Tuesday 13 September 2022

Edificio 18

9:00 – 9:25


9:25 – 10:25

10:25 – 10:50


Cooperative quantum information erasure

We demonstrate an information erasure protocol that resets N qubits at once. The method works (within the estimated error) at Landauer energy cost and sets the current record of energy-x-time cost. The method departs from the standard algorithmic cooling paradigm by exploiting cooperative effects associated to the mechanism of spontaneous symmetry breaking which are amplified by quantum tunnelling phenomena. Such cooperative quantum erasure protocol is demonstrated on a commercial quantum annealer and could be readily applied in next generation hybrid gate-based/quantum-annealing quantum computers, for fast, effective, and energy efficient initialisation of quantum processing units.

10:50 – 11:20

Coffee Break

11:20 – 11:45

De Chiara

Exploiting Coherence in Quantum Thermodynamics

The study of out-of-equilibrium thermodynamics of quantum systems has re- ceived increasing attention in recent years thanks to tremendous theoretical and experimental progress. While most of the studies in quantum thermodynam- ics bear a close resemblance to their classical counterparts, especially close to equilibrium, there are only a few examples of genuine quantum features, e.g. co- herence, squeezing and entanglement, that provide an advantage over classical thermodynamic devices. In this contribution, I will show how thermal equilib- rium reservoirs equipped with an infinitesimal amount of quantum coherence exhibit such an advantage. In fact, reservoir quantum coherence allows the de- sign of engine and refrigerators with efficiencies that exceed the corresponding Carnot’s efficiency of a classical machine operating with the same temperatures. Such thermal machines provide efficiencies at maximum power that exceed the classical Curzon-Ahlborn value. Moreover, the injected coherence allows for a hybrid refrigerators which extracts heat from the coldest bath and simultane- ously produces work. [1] K. Hammam, H. Leitch, Y. Hassouni and G. De Chiara, arXiv:2202.07515

11:45 – 12:35

12:35 – 13:00


Power maximization of two-stroke quantum thermal machines

Two different versions of a two-stroke quantum thermal machine are studied [1]. In both versions, two collections of identical systems with evenly spaced nonvariable energy levels can be put in contact, respectively, with a cold and a hot thermal bath. In the first version, a system of a collection interacts with a system of the other one, and then they thermalize. In the second one, a mediator system interacts alternately with one or more systems of each collection. We show that the efficiencies of these machines depend only on the energy gaps of the systems and are equal to the efficiency of “equivalent” Otto cycles. Focusing on the cases of qubits or harmonic oscillators for both models, we maximize the engine power and analyze, in the model without the mediator, the role of the waiting time between subsequent interactions. We find that in both cycles, the power peaks of qubit systems can surpass the Curzon-Ahlborn efficiency. Finally, we compare our cycle without the mediator with previous schemes of the quantum Otto cycle showing that high coupling is not required to achieve the same maximum power. [1] N. Piccione, G. De Chiara, and B. Bellomo, Phys. Rev. A 103, 032211 (2021).

13:00 – 15:00


15:00 – 15:25


Quantum and classical phase transitions in a two-impurity spin-boson model

The two-impurity spin-boson model (TISBM), consisting in two interacting spin- 1/2’s coupled with a common reservoir composed of quantum harmonic oscilla- tors, is analysed. Several crucial and still open questions concerning the TISBM are currently under active consideration: i) the existence of critical points and then the presence of quantum and/or classical phase transitions; ii) if the even- tual quantum phase transitions are of the Kosterlitz-Thouless type. We consider the class of TISBMs with no transverse fields applied to the spin pair. We show that the role of the tunneling parameter is played by the spin-spin coupling. We demonstrate the presence of: – A classical (finite temperature) phase transi- tion characterized by a spin-spin coupling-dependent critical temperature. – A spin-spin coupling-based quantum (zero temperature) phase transition due to a level crossing with respect to the spin-bath coupling strength(s). – A Kosterlitz- Thouless (quantum) phase transition with a consequent localization phenomenon of the spin-system. – A decoherence-free subspace where the two coupled spins experience a dissipationless dynamics.


Driving quantum thermal machines with optimal power/efficiency trade-offs using reinforcement learning

The optimal control of non-equilibrium open quantum systems is a challenging task but has a key role in improving existing quantum information processing technologies. We introduce a model-free framework, based on Reinforcement Learning (RL), to identify optimal protocols for driven quantum thermal machines [1,2]. Quantum thermal machines, such as heat engines and refrigerators, are quantum devices that convert between heat and work through time-dependent controls that are periodically driven to implement thermodynamic cycles. We introduce a general framework, based on state-of-the-art RL algorithms, to dis- cover optimal cycles that are Pareto optimal trade-offs between power and effi- ciency. The method is model-free, and only requires the observation of the heat fluxes. We test our method on an experimentally realistic refrigerator based on a superconducting qubit, and on a heat engine based on a quantum harmonic oscillator. In both cases we find elaborate cycles that outperform previous liter- ature proposals and the optimized Otto cycle. [1] P.A. Erdman and F. Noé, NPJ Quantum Inf. 8, 1 (2022). [2] P.A. Erdman and F. Noé, arXiv:2204.04785 (2022).

15:25 – 15:50


Environment-induced quantum phase transitions in dissipative systems

Quite recently,the issue of achieving a reliable theoretical description of open quantum systems in theintermediate up to strong system-environment coupling regime has regained interest.Indeed,it is relevant to a wide range of quantum technologies,including quantum annealers andfar-reaching quantum thermody- namics devices.Here we present a theoretical study of several prototypical in- stances of quantum systems that arestrongly interacting with their surroundings
at low temperature.By employing the Caldeira-Leggett model,we analyze how
their equilibrium and out-of-equilibrium properties change with increasing cou- plingstrength compared to the conventional weak-coupling scenario.We address
the typical setting of single to many two-level systems coupled to a bath,as well
as theinstance of a single two-level system coupled to a dissipative quantum harmonic oscillator.Our results point towards the occurrence of environment- induced quantum phase transition of theBerezinskii-Kosterlitz-Thouless (BKT) kind,which eventually leads to the absence of quantum tunneling and to the localization of
the quantum degrees of freedom.


A learning theory for quantum photonic processors and beyond

We establish that quantum information processing architectures based on bosonic oscillators, such as photonic or mechanical ones, can be efficiently trained from examples to solve a broad class of tasks, including state reconstruction, dis- crimination and synthesis.,We consider a family of Gaussian and non-Gaussian continuous-variable (CV) circuits, suitable to describe state-of-the-art photonic processors, and evaluate its learning capabilities. ,Our basic learning problem is: given copies of an unknown quantum state, we apply to each of them a ran- dom quantum circuit from a given set C and obtain measurement outcomes. The objective is to approximate the unknown state using a state from a given hypothesis set S, using a small number of samples. The approximation is good if it reproduces the measurement statistics of all circuits in C on the unknown state, with small error and high probability. We show that a good approximation can be found with a number of samples polynomial in the number of modes of the CV circuit, which is a measure of its size. ,We apply our results to learning a decoder for optical communication, outperforming the state of the art.

15:50 – 16:15


Observation of a non-equilibrium superradiant phase transition in free space

We observe a non-equilibrium phase transition in a driven dissipative quantum system consisting of an elongated cloud of N laser-cooled atoms in free space, optically excited along its main axis. We find that our data are well reproduced by the iconic Driven Dicke model, which assumes a sub-wavelength sample vol- ume, by simply using an effective atom number. By measuring the excited state population of the atoms and the light emitted in the superradiant mode, we characterize the dynamics of the system and its steady-state properties. In par- ticular, we observe the characteristic N2 scaling of the photon emission rate in the superradiant phase, thus demonstrating steady-state superradiance in free space. Finally, we observe a modification of the statistics of the superradiant light as we cross the phase transition.


On the potential and limitations of quantum extreme learning machines

Quantum reservoir computers (QRC) and quantum extreme learning machines (QELM) aim to efficiently post-process the outcome of fixed — generally uncali- brated — quantum devices to solve tasks such as the estimation of the prop- erties of quantum states. The characterisation of their potential and limitations, which is currently lacking, will enable the full deployment of such approaches to problems of system identification, device performance optimization, and state or process reconstruction. We present a framework to model QRCs and QELMs, showing that they can be concisely described via single effective measurements, and provide an explicit characterisation of the information exactly retrievable with such protocols. We furthermore find a close analogy between the training process of QELMs and that of reconstructing the effective measurement char- acterising the given device. Our analysis paves the way to a more thorough understanding of the capabilities and limitations of both QELMs and QRCs, and has the potential to become a powerful measurement paradigm for quantum state estimation that is more resilient to noise and imperfections.

16:15 – 16:40


Photon Condensation: No-go and counter no-go theorems

Equilibrium phase transitions between a normal and a photon condensate state (also known as superradiant phase transitions) are a highly debated research topic, where proposals for their occurrence and no-go theorems have chased each other for the past four decades. Previous no-go theorems have demon- strated that gauge invariance forbids phase transitions to a photon condensate state when the cavity-photon mode is assumed to be spatially uniform. How- ever, it has been theoretically predicted that a collection of three-level systems coupled to light can display a first-order phase transition to a photon condensate state. It has also been recently shown that truncation of the Hilbert space of the matter system can affect the gauge invariance of the theory. However, it is al- ways possible to obtain approximate Hamiltonians obeying the gauge principle in the truncated Hilbert space. Here, we demonstrate a general no-go theo- rem for truncated, gauge-invariant models, which forbids first-order as well as second-order superradiant phase transitions in the absence of a magnetic field, in agreement with the general theory.


Machine Learning-Assisted Read out of Molecular Spin Qubits

We have recently implemented Storage/Retrieval protocols [NPJ QuantInf 6, 68 (2020)] and dispersive readout [AdvQuantTech 2100039 (2021)] on molecular spin qubits embedded into planar superconducting microwave resonators [AdvPhysX3 1435305 (2018)]. Along this line, we present our results on the use of machine learning to assist the
readout of the spin echo signal of an Oxovanadyl (VO(TPP)) molecular spin en- semble at low temperature. We first revisit our Storage/Retrieval protocol using trains of 4 input pulses, which allows us to codify up to 16 binary numbers. We first show that an Artificial Neural Network can be used to correctly recognize the output echoes from the raw measured traces without any prior information on their number or positions. Further post selection by means of a Clustering method allows us to successfully infer the initial input bit sequence. We then consider the phase of the Hahn echo signal, showing that it is possible to use an Artificial Neural Network to correctly infer the initial input phase from raw mea- sured data. Our approach is found to detect the effect of additional single-pulse phase control, holding potential to assist single-qubit gate operations.

16:40 – 17:10

Coffee Break

17:10 – 17:35


Learning Feedback Control Strategies for Quantum Metrology

In this talk I will discuss the usefulness of reinforcement learning techniques (RL) for enhancing the performance in a quantum metrology protocol. In particular we consider the problem of frequency estimation for a single bosonic field evolv- ing under a squeezing Hamiltonian and continuously monitored via homodyne detection. In the first part of the talk I will introduce continuously monitored quantum systems along with the mathematical formalism needed to describe them and the corresponding quantum Cramer-Rao bound that sets the ultimate precision in parameter estimation. Then I will (very) briefly discuss the concept of RL and in particular how to exploit these techniques for our particular metrologi- cal problem. Finally I will present our results: we show that the feedback control determined by the neural network greatly surpasses in the long-time limit the performances of both the “no-control” strategy and the standard “open-loop control” strategy, which we considered as benchmarks. We indeed observe how the devised strategy is able to optimize the nontrivial estimation problem by preparing a large fraction of trajectories corresponding to more sensitive quan- tum conditional states.

17:35 – 18:00


Quantum Asymmetry and Noisy Multimode Interferometry

Quantum asymmetry is a physical resource that coincides with the amount of coherence between the eigenspaces of a gen- erator responsible for phase encoding in interferometric experiments. We high- light an apparently counterintuitive behavior that the asymmetry may increase as a result of a decrease of coherence inside a degenerate subspace. We intu- itively explain and illustrate the phenomena by performing a three-mode single- photon interferometric experiment, where one arm carries the signal and two noisy reference arms have fluctuating phases. We show that the source of the observed sensitivity improvement is the reduction of correlations between these fluctuations and comment on the impact of the effect when moving from the single-photon quantum level to the classical regime. Finally, we also establish the analogy of the effect in the case of entanglement resource theory.

18:00 – 18:25


An Adversarial Learning Approach to Quantum Noise Sensing

One of the recent breakthroughs of Machine Learning (ML) has been Genera- tive Adversarial Networks (GANs). By exploiting results from Nash’s Game The- ory, these learning models can generate fake data with the same distribution of some target ones. In the spirit of early-day Quantum ML (QML), a direct ”quanti- zation” of this learning paradigm, dubbed Q-GANs, has been proposed, where the agents competing in the adversarial game are quantum and can be modeled by Parametrized Quantum Circuits (PQCs). We show how to solve typical issues that arise when the quantum states involved in QGANs training are mixed, and equipped with that piece of knowledge we move on to deploy SuperQGANs, a framework for learning quantum Super Operators in an adversarial fashion. Par- ticularly, we show a proof of concept of how this new paradigm could be used to sense noise affecting real quantum hardware, by modeling the noise as a Pauli channel and learning the associated error-rates. We show numerical evidence of the success of our method, even in the case of spatially and temporally correlated noise.