arXiv subjects

Edoardo Altamura

Publications and source records attributed to Edoardo Altamura.

14 recordsLinked to original sources

arXiv2026

Impact of Data Loss in Postprocessing on Training and Inference of Quantum Neural Networks

As quantum hardware scales to larger devices, the classical software layers that interface with it must evolve in step. Postprocessing routines developed and tested primarily in simulator settings can encode assumptions that no longer hold on utility-scale devices, leading to data loss that can be difficult to detect from high-level model outputs alone. We present a case study of \texttt{SamplerQNN}, the sampling-based quantum neural network class in the Qiskit Machine Learning library. Here, the postprocessing method applies a filter that assumes measurement bit-strings are in virtual qubit space. On our quantum hardware runs, where bit-strings span over 100 physical qubits, this filter led to the loss of 85 to 99.6\% of valid measurement shots, depending on the transpiler's qubit placement. The resulting probability vector is unnormalised, allowing distorted prediction and loss values to propagate through the model without an API-level warning. We demonstrate the impact across five experiments on two IBM backends: for inference, accuracy drops from 0.94 to 0.39 on the same raw measurements; for training, the loss signal is compressed by 22 to 27$\times$, substantially reducing the sensitivity of the optimiser to the objective landscape. The behaviour arises in all released versions of the library (0.8.4 to 0.9.0). We implemented a layout-based marginalisation fix, merged into the GitHub codebase as Pull Request \#1041, that makes \texttt{SamplerQNN} postprocessing forward-compatible with current and upcoming hardware.

quant-ph↗

arXiv2026

Sensitivity of Next-Generation CMB Surveys to Neutrinos and Other Light Relics

Neutrinos and other light relics leave characteristic imprints in the cosmic microwave background anisotropies, making their observation a sensitive probe of the particle content and thermal history of the early universe. The energy density in these relativistic species is parameterized by their effective number $N_\mathrm{eff}$. Measuring this parameter at the percent level, which is a long-standing science goal of CMB-S4 and other experiments, would test a wide range of well-motivated physics within and beyond the Standard Model of particle physics. In this paper, we present Fisher-matrix forecasts of the projected sensitivity to $N_\mathrm{eff}$ of several CMB-S4 survey configurations considered during its extensive design phase. The conceptual design reaches $\sigma(N_\mathrm{eff}) < 0.03$ over its seven-year observing period, while the revised configuration achieves the same precision over a longer timescale. We complement these results with a cosmic-variance-limited survey over the same multipole range to quantify the room for improvement accessible with additional instrumental, observational, and theoretical efforts. Finally, we discuss the broad implications of precise $N_\mathrm{eff}$ measurements for the radiation sector, big bang nucleosynthesis, light thermal relics, and other early-universe physics. The forecasts presented in this work are performed with the publicly released DRAFT (Dark Radiation Anisotropy Flowdown Team) tool. It provides an end-to-end pipeline from simulated foreground maps and component separation to delensing and projected sensitivities for any cosmological parameter, and it can be directly applied to other cosmic microwave background survey designs.

astro-ph.CO↗

arXiv2025

Quantum chemistry with provable convergence via randomized sample-based Krylov quantum diagonalization

Quantum algorithms based on classical processing of individual samples have recently emerged as the most effective and robust methods to approximate ground-state wave functions of many-body quantum systems on pre-fault-tolerant and early-fault-tolerant quantum devices. In these algorithms, the quantum computer acts as a sampling engine that generates the subspace in which the Hamiltonian is classically diagonalized. The recently proposed Sample-based Krylov Quantum Diagonalization (SKQD), uses quantum Krylov states as circuits from which samples are collected. Convergence guarantees can be derived for SKQD under similar assumptions to those of quantum phase estimation, provided that the ground-state wave function is well approximated by a polynomial subset of the full Hilbert space. However, implementations of SKQD for complex many-body Hamiltonians, such as quantum chemistry ones, are limited by the depths of time-evolution circuits needed to generate Krylov vectors. In this work, we introduce a method that combines SKQD with a qDRIFT randomized compilation of the Hamiltonian propagator. The resulting algorithm, termed SqDRIFT, enables quantum chemistry experiments on quantum processors, while preserving the convergence guarantees similar to the phase estimation algorithm. We demonstrate its viability by applying SqDRIFT to calculate the electronic ground-state energy of several polycyclic aromatic hydrocarbons, up to system sizes beyond the reach of exact diagonalization.

quant-ph↗

arXiv2025

Encoding molecular structures in quantum machine learning

Quantum machine learning (QML) has great potential for the analysis of chemical datasets. However, conventional quantum data-encoding schemes, such as fingerprint encoding, are generally unfeasible for the accurate representation of chemical moieties in such datasets. In this contribution, we introduce the quantum molecular structure encoding (QMSE) scheme, which encodes the molecular bond orders and interatomic couplings expressed as a hybrid Coulomb-adjacency matrix, directly as one- and two-qubit rotations within parameterised circuits. We show that this strategy provides an efficient and interpretable method in improving state separability between encoded molecules compared to other fingerprint encoding methods, which is especially crucial for the success in preparing feature maps in QML workflows. To benchmark our method, we train a parameterised ansatz on molecular datasets to perform classification of state phases and regression on boiling points, demonstrating the competitive trainability and generalisation capabilities of QMSE. We further prove a fidelity-preserving chain-contraction theorem that reuses common substructures to cut qubit counts, with an application to long-chain fatty acids. We expect this scalable and interpretable encoding framework to greatly pave the way for practical QML applications of molecular datasets.

quant-ph↗

arXiv2025

Approximate quantum circuit compilation for proton-transfer kinetics on quantum processors

Proton transfer reactions are fundamental to many chemical and biological systems, where quantum effects such as tunneling, delocalization, and zero-point motion play key kinetic control roles. However, classical methods capable of accurately capturing these phenomena scale prohibitively with system size. Here, we develop and demonstrate quantum computing algorithms based on the Nuclear-Electronic Orbital framework, treating the transferring proton quantum mechanically. We assess the potential of current quantum devices for simulating proton transfer kinetics with high accuracy. We first construct a deep initial ans\"atze within a truncated orbital space by employing the frozen natural orbital approximation. Then, to balance circuit depth against state fidelity, we implement an adaptive form of approximate quantum compiling. Using resulting circuits at varying compression levels transpiled for the ibm_fez device, we compute barrier heights and delocalised proton densities along the proton transfer pathway using a realistic hardware noise model. We find that, although current quantum hardware introduces significant noise relative to the demanding energy tolerances involved, our approach allows substantial circuit simplification while maintaining energy barrier estimates within 13% of the reference value. Despite present hardware limitations, these results offer a practical means of approximating key circuit segments in near-term devices and early fault-tolerant quantum computing systems.

quant-ph↗

arXiv2025

Qiskit Machine Learning: an open-source library for quantum machine learning tasks at scale on quantum hardware and classical simulators

We present Qiskit Machine Learning (ML), a high-level Python library that combines elements of quantum computing with traditional machine learning. The API abstracts Qiskit's primitives to facilitate interactions with classical simulators and quantum hardware. Qiskit ML started as a proof-of-concept code in 2019 and has since been developed to be a modular, intuitive tool for non-specialist users while allowing extensibility and fine-tuning controls for quantum computational scientists and developers. The library is available as a public, open-source tool and is distributed under the Apache version 2.0 license.

quant-ph↗

arXiv2025

Entropy plateaus can emerge from gas replacement at a characteristic halo mass in simulated groups and clusters of galaxies

The evolution of the intergalactic medium (IGM) is influenced by gravitational collapse, radiative cooling, and baryonic feedback. Using cosmological hydrodynamic zoom-in simulations of a $8.83 \times 10^{12}$ M$_\odot$ group and a $2.92 \times 10^{14}$ M$_\odot$ cluster at $z=0$, we investigate the emergence of entropy plateaus and their connection to feedback mechanisms. This set-up uses the SWIFT-EAGLE model with three resolutions, down to an initial particle gas mass of $2.29 \times 10^5$ M$_\odot$ and $1.23 \times 10^6$ M$_\odot$ for dark matter. We find that, when halos reach the characteristic mass of $\sim 10^{12}$ M$_{\odot}$, their entropy profiles flatten at the virial radius, marking a transition from supernova to AGN feedback-driven regulation. As halos grow into groups ($\sim 10^{13}$ M$_{\odot}$), the entropy plateau extends inward and isentropic cores form in massive systems ($\sim 10^{14}$ M$_{\odot}$). By tracking the Lagrangian history of gas particles, we demonstrate that this entropy buildup is primarily driven by AGN feedback, which efficiently removes low-entropy gas from progenitors of groups and clusters, redistributing it throughout the IGM before falling into the core. Recent observations of X-GAP groups reveal large entropy excesses and plateaus, in line with our findings and in contrast to the power-law-like profiles of most previous observations. While entropy plateaus and large entropy excesses may be observationally confirmed in unbiased samples, reproducing the full diversity of entropy profiles remains an outstanding challenge for next-generation feedback models. Our results suggest that current feedback models may be overly efficient in expelling low-entropy gas from the potential cool-core progenitors, disrupting the balance between heating and cooling required for long-lived cool cores.

astro-ph.CO↗

arXiv2024

Evolving a multi-population evolutionary-QAOA on distributed QPUs

Our work integrates an Evolutionary Algorithm (EA) with the Quantum Approximate Optimization Algorithm (QAOA) to optimize ansatz parameters in place of traditional gradient-based methods. We benchmark this Evolutionary-QAOA (E-QAOA) approach on the Max-Cut problem for $d$-3 regular graphs of 4 to 26 nodes, demonstrating equal or higher accuracy and reduced variance compared to COBYLA-based QAOA, especially when using Conditional Value at Risk (CVaR) for fitness evaluations. Additionally, we propose a novel distributed multi-population EA strategy, executing parallel, independent populations on two quantum processing units (QPUs) with classical communication of 'elite' solutions. Experiments on quantum simulators and IBM hardware validate the approach. We also discuss potential extensions of our method and outline promising future directions in scalable, distributed quantum optimization on hybrid quantum-classical infrastructures.

quant-ph↗

arXiv2023

Building models of the Universe with hydrodynamic simulations

Hydrodynamic simulations have become irreplaceable in modern cosmology for exploring complex systems and making predictions to steer future observations. In Chapter 1, we begin with a philosophical discussion on the role of simulations in science. We argue that simulations can bridge the gap between empirical and fundamental knowledge. The validation of simulations stresses the importance of achieving a balance between trustworthiness and scepticism. Next, Chapter 2 introduces the formation of structures and comparisons between synthetic and observational data. Chapter 3 describes the production pipeline of zoom-in simulations used to model individual objects and novel methods to mitigate known shortcomings. Then, we assessed the weak scaling of the SWIFT code and found it to be one of the hydrodynamic codes with the highest parallel efficiency. In Chapter 4, we study the rotational kinetic Sunyaev-Zeldovich (rkSZ) effect for high-mass galaxy clusters from the MACSIS simulations. We find a maximum signal greater than 100 $\mu$K, 30 times stronger than early predictions from self-similar models, opening prospects for future detection. In Chapter 5, we address a tension between the distribution of entropy measured from observations and predicted by simulations of groups and clusters of galaxies. We find that most recent hydrodynamic simulations systematically over-predict the entropy profiles by up to one order of magnitude, leading to profiles that are shallower and higher than the power-law-like entropy profiles that have been observed. We discuss the dependence on different hydrodynamic and sub-grid parameters using variations of the EAGLE model. Chapter 6 explores the evolution of the profiles as a function of cosmic time. We report power-law-like entropy profiles at high redshift for both objects. However, at late times, an entropy plateau develops and alters the shape of the profile.

astro-ph.CO↗

arXiv2023

Inferring the dark matter splashback radius from cluster gas and observable profiles in the FLAMINGO simulations

The splashback radius, coinciding with the minimum in the dark matter radial density gradient, is thought to be a universal definition of the edge of a dark matter halo. Observational methods to detect it have traced the dark matter using weak gravitational lensing or galaxy number counts. Recent attempts have also claimed the detection of a similar feature in Sunyaev-Zel'dovich (SZ) observations of the hot intracluster gas. Here, we use the FLAMINGO simulations to investigate whether an extremum gradient in a similar position to the splashback radius is predicted to occur in the cluster gas profiles. We find that the minimum in the gradient of the stacked 3D gas density and pressure profiles, and the maximum in the gradient of the entropy profile, broadly align with the splashback feature though there are significant differences. While the dark matter splashback radius varies with specific mass accretion rate, in agreement with previous work, the radial position of the deepest minimum in the log-slope of the gas density is more sensitive to halo mass. In addition, we show that a similar minimum is also present in projected 2D pseudo-observable profiles: emission measure (X-ray); Compton-$y$ (SZ) and surface mass density (weak lensing). We find that the latter traces the dark matter results reasonably well albeit the minimum occurs at a slightly smaller radius. While results for the gas profiles are largely insensitive to accretion rate and various observable proxies for dynamical state, they do depend on the strength of the feedback processes.

astro-ph.CO↗

arXiv2023

Galaxy cluster rotation revealed in the MACSIS simulations with the kinetic Sunyaev-Zeldovich effect

The kinetic Sunyaev-Zeldovich (kSZ) effect has now become a clear target for ongoing and future studies of the cosmic microwave background (CMB) and cosmology. Aside from the bulk cluster motion, internal motions also lead to a kSZ signal. In this work, we study the rotational kSZ effect caused by coherent large-scale motions of the cluster medium using cluster hydrodynamic cosmological simulations. To utilise the rotational kSZ as a cosmological probe, simulations offer some of the most comprehensive data sets that can inform the modelling of this signal. In this work, we use the MACSIS data set to investigate the rotational kSZ effect in massive clusters specifically. Based on these models, we test stacking approaches and estimate the amplitude of the combined signal with varying mass, dynamical state, redshift and map-alignment geometry. We find that the dark matter, galaxy and gas spins are generally misaligned, an effect that can cause a sub-optimal estimation of the rotational kSZ effect when based on galaxy motions. Furthermore, we provide halo-spin-mass scaling relations that can be used to build a statistical model of the rotational kSZ. The rotational kSZ contribution, which is largest in massive unrelaxed clusters ($\gtrsim$100 $\mu$K), could be relevant to studies of higher-order CMB temperature signals, such as the moving lens effect. The limited mass range of the MACSIS sample strongly motivates an extended investigation of the rotational kSZ effect in large-volume simulations to refine the modelling, particularly towards lower mass and higher redshift, and provide forecasts for upcoming cosmological CMB experiments (e.g. Simons Observatory, SKA-2) and X-ray observations (e.g. Athena/X-IFU).

astro-ph.CO↗

arXiv2022

The impact of stochastic modeling on the predictive power of galaxy formation simulations

All modern galaxy formation models employ stochastic elements in their sub-grid prescriptions to discretise continuous equations across the time domain. In this paper, we investigate how the stochastic nature of these models, notably star formation, black hole accretion, and their associated feedback, that act on small ($<$ kpc) scales, can back-react on macroscopic galaxy properties (e.g. stellar mass and size) across long ($>$ Gyr) timescales. We find that the scatter in scaling relations predicted by the EAGLE model implemented in the SWIFT code can be significantly impacted by random variability between re-simulations of the same object, even when galaxies are resolved by tens of thousands of particles. We then illustrate how re-simulations of the same object can be used to better understand the underlying model, by showing how correlations between galaxy stellar mass and black hole mass disappear at the highest black hole masses ($M_{\rm BH} > 10^8$ M$_\odot$), indicating that the feedback cycle may be interrupted by external processes. We find that although properties that are collected cumulatively over many objects are relatively robust against random variability (e.g. the median of a scaling relation), the properties of individual galaxies (such as galaxy stellar mass) can vary by up to 25\%, even far into the well-resolved regime, driven by bursty physics (black hole feedback) and mergers between galaxies. We suggest that studies of individual objects within cosmological simulations be treated with caution, and that any studies aiming to closely investigate such objects must account for random variability within their results.

astro-ph.GA↗

arXiv2022

Gas clumping and its effect on hydrostatic bias in the MACSIS simulations

We use the MACSIS hydrodynamical simulations to estimate the extent of gas clumping in the intracluster medium of massive galaxy clusters and how it affects the hydrostatic mass bias. By comparing the clumping to the azimuthal scatter in the emission measure, an observational proxy, we find that they both increase with radius and are larger in higher-mass and dynamically perturbed systems. Similar trends are also seen for the azimuthal temperature scatter and non-thermal pressure fraction, both of which correlate with density fluctuations, with these values also increasing with redshift. However, in agreement with recent work, we find only a weak correlation between the clumping, or its proxies, and the hydrostatic mass bias. To reduce the effect of clumping in the projected profiles, we compute the azimuthal median following recent observational studies, and find this reduces the scatter in the bias. We also attempt to correct the cluster masses by using a non-thermal pressure term and find over-corrected mass estimates ($1-b=0.86$ to $1-b=1.15$) from 3D gas profiles but improved mass estimates ($1-b=0.75$ to $1-b=0.85$) from projected gas profiles, with the caveat of systematically increased scatter. We conclude that the cluster-averaged mass bias is minimised from applying a non-thermal pressure correction ($1-b=0.85$) with more modest reductions from selecting clusters that have low clumping ($1-b=0.79$) or are dynamically relaxed ($1-b=0.80$). However, the latter selection is most effective at minimising the scatter for individual objects. Such results can be tested with next generation X-ray missions equipped with high-resolution spectrometers such as Athena.

astro-ph.CO↗

arXiv2022

EAGLE-like simulation models do not solve the entropy core problem in groups and clusters of galaxies

Recent high-resolution cosmological hydrodynamic simulations run with a variety of codes systematically predict large amounts of entropy in the intra-cluster medium at low redshift, leading to flat entropy profiles and a suppressed cool-core population. This prediction is at odds with X-ray observations of groups and clusters. We use a new implementation of the EAGLE galaxy formation model to investigate the sensitivity of the central entropy and the shape of the profiles to changes in the sub-grid model applied to a suite of zoom-in cosmological simulations of a group of mass $M_{500} = 8.8 \times 10^{12}~{\rm M}_\odot$ and a cluster of mass $2.9 \times 10^{14}~{\rm M}_\odot$. Using our reference model, calibrated to match the stellar mass function of field galaxies, we confirm that our simulated groups and clusters contain hot gas with too high entropy in their cores. Additional simulations run without artificial conduction, metal cooling or AGN feedback produce lower entropy levels but still fail to reproduce observed profiles. Conversely, the two objects run without supernova feedback show a significant entropy increase which can be attributed to excessive cooling and star formation. Varying the AGN heating temperature does not greatly affect the profile shape, but only the overall normalisation. Finally, we compared runs with four AGN heating schemes and obtained similar profiles, with the exception of bipolar AGN heating, which produces a higher and more uniform entropy distribution. Our study leaves open the question of whether the entropy core problem in simulations, and particularly the lack of power-law cool-core profiles, arise from incorrect physical assumptions, missing physical processes, or insufficient numerical resolution.

astro-ph.CO↗