Search

In astrophysics, understanding the evolution of galaxies in primarily through imaging data is fundamental to comprehending the formation of the Universe. This paper introduces a novel approach to conditioning Denoising Diffusion Probabilistic Models (DDPM) on redshifts for generating galaxy images. We explore whether this advanced generative model can accurately capture the physical characteristics of galaxies based solely on their images and redshift measurements. Our findings demonstrate that this model not only produces visually realistic galaxy images but also encodes the underlying changes in physical properties with redshift that are the result of galaxy evolution. This approach marks a significant advancement in using generative models to enhance our scientific insight into cosmic phenomena.

Text-to-image (T2I) diffusion models have revolutionized generative modeling by producing high-fidelity, diverse, and visually realistic images from textual prompts. Despite these advances, existing models struggle with complex prompts involving multiple objects and attributes, often misaligning modifiers with their corresponding nouns or neglecting certain elements. Recent attention-based methods have improved object inclusion and linguistic binding, but still face challenges such as attribute misbinding and a lack of robust generalization guarantees. Leveraging the PAC-Bayes framework, we propose a Bayesian approach that designs custom priors over attention distributions to enforce desirable properties, including divergence between objects, alignment between modifiers and their corresponding nouns, minimal attention to irrelevant tokens, and regularization for better generalization. Our approach treats the attention mechanism as an interpretable component, enabling fine-grained control and improved attribute-object alignment. We demonstrate the effectiveness of our method on standard benchmarks, achieving state-of-the-art results across multiple metrics. By integrating custom priors into the denoising process, our method enhances image quality and addresses long-standing challenges in T2I diffusion models, paving the way for more reliable and interpretable generative models.

We study in detail the fully inhomogeneous non-linear dynamics of axion inflation, identifying three regimes: weak-, mild-, and strong-backreaction, depending on the duration of inflation. We use lattice techniques that explicitly preserve gauge invariance and shift symmetry, and which we validate against other computational methods of the linear dynamics and of the homogeneous backreaction regime. Notably, we demonstrate that the latter fails to accurately describe the truly local dynamics of strong backreaction. We investigate the convergence of simulations of local backreaction, determining the requirements to achieve an accurate description of the dynamics, and providing useful parametrizations of the delay of the end of inflation. Additionally, we identify key features emerging from a proper local treatment of strong backreaction: the dominance of magnetic energy against the electric counterpart, the excitation of the longitudinal mode, and the generation of a scale-dependent chiral (im)balance. Our results underscore the necessity to accurately capture the local nature of the non-linear dynamics of the system, in order to correctly assess phenomenological predictions, such as e.g. the production of gravitational waves and primordial black holes., Comment: 39 pages, 25 figures

We report on a study of axion string networks using fixed-grid simulations of up to $16384$ points per side. The length of string can be characterised in terms of standard dimensionless parameters $\zeta_\text{w}$ and $\zeta_\text{r}$, the length density measured in the cosmic rest frame and the string rest frame, scaled with the cosmic time. The motion of the string can be characterised by the root-mean-square (RMS) velocity of the string. Starting from a range of initial length densities and velocities, we analyse the string network in the standard scaling framework and find evolution towards a fixed point with estimated values $\hat{\zeta}_{\text{w},*} = 1.220(57)$ and $\hat{\zeta}_{\text{r},*} = 1.491(93)$. The two measures are related by the RMS velocity, which we estimate to be $\hat{v}_{*} = 0.5705(93)$. The length density is consistent with previous measurements, while the velocity is about 5% lower. For simulations starting from low enough density, the length density parameters $\zeta_\text{w}$ and $\zeta_\text{r}$ remain below their fixed point values throughout, while growing slowly, giving rise to an impression of approximately logarithmic increase with time. This has been proposed as the true long-term behaviour. We find that the growth tends to slow down as the values of $\zeta_\text{w}$ and $\zeta_\text{r}$ identified as fixed points are approached. In the case of $\zeta_\text{r}$, the growth stops for simulations which started close to the fixed point length density. The difference between $\zeta_\text{w}$ and $\zeta_\text{r}$ can be understood to result from the continuing velocity evolution. Our results indicate that the growth of $\zeta_\text{w}$ is a transient appearing at low densities and while the velocity is converging. This highlights the importance of studying the string density and the velocity together, and the preparation of initial conditions., Comment: 19 pp, 18 figures

Advancements in reinforcement learning have led to the development of sophisticated models capable of learning complex decision-making tasks. However, efficiently integrating world models with decision transformers remains a challenge. In this paper, we introduce a novel approach that combines the Dreamer algorithm's ability to generate anticipatory trajectories with the adaptive learning strengths of the Online Decision Transformer. Our methodology enables parallel training where Dreamer-produced trajectories enhance the contextual decision-making of the transformer, creating a bidirectional enhancement loop. We empirically demonstrate the efficacy of our approach on a suite of challenging benchmarks, achieving notable improvements in sample efficiency and reward maximization over existing methods. Our results indicate that the proposed integrated framework not only accelerates learning but also showcases robustness in diverse and dynamic scenarios, marking a significant step forward in model-based reinforcement learning.

We investigate the role of assortative mating in speciation using the sympatric model of Derrida and Higgs. The model explores the idea that genetic differences create incompatibilities between individuals, preventing mating if the number of such differences is too large. Speciation, however, only happens in this mating system if the number of genes is large. Here we show that speciation with small genome sizes can occur if assortative mating is introduced. In our model individuals are represented by three chromosomes: one responsible for reproductive compatibility, one for coding the trait on which assortativity will operate, and a neutral chromosome. Reproduction is possible if individuals are genetically similar with respect to the first chromosome, but among these compatible mating partners, the one with the most similar trait coded by the second chromosome is selected. We show that this type of assortativity facilitates speciation, which can happen with a small number of genes in the first chromosome. Species, classified according to reproductive isolation, dictated by the first chromosome, can display different traits values, as measured by the second and the third chromosomes. Therefore, species can also be identified based on similarity of the neutral trait, which works as a proxy for reproductive isolation., Comment: 23 pages, 7 figures

Reverberation is a key element in spatial audio perception, historically achieved with the use of analogue devices, such as plate and spring reverb, and in the last decades with digital signal processing techniques that have allowed different approaches for Virtual Analogue Modelling (VAM). The electromechanical functioning of the spring reverb makes it a nonlinear system that is difficult to fully emulate in the digital domain with white-box modelling techniques. In this study, we compare five different neural network architectures, including convolutional and recurrent models, to assess their effectiveness in replicating the characteristics of this audio effect. The evaluation is conducted on two datasets at sampling rates of 16 kHz and 48 kHz. This paper specifically focuses on neural audio architectures that offer parametric control, aiming to advance the boundaries of current black-box modelling techniques in the domain of spring reverberation., Comment: 8 pages, 7 figures, 2 tables

Using lattice field simulations of the Abelian-Higgs model, we characterize the simultaneous emission of (scalar and gauge) particles and gravitational waves (GWs) by local string loops. We use {\it network} loops created in a phase transition, and {\it artificial} loops formed by either crossing straight-boosted or curved-static infinite strings. Loops decay via both particle and GW emission, on time scales $\Delta t_{\rm dec} \propto L^p$, where $L$ is the loop length. For particle production, we find $p \simeq 2$ for artificial loops and $p \lesssim 1$ for network loops, whilst for GW emission, we find $p \simeq 1$ for all loops. We find that below a critical length, artificial loops decay primarily through particle production, whilst for larger loops GW emission dominates. However, for network loops, which represent more realistic configurations, particle emission always dominates, as supported by our data with length-to-core ratios up to $L/r_\text{c} \lesssim 3500$. Our results indicate that the GW background from a local string network should be greatly suppressed compared to estimations that ignore particle emission., Comment: 5 pages + Supplemental material, 2+3 figures, 2 tables

In this paper, a novel approach based on robust stability concepts and tools is introduced to evaluate the unconditional stability of microwave active $\textit{n}$-port devices. An efficient calculation of the Structured Singular Value of the $\textit{n}$x$\textit{n}$ scattering matrix is proposed to obtain the stability characteristics of the device. The presented method is validated in two ways. First, it is applied to a referential 4x4 scattering parameter set for independent verification. Second, the method is applied to a 4-port GaAs FET amplifier fabricated in hybrid technology. The results confirm the validity and computational efficiency of the proposed approach.

The motion of a charged particle moving on a flat surface is studied through the constants of motion associated to the system, given the magnetic gauge. The usual Landau' solution and the non separable solution for the Landau's gauge are found, and new non separable solution for the symmetric gauge is given. As a consequence of this, the quantization of the magnetic flux results from the invariance of the solutions under the unitary transformations which arise from the operators constants of motion.

The method previously used to solve Schr\"odinger equation by a unitary transformation for a electron under the influence of a constant magnetic field is used to obtain a non-free Landau electron wave function. The physical meaning of this wave function is discussed based on the conserved properties of the transformed Hamiltonian.

Two systems are studied: the first one involves a charged particle under the influence of a constant electric field, and the second one involves a charged particle under the influence of a constant electromagnetic field. For both systems, it is possible to find time-dependent conserved operators that can be used to derive time-dependent solutions to the complete Schr\"odinger equation. These conserved operators are employed to define the symmetries of the system. An argument of invariance of the wave function under the action of a unitary operator leads to the quantization of resistance and resistivity, in integer multiples of the von Klitzing's constant, for the first and second cases respectively., Comment: 6 pages, 0 figures

The Schr\"odinger equation for an electron under the influence of an electromagnetic field is analyzed based on the conserved operators of the system when the magnetic field is described by Landau's gauge. It is shown that the Lorentz force can be recovered only if two conserved generalized momentum operators are considered: one along the $x$-axis and the second one along $y$-axis; otherwise, the system cannot be fully described. Based on the general solution found, a ground state is built which has the characteristic of having quantized resistivity proportional to integer multiples of the von Klitzing's constant when it is invariant under a unitary transform., Comment: 5 pages, 14 references, no figures

In tasks aiming for long-term returns, planning becomes essential. We study generative modeling for planning with datasets repurposed from offline reinforcement learning. Specifically, we identify temporal consistency in the absence of step-wise rewards as one key technical challenge. We introduce the Latent Plan Transformer (LPT), a novel model that leverages a latent variable to connect a Transformer-based trajectory generator and the final return. LPT can be learned with maximum likelihood estimation on trajectory-return pairs. In learning, posterior sampling of the latent variable naturally integrates sub-trajectories to form a consistent abstraction despite the finite context. At test time, the latent variable is inferred from an expected return before policy execution, realizing the idea of planning as inference. Our experiments demonstrate that LPT can discover improved decisions from sub-optimal trajectories, achieving competitive performance across several benchmarks, including Gym-Mujoco, Franka Kitchen, Maze2D, and Connect Four. It exhibits capabilities in nuanced credit assignments, trajectory stitching, and adaptation to environmental contingencies. These results validate that latent variable inference can be a strong alternative to step-wise reward prompting.

We present a scheme for the generation of a high polarization positron beam with continous wave (CW) bunch structure for the Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Laboratory (JLab). The positrons are created in a high average power conversion target and collected by a CW capture linac and DC solenoid., Comment: 4 pages, 4 figures, 1 table, contributed to IPAC'23

Good advising and support services are instrumental in helping college students--especially Black, Latinx, and low-income learners--stay in school and complete a college credential. The authors of this report analyzed KPI (key performance indicator) data from iPASS institutions and chose five colleges with comparatively strong outcomes among underserved student groups for further study, with the aim of gaining a deeper understanding of how the colleges provide advising and student support both inside and outside of the classroom. The authors describe how the climate of the colleges affects student experiences, and examine practices the colleges have been undertaking to improve outcomes and prioritize equity across racial and socioeconomic subgroups. The authors also detail situational factors, including the pandemic and the fraught political climate at the time of data collection, that their research suggests may have both catalyzed efforts to promote racial equity and slowed progress in improving supports aimed at underserved students. The report is accompanied by a standalone resource guide that describes and provides hyperlinks to useful relevant resources at each of the five colleges. [For the resource guide, see ED626122.]

Swarmalators are oscillators that swarm through space as they synchronize in time. Introduced a few years ago to model many systems which mix synchrony with self-assembly, they remain poorly understood theoretically. Here we obtain the first analytic results on swarmalators moving in two-dimensional (2D) plane by enforcing periodic boundary conditions; this simpler topology allows expressions for order parameters, stabilities, and bifurcations to be derived exactly. We suggest some future directions for swarmalator research and point out some connections to the Kuramoto model and the Vicsek model from active matter; these are intended as a call-to-arms for the sync community and other researchers looking for new problems and puzzles to work on.

In this paper, we introduce a novel algorithm - the Skill-Driven Skill Recombination Algorithm (SDSRA) - an innovative framework that significantly enhances the efficiency of achieving maximum entropy in reinforcement learning tasks. We find that SDSRA achieves faster convergence compared to the traditional Soft Actor-Critic (SAC) algorithm and produces improved policies. By integrating skill-based strategies within the robust Actor-Critic framework, SDSRA demonstrates remarkable adaptability and performance across a wide array of complex and diverse benchmarks.

Rare earth-transition metal ferrimagnets, featuring antiferromagnetically coupled, inequivalent magnetic sublattices, have garnered increasing interest in the burgeoning field of ferrimagnetic spintronics. However, controlling their magnetism with low voltages,a key to reducing power consumption,remains challenging, particularly due to the poorly understood mechanisms underlying bulk perpendicular magnetic anisotropy (PMA). Here, we introduce a method involving voltage-driven hydrogen insertion into interstitial sites between Tb and Co atoms, selectively perturbing the atomic structure and enabling ultra-low-voltage (1.2 V) switching of bulk PMA to in-plane directions in ferrimagnetic films. Combining experimental and theoretical analysis, we find that the anisotropy switching originates from the reorientation of Tb orbital moments induced by the distortion of the crystal field. Initially aligned along Tb-Co bonding directions, the easy magnetization axis undergoes reorientation and switches by 90 deg,as substantiated by ab-initio calculations. Our study not only establishes the groundwork for voltage-controlled magnetic anisotropy in ferrimagnetic films, facilitating the development of electrically-programmable ferrimagnetic spintronics, but also elucidates the atomic origins of PMA in amorphous ferrimagnets, shedding light on a long-standing question in this field., Comment: 4 Figures in manuscript (15 pages); 16 Figures, two Tables in Supplementary materials (21 pages)

Choral singing, a widely practiced form of ensemble singing, lacks comprehensive datasets in the realm of Music Information Retrieval (MIR) research, due to challenges arising from the requirement to curate multitrack recordings. To address this, we devised a novel methodology, leveraging state-of-the-art synthesizers to create and curate quality renditions. The scores were sourced from Choral Public Domain Library(CPDL). This work is done in collaboration with a diverse team of musicians, software engineers and researchers. The resulting dataset, complete with its associated metadata, and methodology is released as part of this work, opening up new avenues for exploration and advancement in the field of singing voice research., Comment: Dataset Link: https://doi.org/10.5281/zenodo.10137883

Given the complex geometry of white matter streamlines, Autoencoders have been proposed as a dimension-reduction tool to simplify the analysis streamlines in a low-dimensional latent spaces. However, despite these recent successes, the majority of encoder architectures only perform dimension reduction on single streamlines as opposed to a full bundle of streamlines. This is a severe limitation of the encoder architecture that completely disregards the global geometric structure of streamlines at the expense of individual fibers. Moreover, the latent space may not be well structured which leads to doubt into their interpretability. In this paper we propose a novel Differentiable Vector Quantized Variational Autoencoder, which are engineered to ingest entire bundles of streamlines as single data-point and provides reliable trustworthy encodings that can then be later used to analyze streamlines in the latent space. Comparisons with several state of the art Autoencoders demonstrate superior performance in both encoding and synthesis., Comment: 5 pages, 4 figures, 1 table

Similar to sperm, where individuals self-organize in space while also striving for coherence in their tail swinging, several natural and engineered systems exhibit the emergence of swarming and synchronization. The arising and interplay of these phenomena have been captured by collectives of hypothetical particles named swarmalators, each possessing a position and a phase whose dynamics are affected reciprocally and also by the space-phase states of their neighbors. In this work, we introduce a solvable model of swarmalators able to move in two-dimensional spaces. We show that several static and active collective states can emerge and derive necessary conditions for each to show up as the model parameters are varied. These conditions elucidate, in some cases, the displaying of multistability among states. Notably, in the active regime, individuals behave chaotically, maintaining spatial correlation under certain conditions, and breaking it under others on what we interpret as a dimensionality transition., Comment: 18 pages, 6 figures

We study the simultaneous decay of global string loops into scalar particles (massless and massive modes) and gravitational waves (GWs). Using field theory simulations in flat space-time of %isolated loops with initial length $\sim 80-1700$ times their core width, we determine the power emitted by a loop into scalar particles, $P_{\varphi}$, and GWs, $P_{\rm GW}$, and characterize the loop-decay timescale as a function of its initial length, energy and angular momentum. We quantify infrared and ultraviolet lattice dependencies of our results. For all type of loops and initial conditions considered, GW emission is always suppressed compared to particles as $P_{\rm GW}/P_{\varphi} \approx \mathcal{O}(10)(v/m_\text{p})^2\ll 1$, where $v$ is the vacuum expectation value associated with string formation. Our results suggest that the GW background from a global string network, such as in dark matter axion scenarios, will be highly suppressed., Comment: 12 pages including Supplemental Material; 9 figures, 3 tables

Reaction-nonlinear diffusion (RND) partial differential equations are a fruitful playground to model the formation of sharp travelling fronts, a fundamental pattern in nature. In this work, we demonstrate the utility and scope of regularisation as a technique to investigate shock-fronted solutions of RND PDEs, using geometric singular perturbation theory (GSPT) as the mathematical framework. In particular, we show that composite regularisations can be used to construct families of monotone shock-fronted travelling waves sweeping out distinct generalised area rules, which interpolate between the equal area and extremal area (i.e. algebraic decay) rules that are well-known in the shockwave literature. We further demonstrate that our RND PDE supports other kinds of shock-fronted solutions, namely, nonmonotone shockwaves as well as shockwaves containing slow tails in the aggregation (negative diffusion) regime. Our analysis blends Melnikov methods -- in both smooth and piecewise-smooth settings -- with GSPT techniques applied to the PDE over distinct spatiotemporal scales. We also consider the spectral stability of these new interpolated shockwaves. Using techniques from geometric spectral stability theory, we determine that our RND PDE admits spectrally stable shock-fronted travelling waves. The multiple-scale nature of the regularised RND PDE continues to play an important role in the analysis of the spatial eigenvalue problem., Comment: 43 pages, 18 figures

One of the challenges of excitonic materials is the accurate determination of the exciton binding energy and bandgap. The difficulty arises from the overlap of the discrete and continuous excitonic absorption at the band edge. Many researches have modeled the shape of the absorption edge of such materials on the Elliott model and its several modifications such as non-parabolic bands, magnetic potentials and electro-hole-polaron interactions. However, exciton binding energies obtained from measured data often vary strongly depending on the chosen model. Here, we propose an alternative and rather simple approach, which has previously been successful in the determination of the optical bandgap of amorphous, direct and indirect semiconductors, based on the bands-fluctuations (BF) model. In this model, the fluctuations due to disorder, temperature or lattice vibrations give rise to the well known exponential distribution of band tail states (Urbach tails). This analysis results in an analytic equation with 5 parameters only. The binding energies and optical bandgaps of GaAs and the family of tri-halide perovskites ($\textrm{MAPbX}_{3}$), $\textrm{X=Br,I,Cl}$, over a wide range of temperatures, are obtained with this model. The results for the bandgap, linewidth and exciton binding energy are in good agreement with previous reports. Moreover, due to the polar nature of perovskites, the obtained binding energies can be compared with the ones computed with a theoretical model for polar materials via a model proposed by Kane et al. In this model, the exciton is surrounded by a cloud of virtual phonons interacting via the Fr$\ddot{\textrm{o}}$lich interaction. As a consequence, the upper bound for the binding energy of the exciton-polaron system is calculated. Coincidentally, these results are in good agreement with the optical constants obtained with the EBF model., Comment: Preprint submitted to: J. Phys. D: Appl. Phys

Swarmalators are phase oscillators that cluster in space, like fireflies flashing on a swarm to attract mates. Interactions between particles, which tend to synchronize their phases and align their motion, decrease with the distance and phase difference between them, coupling the spatial and phase dynamics. In this work, we explore the effects of disorder induced by phase frustration on a system of Swarmalators that move on a one-dimensional ring. Our model is inspired by the well-known Kuramoto-Sakaguchi equations. We find, numerically and analytically, the ordered and disordered states that emerge in the system. The active states, not present in the model without disorder, resemble states found previously in numerical studies for the 2D Swarmalators system. One of these states, in particular, shows similarities to turbulence generated in a flattened media. We show that all ordered states can be generated for any values of the coupling constants by tuning the phase frustration parameters only. Moreover, many of these combinations display multi-stability., Comment: 14 pages, 5 figures, 3 appendices

We study the non-linear dynamics of axion inflation, capturing for the first time the inhomogeneity and full dynamical range during strong backreaction, till the end of inflation. Accounting for inhomogeneous effects leads to a number of new relevant results, compared to spatially homogeneous studies: i) the number of extra efoldings beyond slow roll inflation increases very rapidly with the coupling, ii) oscillations of the inflaton velocity are attenuated, iii) the tachyonic gauge field helicity spectrum is smoothed out (i.e.~the spectral oscillatory features disappear), broadened, and shifted to smaller scales, and iv) the non-tachyonic helicity is excited, reducing the chiral asymmetry, now scale dependent. Our results are expected to impact strongly on the phenomenology and observability of axion inflation, including gravitational wave generation and primordial black hole production., Comment: 9 pages including Supplemental Material; 4 figures. v2: Improved discussion to match submitted version to Physical Review Letters

We perform a detailed comparison of the dynamics of cosmic string loops obtained in cosmological field theory simulations with their expected motion according to the Nambu-Goto action. We demonstrate that these loops follow the trajectories predicted within the NG effective theory except in regions of high curvature where energy is emitted from the loop in the form of massive radiation. This energy loss continues for all the loops studied in this simulation until they self-intersect or become small enough that they annihilate and disappear well before they complete a single oscillation. We comment on the relevance of this investigation to the interpretation of the results from cosmological field theory simulations as well as their extrapolation to a cosmological context., Comment: 17 pages + appendix, 8 figures. Published version

It is pointed out that Hantush's \emph{M} function, commonly used in groundwater pumping modeling, is identical to the function known in the problems of heat conduction in the ground. A modified Hantush function $E$ used in the steady-state solutions for heat conduction that take into account advection due to groundwater flow around borehole heat exchangers (BHEs) is introduced. New exact representation and two-types of approximate expressions for this two-parametric function \emph{E} are presented: one is suitable for small values of a parameter and the other for its large values. High accuracy of the approximate formulae is verified by comparison with the exact values provided by modified Hantush's function $E$ in those validity domains. These analytical expressions can be used for designing complex borehole configurations and may be potentially useful in steady-state analysis of partially penetrated wells (PPWs) in leaky aquifer., Comment: 29 pages, 9 figures