028] Figure 6 is a schematic diagram showing aspects of an example system in accordance with an exemplary embodiment of the present disclosure.

[0029] Like reference numerals refer to corresponding parts throughout the several views of the drawings.

DETAILED DESCRIPTION

[0030] Reference will now be made in detail to embodiments, examples of which are il lustrated in the accompanying drawings. In the following detailed description, numerous specihc details are set forth in order to provide a thorough understanding of the present disclosure. However, it will be apparent to one of ordinary skill in the art that the present disclosure may be practiced without these specihc details. In other instances, well-known methods, procedures, components, circuits, and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

[0031] It will also be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a hrst subject could be termed a second subject, and, similarly, a second subject could be termed a hrst subject, without departing from the scope of the present disclosure. The hrst subject and the second subject are both subjects, but they are not the same subject.

[0032] The terminology used in the present disclosure is for the purpose of describing par- ticular embodiments only and is not intended to be limiting of the invention. As used in the description of the invention and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that the term “and/or” as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items. It will be further understood that the terms “comprises” and/or “comprising,” when used in this specihcation, specify the presence of stated features, integers, steps, operations, el- ements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. [0033] As used herein, the term “if’ may be construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context. Similarly, the phrase “if it is determined” or “if [a stated condition or event] is detected” may be construed to mean “upon determining” or “in response to determining” or “upon detecting [the stated condition or event]” or “in response to detecting [the stated condition or event],” depending on the context.

[0034] An aspect of the present disclosure is directed to providing efficient systems and methods for optimizing an implementation of nontrivial parallel operations. Another aspect of the present disclosure is directed to providing efficient systems and methods for optimizing an implementation of nontrivial parallel operations under general crosstalk.

[0035] One aspect of the present disclosure provides general and scalable systems and meth- ods for implementing high-hdelity simultaneous gates on qubits experiencing crosstalk ef- fects. In some embodiments, the present disclosure exploits the tensor product structure of classical crosstalk to simulate and optimize gates efficiently. Accordingly, the present disclosure remedies the problem of constructing on superconducting qubits. Fur- thermore, the systems and methods of the present disclosure optimize concurrent multiqubit gates and modify an approach to account for nonclassical crosstalk. Taking experimental restrictions into account, a simple control ansatz is provided that allows for reducing the effect of crosstalk dramatically. Moreover, average Hamiltonian theory is applied to simplify the optimization procedure further and gain insight into the cancellation. The inhdelity is improved by more than an order of magnitude despite strong crosstalk for a model of superconducting qubits with realistic parameters.

[0036] Furthermore, the present disclosure provides scalable systems and methods for ac- curately modeling idle and operation crosstalk on a plurality of experimental devices. In some embodiments, the present disclosure exploits a tensor product structure of local (e.g., classical) crosstalk to efficiently express its impact on gates. Through a perturbative expan sion, the present disclosure is applicable to non-local (e.g., quantum) crosstalk and captures its effects to arbitrary order (e.g., any desirable order). In some embodiments, provided there is a sufficient degree of control, the present disclosure can minimize an effect of these errors. Moreover, the present disclosure illustrates applications of the systems and methods described herein through a series of simulations of parallel gates on a square array of one hundred superconducting transmon qubits.

[0037] In a hrst example, the present disclosure applies a gradient-based optimization to an experimentally signihcant problem of implementing arbitrary elements of on superconducting transmon qubits. Despite substantial local crosstalk, the present disclosure illustrates that error rates near the crosstalk-free limit are possible with conventional control hardware. Furthermore, the present disclosure illustrates how to tuneup simultaneous cross- resonance gates and, again, obtain dramatically improved error rates. The results described herein suggest that contrary to prevailing opinions, crosstalk need not be a prohibitive limitation on noisy intermediate-scale quantum (NISQ) era devices. See Murali et al ., 2020,”Architecting Nosiy Intermediate-Scale Trapped Ion Quantum Computers,” ArXiv preprint, ArXiv: 2004.04706, print; Preskill, J., 2018, ” Quantum Computing in the NISQ Era and Beyond,” Quantum, 2, pg. 79, each of which is hereby incorporated by reference in their entirety. Based on the present disclosure, higher quality quantum information processors can be made possible by using the systems and methods described herein to better balance the tradeoffs in a fabrication of a device and pulse design.

[0038] Theory of Crosstalk. The present disclosure addresses a foundational problem: what crosstalk directly affects physical qubits (e.g., as opposed to measurement or line crosstalk), and is the crosstalk efficiently simulatable or mitigatable? Typically, crosstalk is classified as either quantum or classical. Classical (local) crosstalk arises if a semiclassical drive field interacts with multiple qubits, introducing unitary errors on nontarget qubits, but not entangling independent systems. Quantum crosstalk entangles qubits and originates from, for example, the effect of a static coupling when it is not being used to induces a cross resonant interaction. See, Chow et al., 2011, ”Simple All-Microwave Entangling Gate for Fixed-Frequency Superconducting Qubits,” Physical Review Letters, 107, pg. 080502, which is hereby incorporated by reference in its entirety.

[0039] Although classical crosstalk produces correlated noise, the crosstalk can be factorized and simulated efficiently on a digital computer. The induced correlations are classical and, therefore, do not entangle the individual subsystems. Concretely, some embodiments of the present disclosure model classical crosstalk via the Hamiltonian

Each local Hamiltonian H _n acts exclusively on subsystem n, and denotes shared classical parameters that result in crosstalk. Typically, consists of variables that define the drive fields. The average process fidelity F between a target operation and the noisy implementation , where , can be expressed as

See, Nielsen et al., 2002, ”Quantum Computation and Quantum Information: 10th Anniver- sary Edition,” Cambridge University Press, 10th edition, print; Carignan- Dugas et al., 2019,”Bounding the average gate fidelity of composite channels using the unitarity,” New Journal of Physics, 21, pg. 053016; each of which is hereby incorporated by reference in its entirety. The equation holds more generally when are CPTP maps.

[0040] Unlike its classical counterpart, a digital computer cannot efficiently simulate an extensive system affected by quantum crosstalk. Thus, the present disclosure provides a perturbative approach for optimizing the controls of a system experiencing localized quantum crosstalk. [0041] For convenience, an approach of the present disclosure is defined through a graphical model of the noise. However, the graph based approach is simply for presentation of the method and the present disclosure is not limited to graph based approaches. Turning to the graph based approach, a graph G is formed where each node represents a strongly interacting subsystem, such as a two-qubit pair entangled by a cross-resonant coupling. The target operation on the entire system is factorizable over the tensor product space dehned by the node subsystem partitioning. Edges denote the quantum crosstalk that inadvertently couple the subsystems. Accordingly, a constraint is imposed that the crosstalk is local: a maximum degree of the graph lies below a constant D _max that quantihes available computational resources and depends on the expansion order. In contemporary architectures, the constraint is satished since a majority of nodes have a degree of at most four.

[0042] The expansion order is designated by a pair (d, o ), with d specifying the environment depth and o the minimum partition order. A further graph G' is formed by repeatedly cutting G into components such that no cut creates a component with an order less than o. The cutting procedure terminates if no further cuts are possible. The behavior of a component in G' is approximated by evolving the component along with all vertices of distance at most d. These neighboring vertices constitute the environment, and tracing them out approximates the reduced dynamics of the component. A factorizable initial system-environment state is assumed and the environmental degrees of freedom determine a maximally noncommutative state. Using the tensor product factorization over G', an objective function analogous to (25) is formed. In general, the graph G' is not unique, and the objective function should average over all valid graphs or utilize additional information about the noise structure. [0043] As discussed above, the present disclosure is not limited to graph based appraoches. For example, an algorithm could simulate a target system (e.g., node in the graph) and other qudits that appear to or are expected to interact with it. For example, one could run simultaneous random benchmark compiling and use the resulting data to hnd “environment” qubits that induce most of the noise on the target system. One could then optimize their controls without ever making the graph. Random benchmarking and related error correction protocols are disclosed in United States Patent Number 10,360,088 and 10,031,791, each of which is hereby incorporated by reference.

[0044] Referring to Figures 1A through 1C, a collective illustration is provided as to how to apply the present disclosure to a chain of coupled systems. Figures 1A through ID collectively show the graph of a quinquepartite chain and describes how to approximate the collective behavior. For a sufficiently large chain, a (d, o)-order expansion requires the complete simulating of at most 2d + o subsystems.

[0045] Figures 1A through 1D collectively depict a schematic of expansions with order (1, 1) and (1, 2) of a quinquepartite system with nearest-neighbor (NN) coupling. In Figure 1A, the collective dynamics is approximated by evolving each subsystem and its nearest neighbors (NNs) before tracing out their environment (e.g., neighbors). In Figures 1B and 1C, a second-order estimate of the overall dynamics is performed by evolving bipartite and tripartite subsystems and their NNs. Figures 1B and 1C show two different partitionings that are averaged over to better approximate the actual dynamics. Further, Figure 1D illustrates the partitioning of Figure 1C after tracing out each environment. By exploiting the multiplicativity of the process fidelity, the systems and methods of the present disclosure estimate and maximize the quantity without constructing the complete system.

[0046] Now that details of the basic theory of crosstalk has been disclosed, details re- garding a practical application for mitigating the effects of crosstalk on superconducting transmon qubits, in accordance with an embodient of the present disclosure, are disclosed. The present disclosure describes the systems and methods with a focus on transmons for the sake of clarity. However, the present disclosure is not limtied thereto. For instance, in some embodiments, the systems and methods of the present disclosure are applicable to genuine two-level qubits and, with appropriate modffication, other multilevel implementations such as trapped ions.

[0047] A typical implementation of single-qubit operations on transmons will now be de- scribed. A local oscillator acts as a single tone microwave source outputting a constant signal cos(w't) that is shaped by an arbitrary waveform generator via an IQ mixer. A description of a transmon qubit is an anharmonic oscillator driven by microwave pulses. In the lab frame, the relevant Hamiltonian is where is the annihilation operator of the oscillator, , a is the anharmonicity, g is the drive's phase, w is the oscillator's resonant frequency, W(t) specihes the drive envelope, and set .

[0048] The hrst two energy levels form the qubit subspace. For a sufficiently large anhar- monicity, the leading-order effect of leakage is modeled by the inclusion of the third energy level, and the qutrit state evolves under the truncated Hamiltonian where , and l is the relative strength at which the control drives the ) leakage transition (assume l = 2 throughout the present disclosure). In the qubit rotating frame, after making a rotating wave approximation, the Hamiltonian becomes

[0049] To understand how the control creates single-qubit gates, consider a resonant pulse (w = w'), and A = 0, which corresponds to an ideal sufficiently long pulse. The control generates X and Y gates by modulating the coupling between the zero and one states, while the drive phase fixes the rotation axis in the XY -plane, and the pulse area sets the rotation angle. For example, g = 0 and g = p /2 result in rotations about the X- and Y-axis. Rotations about the remaining 21-axis correspond to a change in the relative phase between the states. Rather than manipulating the state of the transmon, an equivalence is to rotate the control with respect to the state, realizing a virtual-Z gate. See, Knill et al., 2000, ”An algorithmic benchmark for quantum information processing,” Nature, 404, pg. 368; Knill et al., 2008, ”Randomized benchmarking of quantum gates,” Physical Review A, 77, pg. 012307; Johnson et al., 2015, ”Demonstration of robust quantum gate tomography via randomized benchmarking,” New Journal of Physics, 17, pg. 113019; McKay et al., 2016,”Universal Gate for Fixed- Frequency Qubits via a Tunable Bus,” Physical Review A., 6, pg. 064007; each of which is hereby incorporated by reference in its entirety. This equivalence is physically accomplished by adding a phase offset to all subsequent gates.

[0050] A pulse with an area and a relative phase offset g generates the unitary with the notation A _q = exp(— ίq A/2). Combining two of these phase offset p/2 pulses and a final virtual-Z realizes any element of SU(2),

[0051] Now that details regarding superconducting qubits have been disclosed, details re- garding single-qubit gate engineering, in accordance with an embodient of the present dis- closure, are disclosed. The semiclassical Hamiltonian governing transmon n with classical crosstalk is

The parameter b and q characterize the crosstalk affecting the system. Accordingly, a par- ticular case where each transmon has a local drive will now be described. However, the present disclosure is not limited thereto. The crosstalk parameters are N x N matrices and the constraints and are imposed. These constraints lead to interpreting b as the relative drive strength and Q is the phase lag. Experimental data supports the model. See, Xia et al., 2015, ”Randomized Benchmarking of Single-Qubit Gates in a 2D Array of Neutral- Atom Qubits,” Physical Review Letters, 114, pg. 100503; Magesan et al., 2018, ”Ef- fective Hamiltonian models of the cross-resonance gates,” arXiv preprint arXiv: 1804.04073, print; Xue et al., 2019, ”Benchmarking Gate Fidelities in a Si / SiGe Two-Qubit Device Physical Review X, 9, pg. 21011; each of which is hereby incorporated by reference in its entirety.

[0052] Accordingly, besides b and q , the parameters in (7) with standard Ramsey tomog- raphy can be estimated. First consider the case where there are no extraneous qubit-qubit interactions. Each element of the crosstalk parameters is isolated by only turning on drive j, and measuring transmon n. In the rotating frame of the transmon, neglecting leakage, and making a rotating-wave approximation, the Hamiltonian is with , and = 0.

[0053] If it is not possible to tune the drive frequency, then a numerical fitting procedure is necessary. Otherwise, if the drive frequency is tuned so that and that the drive amplitude is constant, the expectation of the Pauli matrices at a later time t are

The parameter and are determined via fitting, and an exponential rescaling should be incorporated to account for decoherence. Repeating the procedure for all N ² crosstalk interactions gives estimates for all elements of b and q . [0054] Details regarding static qubit-qubit coupling are now provided. In several experimen tal architectures, constant qubit-qubit couplings facilitate two-qubit operations, such as the cross-resonance gate. See, Rigetti et al., 2015, ”Protocol for Universal Gates in Optimally Biased Superconducting Qubits,” Physical Review Letters, 94, pg. 240502, which is hereby incorporated by reference in its entirety. However, such a static coupling can adversely affect an ability to characterize crosstalk precisely, especially if . Therefore, decoupling the interaction, provided , where is the time it takes to apply a p rotation with drive n to qubit n, is desired. In practice, fast p pulses are feasible on transmons by implementing DRAG corrections. Accordingly, a case of a single static ZZ interaction is analyzed. The joint Hamiltonian of the target and coupled systems is where k is the strength of the qubit-qubit coupling, and classical crosstalk affecting the coupled qubit is neglected since any effects on the target are second order.

[0055] The correction scheme for measuring the s ^y observable is analogous to the procedure for s ^x . For simplicity, consider the limiting case t _p 0. A round of decoupled evolution consists of a time Dt where the system evolves under (10), an X _p , further evolution un- der (10) for a time Dt and finally an X _p . The lowest-order contribution to the average Hamiltonian during the round is

Thus, by performing N rounds with a sufficiently small D, the system evolves approximately under (11). The desired observable is and the total experiment time is T = 2 N D. Higher-order cancellation of the coupling is possible with e.g., Richardson extrapolation.

[0056] Now, details regarding the technical aspects of optimizing crosstalk are disclosed. The hrst two orders of the effective Hamiltonian are [0057] Consider a target operation and its approximation obtained via average Hamil- tonian theory where

Rather than maximizing the fidelity, the logarithm of the overlap is maximzed which shares the same extrema but is linear in its subsystems and has the convenient form which is well-defined for non-orthogoal unitaries, and implicity define . The partial derivative with respect to some scalar is

The following lemmas assist in computation of the gradient for a general Hamiltonian. The first provides a numerical routine for evaluating the gradient. The second allows for eliminating the final virtual-Z angle from the optimization problem.

Lemma 1 Let H be a Hermitian matrix with an eigendecomposition then where .

[0058] Virtual-Z correction.

Lemma 2 Let then the overlap between A and Z _q satisfies and the bound is saturated when [0059] The present disclosure considers a problem of implementing an arbitrary element of concurrently on an ensemble of qubits where their respective drive helds weakly interact with other qubits.

[0060] In some embodiments, the present disclosure considers a related problem of how one can implement an arbitrary element concurrently on an ensemble of qubits if their respective drive helds weakly interact with other qubits? Central to an approach of the present disclosure for the problem is the observation that a digital computer can efficiently simulate the dynamics under the experimentally relevant semiclassical drive approximation. Rather than modeling an explicit tensor product space, each subsystem independently is modeled, and there is no inherent obstacle to computing corrections that maximize the hdelity of the entire system.

[0061] It is necessary to pick control parameters to optimize. On the one hand, forming pulse shapes that yield error rates near the decoherence limit for short gate times is desireable. On the other hand, one must consider experimental realities, such as restrictions on engineering pulse transfer functions and the degree of calibration needed to implement complicated pulses accurately. Accordingly, control to the phases in (6) is resrticted. By appropriately adjusting the phases, unitary error is accountable that is independent of the phases by compiling the inverse gate into the implementation. Moreover, since the phases is only changed, the pulse envelopes remain constant and, therefore, require minimal calibration.

[0062] A Magnus expansion trick that avoids time integration over the gate time is applied. Applying the average Hamiltonian theory, the effective Hamiltonian is satisfying during an interval. Upon moving into the qutrit's rotating frame, and making a rotating- wave approximation the hrst two terms have the form where a, b, c and d are calculated by numerical integration and depend on model parameters except for F _j . Thus, control of the phases is optimized without needing to repeatedly solve the system. [0063] A situation where subsystems are somewhat addressable is now described. Consider a driving held intended for one subsystem interacting with others with a relative amplitude on the order of 10%.

[0064] Figure 2 illustrates a plot of the average process inhdelity of the 3-qubit logical subspace for a system of three superconducting qutrits. Figure 2 depicts the raw unoptimized inhdelity (e.g., squares), the hrst-order Magnus-based optimized inhdelity (e.g., circles), the optimal inhdelity found by an exhaustive numerical search (e.g., triangles), and the decoherence-limited bound on inhdelity ((e.g., the dashed line). A minimum mean inhdelity of 6.04 x 10 ^-4 occurs when t _p/2 = 11 ns.

[0065] Accordingly, the systems and methods of the present disclosure is examplihed with a simulation of superconducting qubits. In this hrst example, consider N = 3 transmons that evolve under (7), and realistic parameter values. See, Takita et al., 2016, ”Demonstration of Weight-Four Parity Measurements in the Surface Code Architecture,” Physical Review Letters, 117, pg. 210505; Takita et al., 2017, ”Experimental Demonstration of Fault- Tolerant State Preparation with Superconducting Qubits,” Physical Review Letters, 119, pg. 180501; each of which is hereby incorporated by reference in its entirety. Set = 4.0, 4.1, 4.2 GHz, a/2p = 330 MHz. In each iteration of this example, the target gate is chosen uniformly at random from SU(2) ³ . The crosstalk phase lag parameter q _jk is sampled uniformly from (0, 2p], and draw b _jk from a normal distribution centered at zero and [std b _jk ] = 0.05. Each oscillator is assumed to exhibit elementary Markovian dephasing and amplitude damping with T ₁ = 50 ms, and T ₂ = 60 ms . There are two discrete periods of successive evolution, each taking time t _p/2 . The two discrete periods differ in their choice of control envelopes W _j (t), and phase offsets 4> _j . In this example, the control envelopes are gaussian, with stdW _j = t _p/2 /4 and implement p/2 pulses. The phase offsets are optimized to approximate a target operation up to a virtual- Z rotation.

[0066] In Figure 2, a plot of the inhdelity as a function of t _p/2 is illustrated. The raw unoptimized inhdelity, illustrated as squares, without correcting the phases is much larger than the optimal inhdelities, illustrated as a circle and a triangle. The circles are obtained through the Magnus expansion approach described herein, while the triangles correspond to an exhaustive numerical search. An exhaustive search with gradient ascent of analytic controls (GOAT) is performed. See, Machnes et al., 2018, ”Tunable, Flexible, and Efficient Optimization of Control Pulses for Practical Qubits,” Physical Review Letters, 120, pg. 150401, which is hereby incorporated by reference in its entirety. Although the exhaustive method returns marginally lower infidelities, the relative computation time exceeds 10 ² and surpasses 10 ⁴ for more extensive systems on the order of ten qubits. The decoherence limited lower bound on the average process hdelity is further plotted, which is given by where t _g = 2t _{p /2} . Accordingly, an optimal average inhdelity is found to be 6.04 x 10 ^-4 with a gate time of 22 ns for implementing an element of .

[0067] A vital property for the experimental application of a pulse engineering method is its robustness to imprecision in the model parameters. For instance, in some embodiments the model adds normally distributed noise to b and q , applies a least square fit to obtain estimates of b and q .

[0068] The ability to address individual qubits is imperative for scalable quantum infor- mation processing. On practically all experimental implementations, optimal two-qubit gates require closely spaced qubits, but close spacings reduce single-qubit addressability and induce classical-crosstalk in which signals intended for one qubit inadvertently influ- ence others. Accordingly, the present disclosure provides a general and scalable method for executing arbitrary operations in parallel while simultaneously suppressing classical and quantum crosstalk. Furthermore, computational experiments of the present disclosure show drastically lower error rates for superconducting qubits.

[0069] Also, the present disclosure considers what crosstalk acts on physical qubits dur- ing idling or the implementation of gates (e.g., as opposed to preparation or measurement crosstalk), and how one can efficiently simulate and consequently, try to mitigate it. It is natural to classify crosstalk as either local or non-local. See Sarovar et al.. Local crosstalk can arise when a semi-classical drive held interacts with several qubits, causing unitary errors on supposedly idle qubits, but not entangling independent subsystems. Non-local crosstalk creates correlations that are nonfactorizable over system qubits and, in some em- bodiments, originate from, for example, the residual static coupling between two qubits or miscalibration.

[0070] Quantifying and reducing crosstalk requires a hgure of merit. In some embodiments, the present disclosure evaluates the average hdelity of one-qubit or two-qubit gates, rather than the hdelity per clock cycle. While the present disclosure functions in either case, the descriptions presented infra focus on the former situation. For instance, local error measures relate directly to fault-tolerance thresholds, are easier to estimate experimentally, and are more common in the literature. The present disclosure illustrates that the average local hdelity can be approximated in a straight-forward manner.

[0071] Local crosstalk. Although local crosstalk (typically) produces correlated noise, it can be factorized and simulated efficiently on a digital computer. The induced correlations are classical and do not entangle the individual subsystems. As such, some embodiments of the systems and methods of the present disclosure localize crosstalk via the Hamiltonian

Each term H _k acts exclusively on subsystem k. and denotes shared classical parameters that result in crosstalk. In some embodiments, the vector x , for example, includes, or contains, the phases and amplitudes that specify drive helds. The average process hdelity F between a target operation and the noisy implementation , where , can be expressed as

See Neilson, M., 2002, ”A Simple Formula for the Average Gate Fidelity of a Quantum Dynamical Operation,” Physics Letters A, 303(4), pg. 249; Carignan-Dugas et al., 2019, ”A Polar Decomposition for Quantum Channels (with Applications to Bounding Error Propa- gation in Quantum Circuits),” Quantum, 3, pg. 173, each of which is hereby incorporated by reference in its entirety. The equation holds more generally when are CPTP maps, for example, when a dissipative process also affects the system or the control parameters fluctuate over time.

[0072] Non-local crosstalk. Unlike local crosstalk, a digital computer cannot usually exactly simulate a large system affected by non-local crosstalk. Thus, the present disclosure provides a perturbative technique for simulating non-local crosstalk. An approximation scheme of the present disclosure characterizes a noise channel by estimating a portion of the associated Pauli error rates . The Pauli-twirled noise channel is where is the Pauli group on n qubits. These error rates provide a partial description of the noise affecting a quantum system and can assist in experimental device calibration. On large experimental devices, the systems and methods of the present disclosure can scalably estimate the parameters in a way that is robust to state preparation and measurement (SPAM) errors. See Flammia et al., 2019, ”Efficient Estimation of Pauli Channels,” arXiv preprint, arXiv: 1907.12976, print, which is hereby incorporated by reference in its entirety. In some embodiments the quantities are combined to calculate holistic measures of device performance, such as the average two-qubit fidelity or global fidelity.

[0073] It is helpful to illustrate the systems and methods of the present disclosure using a graphical model of the noise. See Diestel, R., 2005, ”Graph Theory. 2005,” Grad. Texts in Math, 101, print, for basic graph theory definitions, which is hereby incorporated by reference in its entirety. Some embodiments of the present disclosure construct a graph G where each node is a strongly interacting subsystem during an operation of interest, such as a qubit during a single-qubit gate or a two-qubit pair entangled by a cross-resonance interaction. The entire target operation is factorizable over the tensor product space partitioning defined by the nodes. Edges denote non-local crosstalk that couples subsystems, and the systems and methods of the present disclosure only allow two-body coupling. Some embodiments of the systems and methods of the present disclosure impose a constraint that the graph has limited connectivity (e.g., in a spatial and spectral sense) since the present disclosure relies on simulating subsystems. The constraint is satisfied in contemporary architectures where a majority of nodes have a degree of at most four.

[0074] Referring to Figure 3, a graphical model depicting a system of 20 qubits on a grid with nearest-neighbor non-local crosstalk is illustrated. One of thirty-one (there are 31 edges; one for each non-local interaction) noise simulations needed to estimate all marginal weight-2 Pauli error probabilities with the approximation (d, o ) = (1, 2) is highlighted. The solid black and hashed nodes denote the target subsystem and the environment,.

[0075] A pair of positive integers (d , o) specifies the expansion order of the noise approxi- mation; d designates the ‘environment’ distance and o the maximum component order. The present disclosure considers the set of all components of all induced sub-graphs of G such that the order of every component is less than or equal to o, and any component with an order less than o has the same edges as in G. Said otherwise, the present disclosure does not look at induced components with order less than o. The idea of the simulation scheme is to calculate the Pauli errors that occur on each component.

[0076] The systems and methods of the present disclosure approximate the behavior of a component by evolving it along with all vertices of distance at most d. generating a map - Next, the diagonal of the Pauli-Liouville representation of the channel is deter- mined, which is sometimes also known as the Pauli hdelity vector . A Walsh-Hadamard transformation W relates to the Pauli probability vector , with . See Flammia et al., 2019. The vector is the error probability distribution for a Pauli twirled copy of - Marginalizing the error distribution over the environment produces an estimate of the local error distribution on the target component. After calculating the marginal distributions for all of the components in , the systems and methods described herein can use the theory of probabilistic graphical models to construct an estimate of the entire Pauli error distribution up to some specihed error weight. See Koller et al., 2009, ” Probablistictic Graphical Models: Principles and Techniques,” MIT Press, print, which is hereby incorpo- rated by reference in its entirety. By truncating the distribution at some error weight, the size of the distribution scales polynomially in the number of qubits.

[0077] In some embodiments, including the nearest environmental nodes is sufficient to compute the local error distribution with high relative precision. The systems and meth- ods described herein allow for one to intuitively understand the limited-depth requirement from the fact that intermediate systems must mediate the influence of one subsystem on a nonadjacent subsystem. In some embodiments, these effects are formally bound with Lieb- Robinson bounds. See Lieb et al. , 1972, Communications in Mathematical Physics, 28, pg. 251, which is hereby incorporated by reference in its entirety.

[0078] Single-qubit gate engineering. The present disclosure provides a review of an em- bodiment of single-qubit operations on transmons. A local oscillator acts as a single tone microwave source outputting a constant signal cos(w't) that is shaped by an arbitrary wave- form generator via an IQ mixer. A good description of a transmon qubit is an anharmonic oscillator driven by microwave pulses. In the lab frame, the relevant Hamiltonian is where a is a annihilation operator of an oscillator, . a is an anharmonicity, g is a drive phase, w; is a resonant frequency of the oscillator, W(t) specihes a drive envelope, and setting . [0079] The lowest two energy levels form the qubit subspace. Including the third energy level models the leading-order effect of leakage provided the anharmonicity is sufficiently large. After making a rotating wave approximation (RWA) and moving into the rotating frame of the qubit, the Hamiltonian projected into the qubit subspace is

To see how the control induces single-qubit gates, consider a resonant pulse (w = w'), and l = 0, which corresponds to an ideal sufficiently long pulse. The control generates X and Y gates by modulating the coupling between the zero and one states, while the drive phase fixes the rotation axis in the XY-plane, and the pulse area sets the rotation angle. Rotations about the remaining Z-axis correspond to a change in the relative phase between the states. Rather than manipulating the transmon's state, it is equivalent to rotate the control with respect to the state, realizing a virtual- Z gate. See, Knill et al., 2000, ”An Algorithmic Benchmark for Quantum Information Processing,” Nature, 404(67776), pg. 368; Knill et al. , 2008, ”Randomized Benchmarking of Quantum Gates,” Physical Review A, 77(1), pg. 012307; McKay et al., 2017, ”Efficient Z Gates for Quantum Computing,” Physical Review A, 96(2), pg. 022330, each of which is hereby incorporated by reference in its entirety. The systems and methods of the present disclosure accomplish this physically by adding a phase offset to all subsequent gates. A pulse with an area and a relative phase offset 7 generates the unitary with the notation A _q = exp(— iq A/2). Combining two of these phase-offset p/2 pulses and a final virtual-Z realizes any element of SU(2). See McKay et al., 2017, ”Efficient Z Gates for Quantum Computing,” Physical Review A, 96, pg. 022330, which is hereby incorporated by reference in its entirety.

[0080] Referring to Figure 4, a plot is provided illustrating a significant improvement in the average single-qubit process infidelity as a function of the time for a p/2 gate (the total simulation time is 2t _p/2 ). There are 100 qubits in a square 2D array, and each qubit implements a random element of SU(2) via two p/2 pulses with intermediate phase offsets. The diamonds denote the infidelity of the qubits with half-derivative DRAG corrections and no crosstalk. The black circles, squares and triangles are infidelities obtained under the same control with crosstalk. More specifically, the black circles, squares, and triangles denote several relative crosstalk strengths b _ij , that are sampled from a normal distribution with mean zero and standard deviations s = 0.05, s = 0.1, and s = 0.25 respectively. The grey circles, squares, and triangles have identical crosstalk as their black counterparts, but with optimized control parameters.

[0081] Consider the problem of implementing an arbitrary element of concurrently on an ensemble of qubits where their respective drive helds weakly interact with other qubits. The semiclassical Hamiltonian governing transmon k with local drive crosstalk is

The parameters b and q characterize the crosstalk affecting the system. The present disclo- sure focuses on the case where each transmon has a local drive. The crosstalk parameters are n x n matrices, and one can set b _kk = 1 and q _kk = 0 without loss of generality by modifying . These constraints lead to interpret b as a relative drive strength, and Q as a phase lag. Experimental data supports the model, and one can efficiently esti- mate the parameters with standard Rabi and Ramsey experiments. See Xia et al., 2015,”Randomized Benchmarking of Single-Qubit GAtes in a 2D Array of Neutral-Atom Qubits,” Physical Review Letters, 114(10), pg. 100503; Magesan et al., 2020, ”Effective Hamiltonian Models of the Cross-Resonance Gates,” Physical Review A, 101(5), pg. 052308; Xue et al.,”Benchmarking Gate Fidelities in a Si/SiGe Two-Qubit Device,” Physical Review X, 9(2), 2019, pg. 021011, each of which is hereby incorporated by reference in its entirety.

[0082] Some embodiments of the systems and methods of the present disclosure simulate a system of n = 100 transmons that evolve under (30) and include the first three energy levels. The qubits are on a square grid with b _jk , non-zero only for neighboring qubits. Qubits have a frequency of either w/2p = 3 GHz or w/2p = 3.1 GHz, and no two adjacent qubits have the same frequency. All qubits have anharmonicity a/2p = —330 MHz. In each iteration, the target gate is chosen randomly from SU(2) " The crosstalk phase lag parameter q _jk are sampled randomly from the interval [0, 2p ), and draw b _jk from a normal distribution centered at zero. There are two discrete periods of successive evolution, each taking time t _p/2 . It is necessary to pick pulse shapes. On the one hand, pulses that yield error rates near the decoherence limit for short gate times are desired. On the other hand, there are experimental realities, such as power-bandwidth constraints and the degree of calibration needed to implement complicated pulses accurately. Balancing these constraints, select Gaussian pulses with stdW ^(x ₎ = t _p/2 /4, and half-derivative DRAG corrections . See Motzoi et al. , 2009, ”Simple Pulses for Elimination of Leaking in Weakly Nonlinear Qubits,” Physical Review Letters, 103(11), pg. 110501; Gambetta et al., 2011, ”Superconducting Qubit with Purcell Protection and Tunable Coupling,” Physical Review Letters, 106(3), pg. 030502; Chen et al., 2016, ”Measuring and Suppressing Quantum State Leakage in a Superconducting Qubit,” Physical Review Letters, 116(2), 020501, each of which is hereby incorporated by reference in its entirety.

[0083] Figure 4 provides an average single-qubit process infidelity as a func- tion of t _p/2 . Diamonds denote the raw infidelity for a crosstalk-free system , where d _jk is the Kronecker delta). The squares are infidelities obtained using the crosstalk-free control scheme but with various strengths of drive crosstalk. The triangles are infidelities obtained with optimized control and the same drive crosstalk as the squares. In some embod- iments, the present disclosure optimizes control pulses with the method of Machnes et al., 2018, ”Gradient Optimization of Analytic Controls: The Route to High Accuracy Quantum Optimal Control,” arXiv preprint, arXiv: 1507.04261, print, which is hereby incorporated by reference in its entirety. Applying the protocol requires the selection of appropriate optimization parameters. Sticking to a simple control ansatz, the systems and methods described herein tune the overall magnitude of the resonant W ⁽ _x) quadrature, off-resonant W ⁽ _y) quadrature and the carrier signal phase F, for a total of 7n parameters. Accordingly, approximately two orders of magnitude improvement in the infidelity is observed with the crosstalk minimization systems and methods of the present disclosure.

[0084] In some embodiments, such as real experimental devices, decoherence signihcantly reduces the average error rates. Moreover, decoherence errors grow with time, whereas control errors typically decrease. These contrasting effects imply that there is an optimal gate time that minimizes their combined errors. Thus, the prsent disclosure repeats the simulation by implementing with decoherence added to the model. Table I presents data showing the potential benefit of the systems and methods described herein.

[0085] Two-qubit gate engineering. Now, consider continuing the simulations using the ideal system of fixed-frequency transmons and the parameter values described supra. A goal is to implement parallel cross-resonance gates, which are equivalent to CNOTs up to single-qubit operations. See Rigetti et al., 2005, ”Protocol for Universal Gates in Optimally Biased Superconducting Qubits,” Physical Review Letters, 94(24), pg. 240502; Chow et al., 2012, ” Simple All- Microwave Entangling Gate for Fixed-Frequency Superconducting

TABLE I. Data highlighting a dramatic reduction in the average single-qubit process infidelity for a simulation with realistic decoherence on a square array of 100 qubits and various levels of crosstalk (std b _jk ). We model the same system considered in Fig. ??, but with T ₁ ~ W(40 ms, 5 ms), for each qubit, and T ₂ = 3T ₁ /2. Naturally, there is an optimal gate time that minimizes the combined incoherent (increasing) and coherent (approximately decreasing) effects. We optimize the controls for t _p/2 = 1, 2, . . . , 50 ns. The 'Original' column corresponds to the optimal t _p/2 without control tuneup. For all values of std is minimized at t _p/2 = 2 ns. On contemporary experimental devices, t _p/2 = 2 ns exceeds accesible bandwidths, so we also report r _aVg for t _p/2 = 5 ns.

Qubits,” Physical Review Letters, 107(1), pg. 080502; Sheldon, S., 2016, ” Procedure for Systematically Tuning Up Cross-talk in the Cross-Resonance Gate,” Physical Review A, 93(6), pg. 060302; Allen et al., 2019, ”Minimal Time Robust Two Qubit Gates in Circuit QED,” arXiv preprint, arXiv: 1902.08056, each of which is hereby incorporated by reference in its entirety. Constant capacitive coupling provides a mechanism for implementing entangling operations. Assuming equal coupling between all neighboring qubits in the system, the corresponding interaction Hamiltonian is where (j, k) = 1 denotes a sum over all adjacent qubit pairs. The entire system evolves under H .

[0086] An aspect of the cross resonance effect is that if the qubits are defined in a dressed basis, local microwave drive fields drive both single and two-qubit gates. For two ideal coupled qubits, in the dressed basis, a drive applied to qubit 1 at the frequency of qubit 2 yields the effective Hamiltonian where D = w ₁ — w ₂ is the difference of qubit frequencies and we made an RWA. Although the systems and methods provided herein can decouple the direct qubit coupling, higher-levels of the transmon lead to additional terms in the effective Hamiltonian. See Magesan et al., 2018. In some embodiments, the Z ₁ X ₂ term is used to generate a maximally entangling gate.

[0087] Again, a system of n = 100 transmons on a grid is simulated with the first three energy levels of each transmon included. In some embodiments, adjacent qubits are grouped into pairs and tries to implement 50 simultaneous maximally entangling gates using the CR effect. Accordingly, qubits have 5 distinct frequencies (3.0, 3.1, . . . , 3.4 GHz) to ensure each CR pair is addressable. Further, the frequencies are set so that no two neighbors of one qubit have the same frequency. The target CNOT equivalent is determined using Cartan's KAK decomposition and is invariant to local operations. See Tucci, R., ”An Introduction to Cartan's KAK Decomposition for QC Programmers,” Quantum Physics, arXiv:0507171, which is hereby incorporated by reference in its entirety. The qubit coupling strength is J 2p = 3.8 MHz. The present disclosure realizes qubit control with the same drives as above but with variable drive detuning and phase offset. Moreover, the present disclosure independently parameterizes the resonant and off-resonant control envelopes with the first three Hanning window functions

There are a total of 8 n parameters that determine the n drive fields.

[0088] Referring to Figure 5, a graphical chart is provided that illustrates the average two- qubit process infidelity of each entangling gate as a function of the gate duration t _CR for a square array of transmons implementing 50 simultaneous maximally engtangling gates via CR interactions in which each two-qubit pair approximates a CNOT-gate up to local operations. The present disclosure computes all points with optimized pulse parameters but under different system models. See Machnes et al., 2018. Diamonds denote the infidelity obtained using a drive-crosstalk-free model and no undesirable J coupling. The squares are infidelities calculated using the crosstalk-free optimal control but with added drive crosstalk (cr = 0.1) and nonlocal coupling. The triangles are infidelities obtained with controls tuned up under the crosstalk model. The nonlocal crosstalk effects are approximated with d = 1. The deviation caused by including additional neighbors is unresolvable on the plot.

[0089] Accordingly, the systems and methods of the present disclosure efficiently model and minimize crosstalk that occurs during qubit idling and gates. Compared to other quantum control methods such as dynamical decoupling, which attempts to echo out undesirable interactions, the present disclosure change parameters so the effects do not appear in the first place. See Viola et al. , 1999, ” Dynamical Decoupling of Open Quantum Systems,” Physical Review Letters, 82(12), pg. 2417, which is hereby incorporated by reference in its entirety. Results provided herein show how to mitigate such effects on transmons using a fast control tuneup procedure on a digital computer. As such, the present disclosure aids in understanding the role of crosstalk on NISQ devices and validate improved pulse shapes. Furthermore, the present disclosure can extend our simulation capabilities to other platforms such as trapped ions and apply our methods to improve the performance of experimental platforms.

[0090] Figure 6 is a schematic diagram showing aspects of an example system 100 that includes an error estimating system (digital computer system) 106, a quantum computer system 110 and an interface 108. The quantum computer system 110 includes a quantum information processor 114. The interface 108 provides communication between the error estimating system 106 and the quantum computer system 110. The system 100 may include additional or different features. Although not illustrated in Figure 6, in some embodiments the error estimating system controls a plurlity of fields that act upon the qudits of the quantum computer system 110.

[0091] The example error estimating system 106 has computational resources (e.g., hard- ware, software, firmware) that are used to produce executable instructions (e.g., machine code) for the quantum computer system 110. In some implementations, the error estimating system 106 can be implemented as a classical digital computer system, such as, for example, a laptop computer, a desktop computer, a computer workstation, a server, a server cluster, a mainframe, or another type of classical computer system. As shown in FIG. 6, the example error estimating system 106 includes a memory 102 and one or more digital processors 104. The error estimating system 106 may include other components, and may be configured to operate in another manner. [0092] In some instances, the memory 102 of the error estimating system 106 stores quantum-logic source code to be executed by the quantum information processor 114. For instance, the quantum-logic source code may be stored as quantum-logic gate sequences, quantum-logic circuits, quantum algorithms or other representations of quantum-logic. In some instances, the processor 104 of the error estimating system 106 converts the quantum- logic source code into machine code that can be executed by the quantum information processor 114. For instance, the machine code may be stored as a pulse sequence (e.g., radio frequency, microwave frequency, optical frequency), a pulse program, or another type of control sequence conhgured for execution by the quantum information processor 114. The machine code can be stored on the memory 102 of the error estimating system 106, transferred to the quantum computer system 110 via the interface 108, or handled in another manner.

[0093] The example memory 102 can include, for example, random access memory (RAM), a storage device (e.g., a writable read-only memory (ROM) or others), a hard disk, or another type of storage medium. The example memory 102 can store instructions (e.g., computer code, a computer program, etc.) associated with an operating system, computer applications and other resources. The memory 102 can also store application data and data objects that can be interpreted by one or more applications or virtual machines running on the error estimating system 106. In some embodiments, the memory 102 stores computer- readable instructions for software applications, scripts, programs, functions, executables or other modules that are executed by the processor 104. For example, the computer-readable instructions may be conhgured to perform one or more of the methods described in the present disclosure.

[0094] The example processor 104 can execute instructions, for example, to generate output data based on data inputs. For example, the processor 104 can run computer programs by executing the software, scripts, programs, functions, executables, or other modules stored in the memory 102. In some embodiments, the processor 104 may perform one or more of the methods described in the present disclosure. The example processor 104 shown in FIG. 6 can include one or more chips or chipsets that include analog circuitry, digital circuitry or a combination thereof. In some embodiments, the processor 104 includes multiple processor devices such as, for example, one or more main processors and one or more co-processors. [0095] In some aspects of operation, the compiler system 106 obtains quantum-logic source code (e.g., a quantum- logic gate sequence, a quantum-logic circuit, a quantum algorithm, etc.) and provides the quantum-logic source code to the quantum computer system 110 for execution by the quantum information processor 114. In some embodiments, the quantum- logic source code can be modihed to reduce the effects of noise in the quantum information processor 114. The error estimating system 106 may compile portions (e.g., one or more clock cycles) of the quantum-logic source code “on-the-fly” while the quantum information processor 114 executes earlier- compiled portions (e.g., earlier clock cycles). For instance, each cycle of the modihed quantum-logic source code may be provided to the quantum computer system 110 individually.

[0096] The example quantum computer system 110 can be a quantum computer, a quantum sensor, a quantum system, a quantum repeater or another type of device that operates using a quantum information processor. For example, the quantum computer system 110 may be a universal quantum computer that operates in a fault-tolerant regime and can be programmed and reprogrammed to perform computations. As another example, the quantum computer system 110 may be a quantum repeater that performs quantum error correction in an optical network used for quantum key distribution scheme. As another example, the quantum computer system 110 can be a quantum sensor that can measure physical properties of an external system based on quantum interactions or phenomena. [0097] The example quantum information processor 114 comprises any suitable number of qudits (e.g., quantum bits). For example, the quantum information processor 114 can comprise n qudits in a 2n- dimensional space. The quantum information processor 114 can perform quantum computing operations that manipulate the quantum information. For ex- ample, the quantum information processor 114 may coherently control the quantum system and preserve the relative phases of the qudits. If the quantum information processor 114 is a universal quantum computer, in some embodiment it can generally create any coherent state in a Hilbert space. In some embodiments the quantum information processor 114 can be conhgured to measure the state of the quantum system in a suitable basis. For example, the quantum information processor 114 may be conhgured to measure one or more of the qudits in a computational basis.

[0098] The quantum information processor 114 may be subject to noise that can create errors. The noise can include, for example, coherent noise processes, incoherent noise pro cesses, or decoherent noise processes. The noise can comprise local and nonlocal crosstalk. In some embodiments, the quantum information processor 114 can perform a quantum error correction process that effectively counteracts the noise.

[0099] The example interface 108 can include all or part of a connector, a data communi- cation network or another type of communication link. For example, the interface 108 can include one or more wired or wireless connections, one or more wired or wireless networks or other communication channels. In some examples, the interface 108 includes a Local Area Network (LAN), a Wide Area Network (WAN), a private network, a Virtual Private Network (VPN), a public network (such as the Internet), a peer-to-peer network, a cellular network, a Wi-Fi network, a Personal Area Network (PAN) or another type of data com- munication network. In some examples, the interface 108 includes a wired communication interface (e.g., USB, Ethernet, coaxial, optical) that can be connected directly or indirectly between the compiler system 106 and the quantum device 110.

[00100] The foregoing description included example systems, methods, techniques, instruc- tion sequences, and computing machine program products that embody illustrative imple- mentations. For purposes of explanation, numerous specific details were set forth in order to provide an understanding of various implementations of the inventive subject matter. It will be evident, however, to those skilled in the art that implementations of the inventive subject matter may be practiced without these specific details. In general, well-known instruction instances, protocols, structures and techniques have not been shown in detail.

[00101] The foregoing description, for purpose of explanation, has been described with reference to specific implementations. However, the illustrative discussions above are not in- tended to be exhaustive or to limit the implementations to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The implementa- tions were chosen and described in order to best explain the principles and their practical applications, to thereby enable others skilled in the art to best utilize the implementations and various implementations with various modifications as are suited to the particular use contemplated.

[00102] All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication or patent or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes.