Search

The theoretical foundation of quantum sensing is rooted in the Cram\'er-Rao formalism, which establishes quantitative precision bounds for a given quantum probe. In many practical scenarios, where more than one parameter is unknown, the multi-parameter Cram\'er-Rao bound (CRB) applies. Since this is a matrix inequality involving the inverse of the quantum Fisher information matrix (QFIM), the formalism breaks down when the QFIM is singular. In this paper, we examine the physical origins of such singularities, showing that they result from an over-parametrization on the metrological level. This is itself caused by emergent metrological symmetries, whereby the same set of measurement outcomes are obtained for different combinations of system parameters. Although the number of effective parameters is equal to the number of non-zero QFIM eigenvalues, the Cram\'er-Rao formalism typically does not provide information about the effective parameter encoding. Instead, we demonstrate through a series of concrete examples that Bayesian estimation can provide deep insights. In particular, the metrological symmetries appear in the Bayesian posterior distribution as lines of persistent likelihood running through the space of unknown parameters. These lines are contour lines of the effective parameters which, through suitable parameter transformations, can be estimated and follow their own effective CRBs., Comment: 22 pages with 8 figures

Modern artificial intelligence (AI) systems are powered by foundation models. This paper presents a new set of foundation models, called Llama 3. It is a herd of language models that natively support multilinguality, coding, reasoning, and tool usage. Our largest model is a dense Transformer with 405B parameters and a context window of up to 128K tokens. This paper presents an extensive empirical evaluation of Llama 3. We find that Llama 3 delivers comparable quality to leading language models such as GPT-4 on a plethora of tasks. We publicly release Llama 3, including pre-trained and post-trained versions of the 405B parameter language model and our Llama Guard 3 model for input and output safety. The paper also presents the results of experiments in which we integrate image, video, and speech capabilities into Llama 3 via a compositional approach. We observe this approach performs competitively with the state-of-the-art on image, video, and speech recognition tasks. The resulting models are not yet being broadly released as they are still under development.

Critical quantum metrology relies on the extreme sensitivity of a system's eigenstates near the critical point of a quantum phase transition to Hamiltonian perturbations. This means that these eigenstates are extremely sensitive to all the parameters of the Hamiltonian. In practical settings, there always exists a degree of experimental uncertainty in the control parameters - which are approximately known quantities. Despite such uncertainties representing the most relevant source of noise in critical metrology, their impact on the attainable precision has been largely overlooked. In this work we present a general framework, interpolating between the single and multi-parameter estimation settings, allowing for the proper bookkeeping of relevant errors. We apply this framework to the paradigmatic transverse field Ising and Lipkin-Meshkov-Glick models, explicitly showing how uncertainty in control parameters impacts the sensitivity of critical sensors. For finite-size systems, we establish that there exists a trade-off between the amount of uncertainty a many-body probe can withstand while still maintaining a quantum advantage in parameter estimation., Comment: 9 pages, 3 figures

In this paper we prove discrete to continuum convergence rates for Poisson Learning, a graph-based semi-supervised learning algorithm that is based on solving the graph Poisson equation with a source term consisting of a linear combination of Dirac deltas located at labeled points and carrying label information. The corresponding continuum equation is a Poisson equation with measure data in a Euclidean domain $\Omega \subset \mathbb{R}^d$. The singular nature of these equations is challenging and requires an approach with several distinct parts: (1) We prove quantitative error estimates when convolving the measure data of a Poisson equation with (approximately) radial function supported on balls. (2) We use quantitative variational techniques to prove discrete to continuum convergence rates on random geometric graphs with bandwidth $\varepsilon>0$ for bounded source terms. (3) We show how to regularize the graph Poisson equation via mollification with the graph heat kernel, and we study fine asymptotics of the heat kernel on random geometric graphs. Combining these three pillars we obtain $L^1$ convergence rates that scale, up to logarithmic factors, like $O(\varepsilon^{\frac{1}{d+2}})$ for general data distributions, and $O(\varepsilon^{\frac{2-\sigma}{d+4}})$ for uniformly distributed data, where $\sigma>0$. These rates are valid with high probability if $\varepsilon\gg\left({\log n}/{n}\right)^q$ where $n$ denotes the number of vertices of the graph and $q \approx \frac{1}{3d}$.

Quantum systems used for metrology can offer enhanced precision over their classical counterparts. The design of quantum sensors can be optimized by maximizing the quantum Fisher information (QFI), which characterizes the precision of parameter estimation for an ideal measurement. Here we consider the response of a quantum system as a means to estimate the strength of an external perturbation that has been switched on slowly. General expressions for the QFI are derived, which also hold for interacting many-body systems in the thermodynamic limit at finite temperatures, and can be related to linear-response transport coefficients. For quantum dot nanoelectronics devices, we show that electron interactions can lead to exponential scaling of the QFI with system size, highlighting that quantum resources can be utilized in the full Fock space. The precision estimation of voltages and fields can also be achieved by practical global measurements, such as the electric current, making quantum circuits good candidates for metrological applications., Comment: 4+ pages plus supplementary material

Within the coronae of stars, abundances of those elements with low first ionization potential (FIP) often differ from their photospheric values. The coronae of the Sun and solar-type stars mostly show enhancements of low-FIP elements (the FIP effect) while more active stars such as M dwarfs have coronae generally characterized by the inverse-FIP (I-FIP) effect. Highly localized regions of I-FIP effect solar plasma have been observed by Hinode/EIS in a number of highly complex active regions, usually around strong light bridges of the umbrae of coalescing/merging sunspots. These observations can be interpreted in the context of the ponderomotive force fractionation model which predicts that plasma with I-FIP effect composition is created by the refraction of waves coming from below the plasma fractionation region in the chromosphere. A plausible source of these waves is thought to be reconnection in the (high-plasma \b{eta}) subchromospheric magnetic field. In this study, we use the 3D visualization technique of Chintzoglou & Zhang (2013) combined with observations of localized I-FIP effect in the corona of AR 11504 to identify potential sites of such reconnection and its possible consequences in the solar atmosphere. We found subtle signatures of episodic heating and reconnection outflows in the expected places, in between magnetic flux tubes forming a light bridge, within the photosphere of the active region. Furthermore, on either side of the light bridge, we observed small antiparallel horizontal magnetic field components supporting the possibility of reconnection occuring where we observe I-FIP plasma. When taken together with the I-FIP effect observations, these subtle signatures provide a compelling case for indirect observational evidence of reconnection below the fractionation layer of the chromosphere, however, direct evidence remains elusive., Comment: Accepted ApJ

This paper describes the work of the UniBuc Archaeology team for CLPsych's 2024 Shared Task, which involved finding evidence within the text supporting the assigned suicide risk level. Two types of evidence were required: highlights (extracting relevant spans within the text) and summaries (aggregating evidence into a synthesis). Our work focuses on evaluating Large Language Models (LLM) as opposed to an alternative method that is much more memory and resource efficient. The first approach employs a good old-fashioned machine learning (GOML) pipeline consisting of a tf-idf vectorizer with a logistic regression classifier, whose representative features are used to extract relevant highlights. The second, more resource intensive, uses an LLM for generating the summaries and is guided by chain-of-thought to provide sequences of text indicating clinical markers., Comment: https://github.com/nlp-unibuc/clpsych24-task

Quantum systems can be used as probes in the context of metrology for enhanced parameter estimation. In particular, the delicacy of critical systems to perturbations can make them ideal sensors. Arguably the simplest realistic probe system is a spin-1/2 impurity, which can be manipulated and measured in-situ when embedded in a fermionic environment. Although entanglement between a single impurity probe and its environment produces nontrivial many-body effects, criticality cannot be leveraged for sensing. Here we introduce instead the two-impurity Kondo (2IK) model as a novel paradigm for critical quantum metrology, and examine the multiparameter estimation scenario at finite temperature. We explore the full metrological phase diagram numerically and obtain exact analytic results near criticality. Enhanced sensitivity to the inter-impurity coupling driving a second-order phase transition is evidenced by diverging quantum Fisher information (QFI) and quantum signal-to-noise ratio (QSNR). However, with uncertainty in both coupling strength and temperature, the multiparameter QFI matrix becomes singular -- even though the parameters to be estimated are independent -- resulting in vanishing QSNRs. We demonstrate that by applying a known control field, the singularity can be removed and measurement sensitivity restored. For general systems, we show that the degradation in the QSNR due to uncertainties in another parameter is controlled by the degree of correlation between the unknown parameters., Comment: 19 pages, 9 figures

The plasma composition of the solar corona is different from that of the solar photosphere. Elements that have a low first ionisation potential (FIP) are preferentially transported to the corona and, therefore, show enhanced abundances in the corona compared to the photosphere. The level of enhancement is measured using the FIP bias parameter. In this work, we use data from the EUV Imaging Spectrometer (EIS) on Hinode to study the plasma composition in an active region following an episode of significant new flux emergence into the pre-existing magnetic environment of the active region. We use two FIP bias diagnostics: Si X 258.375 A/S X 264.233 A (temperature of approximately 1.5 MK) and Ca XIV 193.874 A/Ar XIV 194.396 A (temperature of approximately 4 MK). We observe slightly higher FIP bias values with the Ca/Ar diagnostic than Si/S in the newly emerging loops, and this pattern is much stronger in the preexisting loops (those that had been formed before the flux emergence). This result can be interpreted in the context of the ponderomotive force model, which proposes that the plasma fractionation is generally driven by Alfv\'en waves. Model simulations predict this difference between diagnostics using simple assumptions about the wave properties, particularly that the fractionation is driven by resonant/non-resonant waves in the emerging/preexisting loops. We propose that this results in the different fractionation patterns observed in these two sets of loops., Comment: Accepted for publication in The Astrophysical Journal

The Slow Solar Wind Connection Solar Orbiter Observing Plan (Slow Wind SOOP) was developed to utilise the extensive suite of remote sensing and in situ instruments on board the ESA/NASA Solar Orbiter mission to answer significant outstanding questions regarding the origin and formation of the slow solar wind. The Slow Wind SOOP was designed to link remote sensing and in situ measurements of slow wind originating at open-closed field boundaries. The SOOP ran just prior to Solar Orbiter's first close perihelion passage during two remote sensing windows (RSW1 and RSW2) between 2022 March 3-6 and 2022 March 17-22, while Solar Orbiter was at a heliocentric distance of 0.55-0.51 and 0.38-0.34 au from the Sun, respectively. Coordinated observation campaigns were also conducted by Hinode and IRIS. The magnetic connectivity tool was used, along with low latency in situ data, and full-disk remote sensing observations, to guide the target pointing of Solar Orbiter. Solar Orbiter targeted an active region complex during RSW1, the boundary of a coronal hole, and the periphery of a decayed active region during RSW2. Post-observation analysis using the magnetic connectivity tool along with in situ measurements from MAG and SWA/PAS, show that slow solar wind, with velocities between 210 and 600 km/s, arrived at the spacecraft originating from two out of the three of the target regions. The Slow Wind SOOP, despite presenting many challenges, was very successful, providing a blueprint for planning future observation campaigns that rely on the magnetic connectivity of Solar Orbiter., Comment: 24 pages, 10 figures