betaFeedback

WorksPhysics and Astronomy

SPAM-robust learnability of quantum dynamics

Abstract

Accurate characterization of quantum dynamics is essential for the calibration and control of quantum devices. Unknown state preparation and measurement (SPAM) errors, however, introduce gauge ambiguities that limit what can be learned. A central question is therefore which properties are learnable despite SPAM errors, and whether physical structure in the dynamics can restrict these ambiguities and expand the set of learnable properties. In this work, we address both questions within a unified framework for quantum channels and continuous-time evolution. First, for general quantum channels and Lindbladian dynamics, we use the cycle structure of a generalized pattern-transfer graph to fully characterize all learnable parameter combinations and show achievability by constructing explicit SPAM-robust learning protocols, establishing a baseline where continuous gauge freedom prevents many individual parameters from being identified. Second, we show that coherent structure drastically restricts this gauge ambiguity, reducing continuous gauge freedom to discrete or even trivial choices. This restriction manifests across both discrete- and continuous-time settings, including purely Hamiltonian evolution, noisy quantum channels close to a target unitary, and Lindbladian dynamics close to a target Hamiltonian. In particular, when the dynamics preserves or remains close to a coherent Hamiltonian form, all coherent coefficients and broad classes of dissipative error parameters become individually learnable. These results establish a systematic route to exploiting physical structure for SPAM-robust characterization, providing a foundation for the reliable calibration and control of quantum devices.

Introduction

Quantum technologies promise advances in computation [1–3], secure communication [4], sensing [5, 6], and simulation [7, 8]. Realizing these applications relies on the execution of highly accurate quantum operations. Because of imperfect control and environmental coupling, noise benchmarking and characterization are essential for diagnosing errors, calibrating devices, and improving overall performance [9–13].

However, unknown state preparation and measurement (SPAM) errors pose a fundamental obstacle to rigorous characterization. Changes in the description of the dynamics can be fully compensated by inverse changes in the preparation and measurement operations, leaving all accessible statistics mathematically unchanged [14]. This gauge freedom strictly limits what can be learned about a quantum system, even given unlimited experimental data. The central questions in benchmarking are therefore: which properties remain intrinsically learnable, and what practical and physical structures can restrict this gauge freedom to make more information accessible?

These questions arise across two complementary paradigms: digital quantum processors modeled by quantum channels, i.e., completely positive trace-preserving (CPTP) maps, and analog simulators described by continuous-time Hamiltonians or Lindbladians. In the digital setting, the learnability of Pauli noise on Clifford gates has been successfully characterized using the cycle space of a pattern transfer graph [15], with recent extensions to mid-circuit measurements [16] and structured gate-set Pauli noise [17]. However, universal computation requires non-Clifford operations whose errors generally lack Pauli form. Similarly, in the continuous-time setting, while Hamiltonian-learning protocols can tolerate SPAM errors below a constant threshold [18] and concurrent work has classified the individually learnable coefficients of local Lindbladians [19], a complete picture is still missing. Specifically, combinations of parameters can remain learnable even when their constituent coefficients are gauge-ambiguous. A unified theory must therefore encompass arbitrary noisy operations and characterize these learnable combinations in both digital and continuous settings.

In this work, we provide a general characterization of SPAM-robust learnability for both arbitrary quantum channels and Lindbladian evolutions, assuming access to noiseless single-qubit Clifford control. We completely identify the learnable multiplicative combinations of nonzero Pauli transfer matrix (PTM) elements. We show that these combinations correspond to cycles of a generalized pattern transfer graph, along which all gauge factors exactly cancel. Furthermore, we construct explicit, SPAM-robust protocols based on Pauli character projection [20] to extract these quantities, establishing both what can be learned and how to learn it.

While individual parameters in general quantum dynamics often retain continuous gauge ambiguities, we show that coherent structure restricts the allowed gauge transformations and makes additional parameters learnable. This constraint is highly relevant in practice, as quantum gates and many analog simulators are explicitly designed to implement unitary evolution. For Hamiltonian dynamics, we first explain how the reversibility of unitary evolution restricts the gauge ambiguity in the Hamiltonian coefficients to discrete sign flips. We then show that every coefficient magnitude, as well as every odd-weight coefficient, is learnable. Furthermore, within any connected group of overlapping Hamiltonian terms, the even-weight coefficients retain at most a single common sign ambiguity, making all their relative signs learnable. Finally, we construct SPAM-robust protocols that estimate these coefficient magnitudes with Heisenberg scaling while recovering all accessible sign information.

Finally, we demonstrate that coherent structure also enhances learnability when characterizing noisy dynamics near a known target operation. Because gauge-equivalent descriptions must remain close to the same target, the ambiguity in the leading-order errors is strictly constrained. For quantum channels near a known unitary, we provide a graph-theoretic characterization of first-order learnability, complementing existing analyses of small gate-set errors [21] and first-order learnable error rates of Pauli channels [15]. For Lindbladian errors around an ideal Hamiltonian whose pattern transfer graph connects all nonzero supports, we prove that every coherent error coefficient and broad classes of dissipative coefficients become individually learnable to first order. Consequently, prior knowledge of the target dynamics restricts the admissible gauges and unlocks additional physical error parameters for device calibration.

The remainder of this paper is organized as follows. Section II summarizes our main results, and Secs. III and IV review related work and introduce the preliminaries. Section V formulates gauge freedom and learnability, followed by the analysis of general quantum channels in Sec. VI and Hamiltonian and Lindbladian evolution in Secs. VII and VIII. Section IX develops the first-order theory, and Sec. X presents case studies. We conclude in Sec. XI with a discussion of open questions and future directions.

Summary of results

We develop a unified framework for SPAM-robust learnability of general quantum channels and continuous-time dynamics generated by time-independent Hamiltonians and Lindbladians. Throughout, we assume noiseless single-qubit Clifford control and fixed SPAM errors dependent on the preparation and measurement settings. We characterize learnability by identifying the gauge freedom allowed by each physical model and determining the parameter functions invariant under these transformations. These gauge-invariant functions are precisely the SPAM-robust learnable quantities. Figure 1 summarizes the main results.

Overview of SPAM-robust learnability results for quantum channels and continuous-time dynamics

Figure 1. Overview of SPAM-robust learnability results for quantum channels and continuous-time dynamics. First-order results characterize learnability for weak perturbations of a known ideal unitary or Hamiltonian.

Gauge freedom arises because an experiment probes the evolution together with imperfect state preparation and measurement. An invertible map TT, which we call a gauge, acts simultaneously on the state, measurement, and every operation in the circuit. Using the Pauli-Liouville representation, writing the circuit as a sequence of operations A1,…,Am\mathcal{A}_1,\ldots,\mathcal{A}_m, this action leaves its outcome probability unchanged:

p=⟨⟨M∣Am⋯A1∣ρ⟩⟩=(⟨⟨M∣T−1)(TAmT−1)⋯(TA1T−1)(T∣ρ⟩⟩).(1)p = \langle\langle M|\mathcal{A}_m\cdots\mathcal{A}_1|\rho\rangle\rangle= \left(\langle\langle M|T^{-1}\right)\left(T\mathcal{A}_mT^{-1}\right)\cdots\left(T\mathcal{A}_1T^{-1}\right)\left(T|\rho\rangle\rangle\right). \tag*{(1)}

The factors T−1TT^{-1}T cancel between successive operations, so a change in the description of the evolution can be compensated by corresponding changes in the unknown preparation and measurement. A gauge is valid for a given model only when its action preserves the physical and structural assumptions defining that model. Learnability therefore depends both on the invariants of a given gauge freedom and on the constraints that determine which gauges are admissible. Keeping the control and SPAM assumptions fixed allows us to isolate how changes in physical structure alter the accessible information. Section V develops the notions of gauge freedom and learnability in detail.

General quantum channel. We begin with a general quantum channel to establish what remains learnable without imposing additional structure on the unknown dynamics. For a quantum channel G\mathcal{G}, the principal restriction on the gauge freedom comes from the noiseless single-qubit Clifford controls. Any valid gauge must leave these trusted operations unchanged, and Lemma 1 shows that it must therefore be diagonal in the Pauli basis,

T=∑Pλpt⁡(P)∣P⟩ ⁣⟩⟨ ⁣⟨P∣,(2)T = \sum_{P} \lambda_{\operatorname{pt}(P)} \lvert P\rangle\!\rangle\langle\!\langle P\rvert, \tag*{(2)}

where the pattern pt⁡(P)∈{0,1}n\operatorname{pt}(P) \in\{0,1\}^{n} is the nn-bit string indicating the support of PP: its qq-th bit is 1 if PP acts nontrivially on qubit qq and is 0 otherwise. Thus, all Pauli operators with the same support pattern are rescaled by the same nonzero real number. This is the common starting point for our analysis. Since noiseless single-qubit Clifford control is assumed throughout this work, the same restriction applies in every setting considered below.

Writing the map in the Pauli basis, let Gij=2−nTr⁡[PiG(Pj)]G_{ij}=2^{-n}\operatorname{Tr}[P_i\mathcal{G}(P_j)] denote its Pauli transfer matrix elements. To build intuition for the resulting invariants, we associate a pattern transfer graph with the channel, as illustrated in Figure 2. Its vertices are the Pauli patterns, and every nonzero Pauli transfer matrix element GijG_{ij} gives a directed edge from the input pattern pt⁡(Pj)\operatorname{pt}(P_j) to the output pattern pt⁡(Pi)\operatorname{pt}(P_i). Distinct matrix elements are retained as distinct edges. Under the gauge in Eq. (2), the weight of this edge transforms as

Pattern transfer graph and a cycle invariant

Figure 2. Pattern transfer graph of an illustrative noisy two-qubit ZZZZ rotation. (a) displays one arrow for each pair of patterns connected by nonzero PTM elements, with parallel edges combined as one edge for visual clarity. (b) highlights four particular elements and fixes the counterclockwise traversal shown at the center. Along this closed route, weights traversed with their arrows are multiplied and a weight traversed against its arrow is inverted, giving the invariant GZI,IIGXZ,YIGIX,ZY/GIZ,IIG_{ZI,II}G_{XZ,YI}G_{IX,ZY}/G_{IZ,II}.

(TGT−1)ij=λpt⁡(Pi)λpt⁡(Pj)Gij.(3)(TGT^{-1})_{ij} = \frac{\lambda_{\operatorname{pt}(P_i)}}{\lambda_{\operatorname{pt}(P_j)}}G_{ij}. \tag*{(3)}

To see how gauge-invariant quantities arise, choose a closed path through the graph and a direction in which to traverse it, as illustrated in Figure 2(b). Multiply the weight of an edge when the route follows its direction and the inverse weight when it runs against the direction. We call the product obtained in this way a cycle quantity. Every gauge factor cancels in such a product, so every cycle quantity is learnable. It turns out that these cycle quantities characterize all learnable multiplicative functions, as summarized in the following theorem.

Theorem 1 (informal). For a general quantum channel, a multiplicative function of nonzero Pauli transfer matrix elements is learnable if and only if it can be expressed as a product of powers of cycle quantities.

Any multiplicative function outside this class is not gauge invariant and is therefore unlearnable in the presence of SPAM errors. Equivalently, after representing multiplicative combinations additively through logarithms, the formal statement of Theorem 1 identifies the learnable combinations with the cycle space of the pattern transfer graph. This gives a complete characterization of learnable multiplicative combinations even when individual matrix elements remain ambiguous.

Channel learning protocols. To extract the information identified by the gauge analysis, we use Pauli character projection to select a prescribed path through the Pauli transfer matrix. For Pauli operators QQ and RR, let χ(Q,R)\chi(Q,R) be +1+1 when they commute and −1-1 when they anticommute, and let R\mathcal{R} be the unitary channel corresponding to the Pauli gate RR. Character orthogonality gives

ER∼Pn[χ(Q,R)R]=∣Q⟩ ⁣⟩⟨ ⁣⟨Q∣.(4)\mathbb{E}_{R\sim\mathcal{P}_n}\left[\chi(Q,R)\mathcal{R}\right]=\lvert Q\rangle\!\rangle\langle\!\langle Q\rvert. \tag*{(4)}

Thus, sampling Pauli gates and weighting the observed outcomes by their characters cancels every Pauli component except the selected one. Applying this projection successively isolates the product of the transfer-matrix elements along a chosen path γ\gamma in the pattern transfer graph. Noiseless local Clifford gates connect consecutive Pauli operators whenever they have the same pattern. Denoting the weight of an edge by geg_e, the resulting signal has the simple form

Wγ=A∏e∈γge.(5)W_\gamma=A\prod_{e\in\gamma}g_e. \tag*{(5)}

where all the SPAM error factors are absorbed into AA. For an oriented cycle, repeating the projected cycle to depth mm gives

F(m)=Arm,r=∏e∈cge.(6)F(m)=Ar^m,\qquad r=\prod_{e\in c}g_e. \tag*{(6)}

Since the SPAM errors enter only through the depth-independent amplitude AA, fitting the rate rr yields the cycle product. The complete protocol is presented in Box 1. For an unoriented cycle, we instead combine projected path signals as a ratio, using common endpoint settings so that their SPAM factors cancel. Box 2 presents the corresponding construction. Thus, Pauli character projection provides an explicit SPAM-robust protocol for every learnable cycle quantity.

Hamiltonian evolution. Preserving coherent structure imposes consistency conditions absent from a general noisy channel. We now show how these conditions reduce the ambiguity in nonidentity Pauli coefficients to discrete sign choices. Consider a time-independent Hamiltonian H=∑acaPaH=\sum_a c_aP_a, where aa runs over the nonzero terms which specify the interaction structure. Let Ut(H)U_t^{(H)} denote the unitary channel obtained by evolving under HH for time tt. We call HH and H′=∑aca′PaH'=\sum_a c'_aP_a gauge equivalent if there exists a valid gauge TT such that

Ut(H′)=TUt(H)T−1,(7)U_t^{(H')}=TU_t^{(H)}T^{-1}, \tag*{(7)}

holds for all t≥0t\ge0. A Hamiltonian coefficient is learnable only if it remains invariant in all gauge equivalent Hamiltonians. Recall from Eq. (2) that TT assigns one gauge factor λu\lambda_u to each Pauli pattern uu. In the meantime, one single Hamiltonian coefficient induces several edges in the pattern transfer graph with such weights. Gauge transformation can rescale these weights, but they must be identical to preserve Hamiltonian form. This therefore imposes stronger gauge constraints. Lemma 3 expresses this observation explicitly. Suppose TT is a valid gauge preserving Hamiltonian form, then for every nonzero Hamiltonian term caPac_aP_a, there exists μa\mu_a such that

λpt⁡(PaPb)λpt⁡(Pb)=μa,(8)\frac{\lambda_{\operatorname{pt}(P_aP_b)}}{\lambda_{\operatorname{pt}(P_b)}}=\mu_a, \tag*{(8)}

for every PbP_b satisfying {Pa,Pb}=0\{P_a,P_b\}=0. Here μa\mu_a may depend on the Hamiltonian term PaP_a, but not on the choice of PbP_b. Replacing PbP_b by PaPbP_aP_b reverses the ratio, giving μa=μa−1\mu_a=\mu_a^{-1} and hence μa∈{+1,−1}\mu_a\in\{+1,-1\}. These μa\mu_a factors completely characterize the ambiguity in the Hamiltonian coefficients induced by gauge transformation of TT, as summarized below.

Theorem 3 (informal). Gauge-equivalent Hamiltonians HH and H′H' can differ only in the signs of their Pauli coefficients:

ca′=μaca,μa∈{+1,−1}.(9)c'_a=\mu_a c_a,\qquad\mu_a\in\{+1,-1\}. \tag*{(9)}

The allowed sign changes can be represented by assigning signs λu∈{+1,−1}\lambda_u\in\{+1,-1\} to the Pauli patterns, subject to Eq. (8). Thus, every coefficient has a learnable magnitude, while the remaining sign choices are constrained by the Hamiltonian terms.

The remaining sign ambiguity depends on the parity of the Pauli weight. Every odd-weight term generates a nonzero transfer between two Pauli operators with the same pattern, forming a self-loop in the pattern transfer graph. Therefore all odd weight coefficients are guaranteed to be learnable.

Corollary 1 (informal). Every odd-weight Hamiltonian coefficient is individually learnable: no valid gauge can change either its magnitude or its sign.

For even-weight terms, the remaining sign choices are constrained by how the Hamiltonian terms overlap. Two terms are connected if they act on a common qubit or are linked by a chain of nonzero Hamiltonian terms with overlapping supports. Such a chain may include both odd- and even-weight terms. It turns out that if two even-weight terms are connected, they must share the same sign flip.

Corollary 2 (informal). Every even-weight coefficient has a learnable magnitude. Within each connected group of Hamiltonian terms, the even-weight coefficients can only change by a simultaneous sign flip,

ca′=σC(a)ca,σC(a)∈{+1,−1},(10)c'_a=\sigma_{C(a)}c_a,\qquad\sigma_{C(a)}\in\{+1,-1\}, \tag*{(10)}

where C(a)C(a) denotes the connected group containing PaP_a. These sign flips may be chosen independently for different disconnected groups.

Consequently, the relative signs, pairwise products, and ratios of even-weight coefficients within the same connected group are learnable. In particular, when all nonzero Hamiltonian terms are connected, the Hamiltonian is determined up to a single collective sign flip of its even-weight terms. Thus, coherent structure leaves only a discrete obstruction to identifying these coefficients, without strengthening the SPAM assumptions.

Hamiltonian learning protocols. To recover the information made learnable by coherent structure, we combine Pauli character projection with fast Pauli control. As in the channel protocol, character projection isolates a closed path whose repeated signal is insensitive to SPAM errors. Hamiltonian reshaping now allows us to select cycles associated with particular coefficients. For any target coefficient cac_a, we first isolate its Pauli term by averaging HH over Pauli gates that commute with PaP_a,

ER∈Cent⁡(Pa)[RHR†]=caPa.(11)\mathbb{E}_{R\in\operatorname{Cent}(P_a)}\left[RHR^\dagger\right]=c_aP_a. \tag*{(11)}

This Hamiltonian reshaping can be implemented with fast Pauli control, and reduces the relevant evolution to that generated by caPac_aP_a. Choose any Pauli PbP_b that anticommutes with PaP_a and let it evolve under the reshaped Hamiltonian for time τ\tau, denote Q=−iPaPbQ=-iP_aP_b, one obtains,

Uτ(caPa)∣Pb⟩=cos⁡(2caτ)∣Pb⟩+sin⁡(2caτ)∣Q⟩.(12)U_{\tau}^{(c_aP_a)}|P_b\rangle=\cos(2c_a\tau)|P_b\rangle+\sin(2c_a\tau)|Q\rangle. \tag*{(12)}

Applying character projector in Eq. (4) to selects PbP_b before and after the evolution forms a cycle with weight cos⁡(2caτ)\cos(2c_a\tau). Repeating this cycle to depth mm gives

Fcos⁡(m)=Acos⁡cos⁡m(2caτ).(13)F_{\cos}(m)=A_{\cos}\cos^m(2c_a\tau). \tag*{(13)}

The unknown SPAM errors enter only through the depth-independent amplitude Acos⁡A_{\cos}, so fitting the signal across depths extracts cos⁡(2caτ)\cos(2c_a\tau). This construction applies to both odd- and even-weight terms and provides access to ∣ca∣|c_a| by varying the evolution time.

For odd-weight PaP_a, we can also extract the sine branch in Eq. (12) by choosing PbP_b to have a different nonidentity Pauli on every qubit in the support of PaP_a and identity elsewhere. Then PbP_b anticommutes with PaP_a, and QQ has the same pattern as PbP_b. Selecting the QQ branch after the evolution and applying a noiseless local Clifford that maps QQ back to PbP_b closes a cycle with weight sin⁡(2caτ)\sin(2c_a\tau). Repeating this cycle gives

Fsin⁡(m)=Asin⁡sin⁡m(2caτ),(14)F_{\sin}(m)=A_{\sin}\sin^{m}(2c_a\tau), \tag*{(14)}

from which the sine is extracted in the same way. Together, the sine and cosine estimates at increasing evolution times allow robust phase estimation (Sec. IV D) to recover the odd-weight coefficient cac_a with Heisenberg scaling. Box 3 presents the full protocol.

To determine the relative signs of connected even-weight coefficients, the required cycles may involve several Hamiltonian terms. Their construction and the corresponding protocols are given in Sec. VII C, in particular Box 4. While these protocols recover the learnable coefficient information one term at a time, Sec. VII D further shows how the learning can be parallelized when the Hamiltonian has additional sparse and kk-local structure.

General Lindbladian evolution. General Lindbladian dynamics provides a contrasting case: a valid gauge must preserve Lindblad form, but can mix the coherent and dissipative contributions. This allows us to distinguish the restrictions imposed by coherent structure from those imposed by continuous-time evolution alone. In the Pauli representation, a time-independent Lindbladian takes the standard form

References

  1. [21]J. G. D. Mendoza, C. Ostrove, T. Proctor, K. Young, E. Nielsen, and R. Blume-Kohout, A gauge-invariant theory of small markovian errors in quantum gate sets (2026), arXiv:2608.14891 [quant-ph].arxiv.org/abs/2608.14891
  2. [22]S. T. Merkel, J. M. Gambetta, J. A. Smolin, S. Poletto, A. D. Córcoles, B. R. Johnson, C. A. Ryan, and M. Steffen, Self-consistent quantum process tomography, Phys. Rev. A 87, 062119 (2013).DOI
  3. [23]C. Stark, Self-consistent tomography of the state-measurement Gram matrix, Phys. Rev. A 89, 052109 (2014).DOI
  4. [24]R. Blume-Kohout, J. K. Gamble, E. Nielsen, J. Mizrahi, J. D. Sterk, and P. Maunz, Robust, self-consistent, closed-form tomography of quantum logic gates on a trapped ion qubit (2013), arXiv:1310.4492 [quant-ph].
  5. [25]Y. Gu, R. Mishra, B.-G. Englert, and H. K. Ng, Randomized linear gate-set tomography, PRX Quantum 2, 030328 (2021).
  6. [26]K. Rudinger, G. J. Ribeill, L. C. Govia, M. Ware, E. Nielsen, K. Young, T. A. Ohki, R. Blume-Kohout, and T. Proctor, Characterizing midcircuit measurements on a superconducting qubit using gate set tomography, Phys. Rev. Appl. 17, 014014 (2022).DOI
  7. [27]R. Brieger, I. Roth, and M. Kliesch, Compressive gate set tomography, PRX Quantum 4, 010325 (2023).DOI
  8. [28]A. Carignan-Dugas, J. J. Wallman, and J. Emerson, Characterizing universal gate sets via dihedral benchmarking, Physical Review A 92, 060302 (2015).
  9. [29]A. W. Cross, E. Magesan, L. S. Bishop, J. A. Smolin, and J. M. Gambetta, Scalable randomised benchmarking of non-clifford gates, npj Quantum Information 2, 10.1038/npjqi.2016.12 (2016).DOI
  10. [30]E. Onorati, A. H. Werner, and J. Eisert, Randomized benchmarking for individual quantum gates, Phys. Rev. Lett. 123, 060501 (2019).DOI
  11. [31]J. Hines, M. Lu, R. K. Naik, A. Hashim, J.-L. Ville, B. Mitchell, J. M. Kriekebaum, D. I. Santiago, S. Seritan, E. Nielsen, R. Blume-Kohout, K. Young, I. Siddiqi, B. Whaley, and T. Proctor, Demonstrating scalable randomized benchmarking of universal gate sets, Phys. Rev. X 13, 041030 (2023).DOI
  12. [32]G. Liu, Z. Xie, Z. Xu, and X. Ma, Group twirling and noise tailoring for multiqubit controlled phase gates, Phys. Rev. Res. 6, 043221 (2024).DOI
  13. [33]H. Ye, G. Liu, and X. Ma, Benchmarking non-clifford gates using pauli twirling, Phys. Rev. A 113, 012626 (2026).DOI
  14. [34]D. Layden, B. Mitchell, and K. Siva, Theory of quantum error mitigation for non-Clifford gates, Quantum 10, 2003 (2026).arxiv.org/abs/2403.18793
  15. [35]S. Endo, S. C. Benjamin, and Y. Li, Practical quantum error mitigation for near-future applications, Physical Review X 8, 032027 (2018).DOI
  16. [36]E. H. Chen, S. Chen, L. E. Fischer, A. Eddins, L. C. Govia, B. Mitchell, A. He, Y. Kim, L. Jiang, and A. Seif, Disambiguating pauli noise in quantum computers, PRX Quantum 7, 033045 (2026).DOI
  17. [37]W. Yu, J. Sun, Z. Han, and X. Yuan, Robust and efficient hamiltonian learning, Quantum 7, 1045 (2023).DOI
  18. [38]D. S. França, L. A. Markovich, V. V. Dobrovitski, A. H. Werner, and J. Borregaard, Efficient and robust estimation of many-qubit hamiltonians (2022), arXiv:2205.09567 [quant-ph].arxiv.org/abs/2205.09567
  19. [39]H.-Y. Hu, M. Ma, W. Gong, Q. Ye, Y. Tong, S. T. Flammia, and S. F. Yelin, Ansatz-free hamiltonian learning with heisenberg-limited scaling, PRX Quantum 6, 040315 (2025).DOI
  20. [40]M. Ma, S. T. Flammia, J. Preskill, and Y. Tong, Learning many-body hamiltonians via compressed sensing, IEEE Transactions on Information Theory , 1–1 (2026).DOI
  21. [41]S. Brahmachari, S. Zhu, I. Maizan, and Y. Tong, Learning hamiltonians in the heisenberg limit with static single-qubit fields (2026), arXiv:2601.10380 [quant-ph].arxiv.org/abs/2601.10380
  22. [42]Y. Liu, J. R. Seddon, T. Kohler, E. Onorati, and T. S. Cubitt, Robust lindbladian estimation for quantum dynamics (2025), arXiv:2507.07912 [quant-ph].arxiv.org/abs/2507.07912
  23. [43]S. D. Sinha, Efficient and spam-robust ansatz-free lindbladian learning, arXiv preprint arXiv:2606.20706 (2026).DOI
  24. [44]P. Ivashkov, N. Romanov, W. Gong, A. Gu, H.-Y. Hu, and S. F. Yelin, Ansatz-free learning of lindbladian dynamics in situ (2026), arXiv:2603.05492 [quant-ph].arxiv.org/abs/2603.05492
  25. [45]S. Liu, Z. Xu, A. V. Gorshkov, X. Wu, Y.-X. Wang, and Z.-Y. Wei, Learning many-body hamiltonians using a local probe (2026), arXiv:2610.02157 [quant-ph].arxiv.org/abs/2610.02157
  26. [46]W. Gong, M. Ma, S. Chen, J. Cotler, and H.-Y. Huang, Out-of-control hamiltonian learning, arXiv preprint arXiv:2610.06709 (2026).arxiv.org/abs/2610.06709
  27. [47]S. Kimmel, G. H. Low, and T. J. Yoder, Robust calibration of a universal single-qubit gate set via robust phase estimation, Physical Review A 92, 062315 (2015), erratum: Phys. Rev. A 104, 069901 (2021), arXiv:1502.02677.DOI
  28. [48]D. Leibfried, B. DeMarco, V. Meyer, D. Lucas, M. Barrett, J. Britton, W. M. Itano, B. Jelenković, C. Langer, T. Rosenband, and D. J. Wineland, Experimental demonstration of a robust, high-fidelity geometric two ion-qubit phase gate, Nature 422, 412 (2003).
  29. [49]L. Isenhower, E. Urban, X. L. Zhang, A. T. Gill, T. Henage, T. A. Johnson, T. G. Walker, and M. Saffman, Demonstration of a neutral atom controlled-NOT quantum gate, Physical Review Letters 104, 010503 (2010).pubmed.ncbi.nlm.nih.gov/20366355
  30. [50]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, and R. J. Schoelkopf, Demonstration of two-qubit algorithms with a superconducting quantum processor, Nature 460, 240 (2009).DOI
  31. [51]Y. Sung, L. Ding, J. Braumüller, A. Vepsäläinen, B. Kannan, M. Kjaergaard, A. Greene, G. O. Samach, C. McNally, D. Kim, A. Melville, B. Niedzielski, M. E. Schwartz, J. L. Yoder, T. P. Orlando, S. Gustavsson, and W. D. Oliver, Realization of high-fidelity CZ and ZZ-free iSWAP gates with a tunable coupler, Physical Review X 11, 021058 (2021).
  32. [52]Z. Xu and G. Liu, Entanglement distillation based on hamiltonian dynamics (2026), arXiv:2603.10843 [quant-ph].arxiv.org/abs/2603.10843
  33. [53]Z. Liu, Z. Hao, and H.-Y. Hu, Predicting arbitrary state properties from single hamiltonian quench dynamics, Phys. Rev. Res. 6, 043118 (2024).arxiv.org/abs/2311.00695
  34. [54]P. Rungta, V. Bužek, C. M. Caves, M. Hillery, and G. J. Milburn, Universal state inversion and concurrence in arbitrary dimensions, Physical Review A 64, 042315 (2001).DOI
  35. [55]H. Rauhut, Compressive sensing and structured random matrices, in Theoretical Foundations and Numerical Methods for Sparse Recovery, Radon Series on Computational and Applied Mathematics, Vol. 9, edited by M. Fornasier (De Gruyter, 2010) pp. 1–92.DOI
  36. [56]E. J. Candès, J. K. Romberg, and T. Tao, Stable signal recovery from incomplete and inaccurate measurements, Communications on Pure and Applied Mathematics 59, 1207 (2006).
  37. [57]C. J. Ballance, T. P. Harty, N. M. Linke, M. A. Sepiol, and D. M. Lucas, High-fidelity quantum logic gates using trapped-ion hyperfine qubits, Physical Review Letters 117, 060504 (2016).DOI
  38. [58]L. Ding, M. Hays, Y. Sung, B. Kannam, J. An, A. Di Paolo, A. H. Karamlou, T. M. Hazard, K. Azar, D. K. Kim, B. M. Niedzielski, A. Melville, M. E. Schwartz, J. L. Yoder, T. P. Orlando, S. Gustavsson, J. A. Grover, K. S. Ranjani, and W. D. Oliver, High-fidelity, frequency-flexible two-qubit fluxonium gates with a transmon coupler, Physical Review X 13, 031035 (2023).DOI
  39. [59]E. van den Berg, Z. K. Minev, A. Kandala, and K. Temme, Probabilistic error cancellation with sparse Pauli–Lindblad models on noisy quantum processors, Nature Physics 19, 116 (2023).DOI
  40. [60]Y. Wu, Y. Wang, X. Qin, X. Rong, and J. Du, A programmable two-qubit solid-state quantum processor under ambient conditions, npj Quantum Information 5, 9 (2019).DOI
  41. [61]Z. Bao, S. Xu, Z. Song, K. Wang, L. Xiang, Z. Zhu, J. Chen, et al., Creating and controlling global Greenberger–Horne–Zeilinger entanglement on quantum processors, Nature Communications 15, 8823 (2024).DOI
  42. [62]D. Grinko, J. Gacon, G. Zoufal, and S. Woerner, Iterative quantum amplitude estimation, npj Quantum Information 7, 52 (2021), arXiv:1912.05559.DOI
  43. [63]S. Fukuzawa, C. Ho, S. Irani, and J. Zion, Modified iterative quantum amplitude estimation is asymptotically optimal, in 2023 Proceedings of the Symposium on Algorithm Engineering and Experiments (ALENEX) (2023) pp. 135–147, arXiv:2208.14612.DOI
  44. [64]A. Bonami, Étude des coefficients de Fourier des fonctions de L^p(G), Annales de l’Institut Fourier 20, 335 (1970).DOI
  45. [65]W. Beckner, Inequalities in Fourier analysis, Annals of Mathematics 102, 159 (1975).DOI

Paper details

vlnc:W_u2v83bft
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Generative artificial intelligence tools

View statement
Looking for related papers…
Contents