Elias Zapusek’s scientific contributions

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (6)


Provable advantages of kernel-based quantum learners and quantum preprocessing based on Grover's algorithm
  • Article

September 2024

·

5 Reads

·

4 Citations

Physical Review A

T. Muser

·

E. Zapusek

·

V. Belis

·

F. Reiter

There is an ongoing effort to find quantum speedups for learning problems. Recently [Y. Liu et al., Nat. Phys. 17, 1013 (2021)] proved an exponential speedup for quantum support vector machines by leveraging the speedup of Shor's algorithm. We expand upon this result and identify a speedup utilizing Grover's algorithm in the kernel of a support vector machine. To show the practicality of the kernel structure we apply it to a problem related to pattern matching, providing a practical yet provable advantage. Moreover, we show that combining quantum computation in a preprocessing step with classical methods for classification further improves classifier performance.


FIG. 4: (a) Standard dissipative cat-state stabilization explained using the nonlinear reservoir engineering method. Rabi frequencies and stabilized boson distribution of the jump operatorâoperatorˆoperatorâ 3 − α 3 with α = 3.5. (b) Relative entropy of a Poisson distribution, parameterized by α, with respect to its approximation using a CMP distribution given in Eq. (10). The Poisson distribution is stabilized usingâusingˆusingâ d − α d , where d represents the dimension of the coherent-state manifold. The low relative entropy suggests that the true and approximate distributions contain nearly identical amounts of information. This demonstrates that the proposed nonlinear reservoir engineering method effectively generalizes wellestablished reservoir engineering cases.
FIG. 5: (a) We evaluate the effective confinement rate κ conf as the exponential rate at which a dark state initially displaced (δx = 0.001) returns to the stabilized manifold. (b) Effective confinement rate with respect to the angle ϑ f + ϑ g between the functions˜ffunctions˜ functions˜f and˜gand˜ and˜g. One can increase this rate by increasing the ratio ϑ f /ϑ g which effectively increases the skewness of the stabilized boson distribution (the height is constant h * = 20). The theoretical curves have been obtained using Eq. (19) additionally corrected for the skewness. (c) The relationship between the height of the crossing point h * and κ conf is linear (ϑ f = ϑ g = π/3). In both (b) and (c), we keep the crossing point at k * = 20.
FIG. 6: Reducing the Mandel Q parameter improves the correction against dephasing. (b) The evolution of the fidelity of a dark state of L with itself when subject of the stabilization and incoherent dephasing processes for (0, 2) schemes (κ ϕ = 10 −2 κ eff ). This evolution is shown for different slopes s f = s g of the linearly decreasing and increasing processes˜fprocesses˜ processes˜f and˜gand˜ and˜g crossing at a fixed Fock state k * = 20 and height h * = 10 as depicted in (a). Changing these slopes changes the Mandel Q parameter (see Fig. 3). We observe an enhanced stabilization when Q → −1 which is explained by a lower variance of the stabilized distribution and thus a higher confinement rate κ conf (see Fig. 5). (c) Steady state infidelity as a function of the Mandel Q parameter for both (0, 2) and (2, 0) schemes.
FIG. 7: Correction capability of (1, 1) models. (a) The fidelity of the initial state |Ξ 0 ⟩ with itself over time under simultaneous stabilization and momentum errors modeled using jump operators √ κ diffˆpdiffˆ diffˆp with κ diff = 0.5κ eff . The stabilization is performed using linear˜flinear˜ linear˜f and˜gand˜ and˜g intersecting at k * = 10, h * = 10 and ϑ f + ϑ g = 2π/3 (see Fig. 3). The evolution is also shown for the (0, 2) model (dashed line). (b) The logical error rate for (1, 1) schemes as a function of the relative angle betweeñ f and˜gand˜ and˜g at the crossing point. For symmetric boson distributions, i.e. ϑ f = ϑ g , the rate stays constant. For asymmetric distributions, the rate can be improved by increasing their variance. However, when the relative angle is too large, the rate increases due to the mixing effect discussed in Sec. III A.
FIG. 10: NLRE method for superconducting circuit QED settings. (a) Schematic representation of cQED with a highquality LC oscillator (black) storing the cat-like state manifold and a nonlinear circuit (blue) coupled to a dissipative environment represented here by a lossy LC oscillator (yellow). The nonlinear elements is a DC-biased asymmetrically threaded superconducting quantum interference device (ATS). The toy model of NLRE is engineered using a two-tone flux drive. (b) Example of the (r, l) = (0, 4) scheme using a voltage-biased ATS circuit with parameters specified in the main text. (c) Wigner quasiprobability of the stabilized cat-like states |Ξ 0 ⟩ for the scheme in (b).

+1

Stabilization of cat-state manifolds using nonlinear reservoir engineering
  • Preprint
  • File available

July 2024

·

22 Reads

Ivan Rojkov

·

Matteo Simoni

·

Elias Zapusek

·

[...]

·

Jonathan Home

We introduce a novel reservoir engineering approach for stabilizing multi-component Schr\"odinger's cat manifolds. The fundamental principle of the method lies in the destructive interference at crossings of gain and loss Hamiltonian terms in the coupling of an oscillator to a zero-temperature auxiliary system, which are nonlinear with respect to the oscillator's energy. The nature of these gain and loss terms is found to determine the rotational symmetry, energy distributions, and degeneracy of the resulting stabilized manifolds. Considering these systems as bosonic error-correction codes, we analyze their properties with respect to a variety of errors, including both autonomous and passive error correction, where we find that our formalism gives straightforward insights into the nature of the correction. We give example implementations using the anharmonic laser-ion coupling of a trapped ion outside the Lamb-Dicke regime as well as nonlinear superconducting circuits. Beyond the dissipative stabilization of standard cat manifolds and novel rotation symmetric codes, we demonstrate that our formalism allows for the stabilization of bosonic codes linked to cat states through unitary transformations, such as quadrature-squeezed cats. Our work establishes a design approach for creating and utilizing codes using nonlinearity, providing access to novel quantum states and processes across a range of physical systems.

Download

Experimental Realization of Nonunitary Multiqubit Operations

January 2024

·

55 Reads

·

2 Citations

Physical Review Letters

We demonstrate a novel experimental tool set that enables irreversible multiqubit operations on a quantum platform. To exemplify our approach, we realize two elementary nonunitary operations: the or and nor gates. The electronic states of two trapped Ca40+ ions encode the logical information, and a cotrapped Sr88+ ion provides the irreversibility of the gate by a dissipation channel through sideband cooling. We measure 87% and 81% success rates for the or and nor gates, respectively. The presented methods are a stepping stone toward other nonunitary operations such as in quantum error correction and quantum machine learning.


FIG. 2. Overview of the gate mechanism. (a) Desired (left) and undesired (right) process for the OR gate. State |01 |0 m is off-resonantly driven to |f 1 |0 m with Rabi frequency Ω f . Due to the coupling ΩSB, |f 1 |0 m is hybridized with states |11 |1 m and |1f |0, resulting in a dressed state splitting. For ∆ = ΩSB/ √ 2, the carrier drive is on resonance with the dressed state, and is therefore excited. The gate action is completed by sympathetically cooling the motional mode. The dressed states |f 0 |0 m and |10 |1 m , accessible from |00 |0 m , are shifted by ±ΩSB/2, and are therefore not in resonance with the carrier drive. Excitation from |00 |0 m is thus suppressed. (b) Pulse sequence of OR and NOR gates. Transitions used in the experiment are indicated in Fig. 1(c). Ω f and Ωe act on the first ion, coupling |0 with |f and |1 with |e. ΩSB acts on both ions, coupling the red sideband of |1 and |f .
FIG. 3. (a) Experimental demonstration of state-dependent population transfer transfer with ∆ = ΩSB/2, shown for each possible initial state. The lines indicate simulated results, which include measured initial phonon number and heating rate as simulation parameters. (b) Demonstration of dissipation, after a maximal probe transfer from |00 (at a time marked by the gray dashed line in (a) ). We show the evolution of P f 0 and P10, and use sideband thermometry on the Strontium ion to infer the ground-state phonon occupation. The lines show simulated results, in which the dissipation rate Γ f is obtained through a least-squares fit between simulated and measured data.
FIG. 4. Measured population truth tables of the OR and NOR gates, with the intended output states marked with dashed lines. Values are in percent, and are determined from 50 experimental shots for each input setting. The OR and NOR gates have an average population fidelity of 87(5)% and 81(5)%.
FIG. S1. Level schemes of the error processes. (a) The error process intrinsic to the scheme. Even in absence of any harmonic oscillator excitations some population from |00 |0 m is off-resonantly excited. By increasing the sideband frequency ΩSB relative to the other couplings this error could be suppressed. (b) If the harmonic oscillator is in state n = 1 the energies of the excited states addressed from |00 |1 m are shifted by δ and the coupling between them is enhanced. This shifts the probe into resonance. (c) The desired process in presence of an harmonic oscillator excitation. The additional excitation enables coupling to the double excited state |f f |0 m and adjusts the frequencies to ∆ + δ. Furthermore, |01 |1 m forms dressed states with |0f |0 m . The lowest frequency dressed state resides at˜∆01at˜ at˜∆01.
Experimental realization of nonunitary multi-qubit operations

March 2023

·

119 Reads

We demonstrate a novel experimental toolset that enables irreversible multi-qubit operations on a quantum platform. To exemplify our approach, we realize two elementary nonunitary operations: the OR and NOR gates. The electronic states of two trapped 40^{40}Ca+^{+} ions encode the logical information, and a co-trapped 88^{88}Sr+^{+} ion provides the irreversibility of the gate by a dissipation channel through sideband cooling. We measure 87%87\% and 81%81\% success rates for the OR and NOR gates, respectively. The presented methods are a stepping stone towards other nonunitary operations such as in quantum error correction and quantum machine learning.


Nonunitary gate operations by dissipation engineering

November 2022

·

46 Reads

·

12 Citations

Irreversible logic is at odds with unitary quantum evolution. Emulating such operations by classical measurements can result in disturbances and high resource demands. To overcome these limitations, we propose protocols that harness dissipation to realize the nonunitary evolution required for irreversible gate operations. Using additional excited states subject to decay, we engineer effective decay processes that perform the desired gate operations on the smallest stable Hilbert space. These operate deterministically and in an autonomous fashion, without the need for measurements. We exemplify our approach considering several classical logic operations, such as the OR, NOR, and XOR gates. Towards experimental realization, we discuss a possible implementation in quantum dots. Our study shows that irreversible logic operations can be efficiently performed on realistic quantum systems and that dissipation engineering is an essential tool for obtaining nonunitary evolutions. The proposed operations expand the quantum engineers' toolbox and have promising applications in NISQ algorithms and quantum machine learning.


Nonunitary Gate Operations by Dissipation Engineering

January 2022

·

22 Reads

Irreversible logic is at odds with unitary quantum evolution. Emulating such operations by classical measurements can result in disturbances and high resource demands. To overcome these limitations, we propose protocols that harness dissipation to realize the nonunitary evolution required for irreversible gate operations. Using additional excited states subject to decay, we engineer effective decay processes that perform the desired gate operations on the smallest stable Hilbert space. These operate deterministically and in an autonomous fashion, without the need for measurements. We exemplify our approach considering several classical logic operations, such as the OR, NOR, and XOR gates. Towards experimental realization, we discuss a possible implementation in quantum dots. Our study shows that irreversible logic operations can be efficiently performed on realistic quantum systems and that dissipation engineering is an essential tool for obtaining nonunitary evolutions. The proposed operations expand the quantum engineers' toolbox and have promising applications in NISQ algorithms and quantum machine learning.

Citations (3)


... The advantage of this approach is improved trainability due to the reduced size of the state space explored by the optimizer [48,69,71]. In quantum machine learning where encoding the right bias into the model is essential and one wants to avoid overfitting, this advantage is amplified [39,52]. Normally, in state preparation overfitting does not present an issue; we, however, train with an approximation of the entropy; therefore, an optimizer could overfit by exploiting errors in the approximation. ...

Reference:

Variational quantum thermalizers based on weakly-symmetric nonunitary multi-qubit operations
Provable advantages of kernel-based quantum learners and quantum preprocessing based on Grover's algorithm
  • Citing Article
  • September 2024

Physical Review A

... Subsequently, the excited state decays, completing the jump. Such operations can be readily implemented in physical systems such as trapped ions [78], superconducting qubits [64], or quantum dots in cavities [91]. Furthermore, the operational primitives have been shown to be universal for implementing Lindbladians on two qubits [78]. ...

Experimental Realization of Nonunitary Multiqubit Operations
  • Citing Article
  • January 2024

Physical Review Letters

... Subsequently, the excited state decays, completing the jump. Such operations can be readily implemented in physical systems such as trapped ions [78], superconducting qubits [64], or quantum dots in cavities [91]. Furthermore, the operational primitives have been shown to be universal for implementing Lindbladians on two qubits [78]. ...

Nonunitary gate operations by dissipation engineering