text
stringlengths 6
128k
|
|---|
The dynamics of accreting and outgoing flows around compact objects depends
crucially on the strengths and configurations of the magnetic fields therein,
especially of the large-scale fields that remain coherent beyond turbulence
scales. Possible origins of these large-scale magnetic fields include flux
advection and disc dynamo actions. However, most numerical simulations have to
adopt an initially strong large-scale field rather than allow them to be
self-consistently advected or amplified, due to limited computational
resources. The situation can be partially cured by using sub-grid models where
dynamo actions only reachable at high resolutions are mimicked by artificial
terms in low-resolution simulations. In this work, we couple thin-disc models
with local shearing-box simulation results to facilitate more realistic
sub-grid dynamo implementations. For helical dynamos, detailed spatial profiles
of dynamo drivers inferred from local simulations are used, and the nonlinear
quenching and saturation is constrained by magnetic helicity evolution. In the
inner disc region, saturated fields have dipole configurations and the plasma
$\beta$ reaches $\simeq 0.1$ to $100$, with correlation lengths $\simeq h$ in
the vertical direction and $\simeq 10h$ in the radial direction, where $h$ is
the disc scale height. The dynamo cycle period is $\simeq 40$ orbital time
scale, compatible with previous global simulations. Additionally, we explore
two dynamo mechanisms which do not require a net kinetic helicity and have only
been studied in shearing-box setups. We show that such dynamos are possible in
thin accretion discs, but produce field configurations that are incompatible
with previous results. We discuss implications for future general-relativistic
magnetohydrodynamics simulations.
|
This paper presents the initial development of a robotic additive
manufacturing technology based on ultraviolet (UV)-curable thermoset polymers.
This is designed to allow free-standing printing through partial UV curing and
fiber reinforcement for structural applications. The proposed system integrates
a collaborative robotic manipulator with a custom-built extruder end-effector
designed specifically for printing with UV-curable polymers. The system was
tested using a variety of resin compositions, some reinforced with milled glass
fiber (GF) or fumed silica (FS) and small-scale, 2D and 3D specimens were
printed. Dimensional stability was analyzed for all formulations, showing that
resin containing up to 50 wt% GF or at least 2.8 wt% FS displayed the most
accurate dimensions.
|
We consider a 1D linear Schr{\"o}dinger equation, on a bounded interval, with
Dirichlet boundary conditions and bilinear control. We study its
controllability around the ground state when the linearized system is not
controllable. More precisely, we study to what extent the nonlinear terms of
the expansion can recover the directions lost at the first order.In previous
works, for any positive integer $n$, assumptions have been formulated under
which the quadratic term induces a drift in the nonlinear dynamics, quantified
by the $H^{-n}$ norm of the control. This drift is an obstruction to the
small-time local controllability (STLC) under a smallness assumption on the
controls in regular spaces. In this paper, we prove that for controls small in
less regular spaces, the cubic term can recover the controllability lost at the
linear level, despite the quadratic drift. The proof is inspired by Sussman's
method to prove the sufficiency of the $\mathcal{S}(\theta)$ condition for STLC
of ODEs. However, it uses a different global strategy relying on a new concept
of tangent vector, better adapted to the infinite-dimensional setting of PDEs.
Given a target, we first realize the expected motion along the lost direction
by using control variations for which the cubic term dominates the quadratic
one. Then, we correct the other components exactly, by using a STLC in
projection result, with simultaneous estimates of weak norms of the control.
These estimates ensure that the new error along the lost direction is
negligible, and we conclude with the Brouwer fixed point theorem.
|
Shared space reduces segregation between vehicles and pedestrians and
encourages them to share roads without imposed traffic rules. The behaviour of
road users (RUs) is then controlled by social norms, and interactions are more
versatile than on traditional roads. Autonomous vehicles (AVs) will need to
adapt to these norms to become socially acceptable RUs in shared spaces.
However, to date, there is not much research into pedestrian-vehicle
interaction in shared-space environments, and prior efforts have predominantly
focused on traditional roads and crossing scenarios. We present a video
observation investigating pedestrian reactions to a small, automation-capable
vehicle driven manually in shared spaces based on a long-term naturalistic
driving dataset. We report various pedestrian reactions (from movement
adjustment to prosocial behaviour) and situations pertinent to shared spaces at
this early stage. Insights drawn can serve as a foundation to support future
AVs navigating shared spaces, especially those with a high pedestrian focus.
|
Ab initio study of magnetic superstructures (e.g., magnetic skyrmion) is
indispensable to the research of novel materials but bottlenecked by its
formidable computational cost. For solving the bottleneck problem, we develop a
deep equivariant neural network method (named xDeepH) to represent density
functional theory Hamiltonian $H_\text{DFT}$ as a function of atomic and
magnetic structures and apply neural networks for efficient electronic
structure calculation. Intelligence of neural networks is optimized by
incorporating a priori knowledge about the important locality and symmetry
properties into the method. Particularly, we design a neural-network
architecture fully preserving all equivalent requirements on $H_\text{DFT}$ by
the Euclidean and time-reversal symmetries ($E(3) \times \{I, T\}$), which is
essential to improve method performance. High accuracy (sub-meV error) and good
transferability of xDeepH are shown by systematic experiments on nanotube,
spin-spiral, and Moir\'{e} magnets, and the capability of studying magnetic
skyrmion is also demonstrated. The method could find promising applications in
magnetic materials research and inspire development of deep-learning ab initio
methods.
|
The mechanism of angular momentum transport in protoplanetary disks is
fundamental to understand the distributions of gas and dust in the disks. The
unprecedented, high spatial resolution ALMA observations taken toward HL Tau
and subsequent radiative transfer modeling reveal that a high degree of dust
settling is currently achieved at the outer part of the HL Tau disk. Previous
observations however suggest a high disk accretion rate onto the central star.
This configuration is not necessarily intuitive in the framework of the
conventional viscous disk model, since efficient accretion generally requires a
high level of turbulence, which can suppress dust settling considerably. We
develop a simplified, semi-analytical disk model to examine under what
condition these two properties can be realized in a single model. Recent,
non-ideal MHD simulations are utilized to realistically model the angular
momentum transport both radially via MHD turbulence and vertically via
magnetically induced disk winds. We find that the HL Tau disk configuration can
be reproduced well when disk winds are properly taken into account. While the
resulting disk properties are likely consistent with other observational
results, such an ideal situation can be established only if the plasma $\beta$
at the disk midplane is $\beta_0 \simeq 2 \times 10^4$ under the assumption of
steady accretion. Equivalently, the vertical magnetic flux at 100 au is about
0.2 mG. More detailed modeling is needed to fully identify the origin of the
disk accretion and quantitatively examine plausible mechanisms behind the
observed gap structures in the HL Tau disk.
|
(abridged) The case can be made for a rather universal stellar IMF form that
can be approximated by a two-part power-law function in the stellar regime.
However, there exists a possible hint for a systematic variation with
metallicity. A picture is emerging according to which the binary properties of
very-low-mass stars (VLMSs) and BDs may be fundamentally different from those
of late-type stars implying the probable existence of a discontinuity in the
IMF, but the surveys also appear to suggest the number of BDs per star to be
independent of the physical conditions of current Galactic star formation.
Star-burst clusters and thus globular cluster may, however, have a much larger
abundance of BDs. Very recent advances have allowed the measurement of the
physical upper stellar mass limit, which also appears to be disconcertingly
robust to variations in metallicity. Furthermore, it now appears that star
clusters may be formed such that the most-massive stars just forming terminate
further star-formation within the particular cluster. Populations formed from
many star clusters, composite populations, would then have steeper IMFs (fewer
massive stars per low-mass star) than the simple populations in the constituent
clusters. A near invariant star-cluster mass function implies the maximal
cluster mass to correlate with the galaxy-wide star-formation rate. This then
leads to the result that the composite-stellar IMFs vary in dependence of
galaxy type, with potentially dramatic implications for theories of galaxy
formation and evolution.
|
String field theory for the non-critical NSR string is described. In
particular it gives string field theory for the 2D super-gravity coupled to a
$\hat{c}=1$ matter field. For this purpose double-step pictures changing
operators for the non-critical NSR string are constructed. Analogues of the
critical supersymmetry transformations are written for $D<10$, they form a
closed on-shell algebra, however their action on vertices is defined only for
discrete value of the Liouville momentum. For D=2 this means that spinor
massless field has its superpartner in the NS sector only if its momentum is
fixed.
Starting from string field theory we calculate string amplitudes. These
amplitudes for D=2 have poles which are related with discrete set of primary
fields, namely 2R$\to$2R amplitude has poles corresponding to the n-level NS
excitations with discrete momenta $p_1=n,~~p_2=-1\pm (n+1)$.
|
The aim of this letter is to clarify the relationships between Hawking
radiation and the scattering of light by matter falling into a black hole. To
this end we analyze the S-matrix elements of a model composed of a massive
infalling particle (described by a quantized field) and the radiation field.
These fields are coupled by current-current interactions and propagate in the
Schwarzschild geometry. As long as the photons energy is much smaller than the
mass of the infalling particle, one recovers Hawking radiation since our
S-matrix elements identically reproduce the Bogoliubov coefficients obtained by
treating the trajectory of the infalling particle classically. But after a
brief period, the energy of the `partners' of Hawking photons reaches this mass
and the production of thermal photons through these interactions stops. The
implications of this result are discussed.
|
Let $k$ be a field, $Q$ a finite directed graph, and $kQ$ its path algebra.
Make $kQ$ an $\NN$-graded algebra by assigning each arrow a positive degree.
Let $I$ be a homogeneous ideal in $kQ$ and write $A=kQ/I$. Let $\QGr A$ denote
the quotient of the category of graded right $A$-modules modulo the Serre
subcategory consisting of those graded modules that are the sum of their finite
dimensional submodules. This paper shows there is a finite directed graph $Q'$
with all its arrows placed in degree 1 and a homogeneous ideal $I'\subset kQ'$
such that $\QGr A \equiv \QGr kQ'/I'$. This is an extension of a result
obtained by the author and Gautam Sisodia.
|
The most abundantly produced hadron species in $Si\!-\!Au$ collisions at the
BNL-AGS (nucleons, pions, kaons, antikaons and hyperons) are shown to be in
accord with emission from a thermal resonance gas source of temperature
$T\simeq 110$ MeV and baryochemical potential $\mu_B \simeq 540$ MeV,
corresponding to about 1/3 standard nuclear density. Our analysis takes the
isopin asymmetry of the initial state fully into account.
|
Dark matter may be in the form of non-baryonic structures such as compact
subhalos and boson stars. Structures weighing between asteroid and solar masses
may be discovered via gravitational microlensing, an astronomical probe that
has in the past helped constrain the population of primordial black holes and
baryonic MACHOs. We investigate the non-trivial effect of the size of and
density distribution within these structures on the microlensing signal, and
constrain their populations using the EROS-2 and OGLE-IV surveys. Structures
larger than a solar radius are generally constrained more weakly than
point-like lenses, but stronger constraints may be obtained for structures with
mass distributions that give rise to caustic crossings or produce larger
magnifications.
|
We present an experimental investigation of the statistical properties of
spherical granular particles on an inclined plane that are excited by an
oscillating side-wall. The data is obtained by high-speed imaging and particle
tracking techniques. We identify all particles in the system and link their
positions to form trajectories over long times. Thus, we identify particle
collisions to measure the effective coefficient of restitution and find a broad
distribution of values for the same impact angles. We find that the energy
inelasticity can take on values greater than one, which implies that the
rotational degrees play an important role in energy transfer. We also measure
the distance and the time between collision events in order to directly
determine the distribution of path lengths and the free times. These
distributions are shown to deviate from expected theoretical forms for elastic
spheres, demonstrating the inherent clustering in this system. We describe the
data with a two-parameter fitting function and use it to calculated the mean
free path and collision time. We find that the ratio of these values is
consistent with the average velocity. The velocity distribution are observed to
be strongly non-Gaussian and do not demonstrate any apparent universal
behavior. We report the scaling of the second moment, which corresponds to the
granular temperature, and higher order moments as a function of distance from
the driving wall. Additionally, we measure long time correlation functions in
both space and in the velocities to probe diffusion in a dissipative gas.
|
In recent years, generative adversarial network (GAN)-based image generation
techniques design their generators by stacking up multiple residual blocks. The
residual block generally contains a shortcut, \ie skip connection, which
effectively supports information propagation in the network. In this paper, we
propose a novel shortcut method, called the gated shortcut, which not only
embraces the strength point of the residual block but also further boosts the
GAN performance. More specifically, based on the gating mechanism, the proposed
method leads the residual block to keep (or remove) information that is
relevant (or irrelevant) to the image being generated. To demonstrate that the
proposed method brings significant improvements in the GAN performance, this
paper provides extensive experimental results on the various standard datasets
such as CIFAR-10, CIFAR-100, LSUN, and tiny-ImageNet. Quantitative evaluations
show that the gated shortcut achieves the impressive GAN performance in terms
of Frechet inception distance (FID) and Inception score (IS). For instance, the
proposed method improves the FID and IS scores on the tiny-ImageNet dataset
from 35.13 to 27.90 and 20.23 to 23.42, respectively.
|
Knowledge probing assesses to which degree a language model (LM) has
successfully learned relational knowledge during pre-training. Probing is an
inexpensive way to compare LMs of different sizes and training configurations.
However, previous approaches rely on the objective function used in
pre-training LMs and are thus applicable only to masked or causal LMs. As a
result, comparing different types of LMs becomes impossible. To address this,
we propose an approach that uses an LM's inherent ability to estimate the
log-likelihood of any given textual statement. We carefully design an
evaluation dataset of 7,731 instances (40,916 in a larger variant) from which
we produce alternative statements for each relational fact, one of which is
correct. We then evaluate whether an LM correctly assigns the highest
log-likelihood to the correct statement. Our experimental evaluation of 22
common LMs shows that our proposed framework, BEAR, can effectively probe for
knowledge across different LM types. We release the BEAR datasets and an
open-source framework that implements the probing approach to the research
community to facilitate the evaluation and development of LMs.
|
Highly transmissive ballistic junctions are demonstrated between Nb and the
two-dimensional electron gas formed at an InAs/AlSb heterojunction. A
reproducible fabrication protocol is presented yielding high critical
supercurrent values. Current-voltage characteristics were measured down to 0.4
K and the observed supercurrent behavior was analyzed within a ballistic model
in the clean limit. This investigation allows us to demonstrate an intrinsic
interface transmissivity approaching 90%. The reproducibility of the
fabrication protocol makes it of interest for the experimental study of
InAs-based superconductor-semiconductor hybrid devices.
|
The rational map ansatz of Houghton et al \cite{HMS} is generalised by
allowing the profile function, usually a function of $r$, to depend also on $z$
and $\bar{z}$. It is shown that, within this ansatz, the energies of the lowest
$B=2,3,4$ field configurations of the SU(2) Skyrme model are closer to the
corresponding values of the true solutions of the model than those obtained
within the original rational map ansatz. In particular, we present plots of the
profile functions which do exhibit their dependence on $ z$ and $\bar{z}$.
The obvious generalisation of the ansatz to higher SU(N) models involving the
introduction of more projectors is briefly mentioned.
|
We devise a hierarchy of computational algorithms to enumerate the
microstates of a system comprising N independent, distinguishable particles. An
important challenge is to cope with integers that increase exponentially with
system size, and which very quickly become too large to be addressed by the
computer. A related problem is that the computational time for the most obvious
brute-force method scales exponentially with the system size which makes it
difficult to study the system in the large N limit. Our methods address these
issues in a systematic and hierarchical manner. Our methods are very general
and applicable to a wide class of problems such as harmonic oscillators, free
particles, spin J particles, etc. and a range of other models for which there
are no analytical solutions, for example, a system with single particle energy
spectrum given by {\epsilon}(p,q) = {\epsilon}0 (p^2 + q^4), where p and q are
non-negative integers and so on. Working within the microcanonical ensemble,
our methods enable one to directly monitor the approach to the thermodynamic
limit (N \rightarrow \infty), and in so doing, the equivalence with the
canonical ensemble is made more manifest. Various thermodynamic quantities as a
function of N may be computed using our methods; in this paper, we focus on the
entropy, the chemical potential and the temperature.
|
A fingerprint region of interest (roi) segmentation algorithm is designed to
separate the foreground fingerprint from the background noise. All the learning
based state-of-the-art fingerprint roi segmentation algorithms proposed in the
literature are benchmarked on scenarios when both training and testing
databases consist of fingerprint images acquired from the same sensors.
However, when testing is conducted on a different sensor, the segmentation
performance obtained is often unsatisfactory. As a result, every time a new
fingerprint sensor is used for testing, the fingerprint roi segmentation model
needs to be re-trained with the fingerprint image acquired from the new sensor
and its corresponding manually marked ROI. Manually marking fingerprint ROI is
expensive because firstly, it is time consuming and more importantly, requires
domain expertise. In order to save the human effort in generating annotations
required by state-of-the-art, we propose a fingerprint roi segmentation model
which aligns the features of fingerprint images derived from the unseen sensor
such that they are similar to the ones obtained from the fingerprints whose
ground truth roi masks are available for training. Specifically, we propose a
recurrent adversarial learning based feature alignment network that helps the
fingerprint roi segmentation model to learn sensor-invariant features.
Consequently, sensor-invariant features learnt by the proposed roi segmentation
model help it to achieve improved segmentation performance on fingerprints
acquired from the new sensor. Experiments on publicly available FVC databases
demonstrate the efficacy of the proposed work.
|
Third sound measurements of superfluid $^4$He thin films adsorbed on 10 nm
diameter multiwall carbon nanotubes are used to probe the superfluid onset
temperature as a function of the film thickness, and to study the temperature
dependence of the film compressibility. The nanotubes provide a highly ordered
carbon surface, with layer-by-layer growth of the adsorbed film as shown by
oscillation peaks in the third sound velocity at the completion of the third,
fourth, and fifth atomic layers, arising from oscillations in the
compressibility. In temperature sweeps the third sound velocity at very low
temperatures is found to be linear with temperature, but oscillating between
positive and negative slope depending on the film thickness. Analysis shows
that this can be attributed to a linearly decreasing compressibility of the
film with temperature that appears to hold even near zero temperature. The
superfluid onset temperature is found to be linear in the film thickness, as
predicted by the Kosterlitz-Thouless theory, but the slope is anomalous, a
factor of three smaller than the predicted universal value.
|
Although High Performance Computing (HPC) users understand basic resource
requirements such as the number of CPUs and memory limits, internal
infrastructural utilization data is exclusively leveraged by cluster operators,
who use it to configure batch schedulers. This task is challenging and
increasingly complex due to ever larger cluster scales and heterogeneity of
modern scientific workflows. As a result, HPC systems achieve low utilization
with long job completion times (makespans). To tackle these challenges, we
propose a co-scheduling algorithm based on an adaptive reinforcement learning
algorithm, where application profiling is combined with cluster monitoring. The
resulting cluster scheduler matches resource utilization to application
performance in a fine-grained manner (i.e., operating system level). As opposed
to nominal allocations, we apply decision trees to model applications' actual
resource usage, which are used to estimate how much resource capacity from one
allocation can be co-allocated to additional applications. Our algorithm learns
from incorrect co-scheduling decisions and adapts from changing environment
conditions, and evaluates when such changes cause resource contention that
impacts quality of service metrics such as jobs slowdowns. We integrate our
algorithm in an HPC resource manager that combines Slurm and Mesos for job
scheduling and co-allocation, respectively. Our experimental evaluation
performed in a dedicated cluster executing a mix of four real different
scientific workflows demonstrates improvements on cluster utilization of up to
51% even in high load scenarios, with 55% average queue makespan reductions
under low loads.
|
A comprehensive methodology for inference in vector autoregressions (VARs)
using sign and other structural restrictions is developed. The reduced-form VAR
disturbances are driven by a few common factors and structural identification
restrictions can be incorporated in their loadings in the form of parametric
restrictions. A Gibbs sampler is derived that allows for reduced-form
parameters and structural restrictions to be sampled efficiently in one step. A
key benefit of the proposed approach is that it allows for treating parameter
estimation and structural inference as a joint problem. An additional benefit
is that the methodology can scale to large VARs with multiple shocks, and it
can be extended to accommodate non-linearities, asymmetries, and numerous other
interesting empirical features. The excellent properties of the new algorithm
for inference are explored using synthetic data experiments, and by revisiting
the role of financial factors in economic fluctuations using identification
based on sign restrictions.
|
Cold atom developments suggest the prospect of measuring scaling properties
and long-range fluctuations of continuous phase transitions at
zero-temperature. We discuss the conditions for characterizing the phase
separation of Bose-Einstein condensates of boson atoms in two distinct
hyperfine spin states. The mean-field description breaks down as the system
approaches the transition from the miscible side. An effective spin description
clarifies the ferromagnetic nature of the transition. We show that a difference
in the scattering lengths for the bosons in the same spin state leads to an
effective internal magnetic field. The conditions at which the internal
magnetic field vanishes (i.e., equal values of the like-boson scattering
lengths) is a special point. We show that the long range density fluctuations
are suppressed near that point while the effective spin exhibits the long-range
fluctuations that characterize critical points. The zero-temperature system
exhibits critical opalescence with respect to long wavelength waves of impurity
atoms that interact with the bosons in a spin-dependent manner.
|
We investigate the effect of a small, gauge-invariant mass of the gluon on
the anomalous chromomagnetic moment of quarks (ACM) by perturbative
calculations at one loop level. The mass of the gluon is taken to have been
generated via a topological mass generation mechanism, in which the gluon
acquires a mass through its interaction with an antisymmetric tensor field
$B_{\mu \nu}$. For a small gluon mass $(<10$ MeV), we calculate the ACM at
momentum transfer $q^2=-M_Z^2$. We compare those with the ACM calculated for
the gluon mass arising from a Proca mass term. We find that the ACM of up,
down, strange and charm quarks vary significantly with the gluon mass, while
the ACM of top and bottom quarks show negligible gluon mass dependence. The
mechanism of gluon mass generation is most important for the strange quarks
ACM, but not so much for the other quarks. We also show the results at
$q^2=-m_t^2$. We find that the dependence on gluon mass at $q^2=-m_t^2$ is much
less than at $q^2=-M_Z^2$ for all quarks.
|
Rare-event search experiments located on-surface, such as short-baseline
reactor neutrino experiments, are often limited by muon-induced background
events. Highly efficient muon vetos are essential to reduce the detector
background and to reach the sensitivity goals. We demonstrate the feasibility
of deploying organic plastic scintillators at sub-Kelvin temperatures. For the
NUCLEUS experiment, we developed a cryogenic muon veto equipped with wavelength
shifting fibers and a silicon photo multiplier operating inside a dilution
refrigerator. The achievable compactness of cryostat-internal integration is a
key factor in keeping the muon rate to a minimum while maximizing coverage. The
thermal and light output properties of a plastic scintillation detector were
examined. We report first data on the thermal conductivity and heat capacity of
the polystyrene-based scintillator UPS-923A over a wide range of temperatures
extending below one Kelvin. The light output was measured down to 0.8K and
observed to increase by a factor of 1.61$\pm$0.05 compared to 300K. The
development of an organic plastic scintillation muon veto operating in
sub-Kelvin temperature environments opens new perspectives for rare-event
searches with cryogenic detectors at sites lacking substantial overburden.
|
This article is devoted to Kato's Euler system, which is constructed from
modular unites, and to its image by the dual exponential map (so called Kato's
reciprocity law). The presentation in this article is different form Kato's
original one, and dual exponential map in this article is a modification of
Colmez's construction in his Bourbaki talk.
|
The statistical inference of stochastic block models as emerged as a
mathematicaly principled method for identifying communities inside networks.
Its objective is to find the node partition and the block-to-block adjacency
matrix of maximum likelihood i.e. the one which has most probably generated the
observed network. In practice, in the so-called microcanonical ensemble, it is
frequently assumed that when comparing two models which have the same number
and sizes of communities, the best one is the one of minimum entropy i.e. the
one which can generate the less different networks. In this paper, we show that
there are situations in which the minimum entropy model does not identify the
most significant communities in terms of edge distribution, even though it
generates the observed graph with a higher probability.
|
The short term passenger flow prediction of the urban rail transit system is
of great significance for traffic operation and management. The emerging deep
learning-based models provide effective methods to improve prediction accuracy.
However, most of the existing models mainly predict the passenger flow on
general weekdays or weekends. There are only few studies focusing on predicting
the passenger flow on holidays, which is a significantly challenging task for
traffic management because of its suddenness and irregularity. To this end, we
propose a deep learning-based model named Spatial Temporal Attention Fusion
Network comprising a novel Multi-Graph Attention Network, a Conv-Attention
Block, and Feature Fusion Block for short-term passenger flow prediction on
holidays. The multi-graph attention network is applied to extract the complex
spatial dependencies of passenger flow dynamically and the conv-attention block
is applied to extract the temporal dependencies of passenger flow from global
and local perspectives. Moreover, in addition to the historical passenger flow
data, the social media data, which has been proven that they can effectively
reflect the evolution trend of passenger flow under events, are also fused into
the feature fusion block of STAFN. The STAFN is tested on two large-scale urban
rail transit AFC datasets from China on the New Year holiday, and the
prediction performance of the model are compared with that of several
conventional prediction models. Results demonstrate its better robustness and
advantages among benchmark methods, which can provide overwhelming support for
practical applications of short term passenger flow prediction on holidays.
|
We study the luminosity function and formation rate of long gamma-ray bursts
(GRBs) by using a maximum likelihood method. This is the first time this method
is applied to a well-defined sample of GRBs that is complete in redshift. The
sample is composed of 99 bursts detected by the $Swift$ satellite, 81 of them
with measured redshift and luminosity for a completeness level of $82\%$. We
confirm that a strong redshift evolution in luminosity (with an evolution index
of $\delta=2.22^{+0.32}_{-0.31}$) or in density ($\delta=1.92^{+0.20}_{-0.21}$)
is needed in order to reproduce the observations well. But since the predicted
redshift and luminosity distributions in the two scenarios are very similar, it
is difficult to distinguish between these two kinds of evolutions only on the
basis of the current sample. Furthermore, we also consider an empirical density
case in which the GRB rate density is directly described as a broken power-law
function and the luminosity function is taken to be non-evolving. In this case,
we find that the GRB formation rate rises like $(1+z)^{3.85^{+0.48}_{-0.45}}$
for $z\leq2$ and is proportional to $(1+z)^{-1.07^{+0.98}_{-1.12}}$ for
$z\geq2$. The local GRB rate is $1.49^{+0.63}_{-0.64}$ Gpc$^{-3}$ yr$^{-1}$.
The GRB rate may be consistent with the cosmic star formation rate (SFR) at
$z\leq2$, but shows an enhancement compared to the SFR at $z\geq2$.
|
Heavy-flavour hadrons, i.e. hadrons containing charm or beauty quarks, are
effective probes to test perturbative-QCD (pQCD) calculations, to investigate
the different hadronisation mechanisms, and to study the quark-gluon plasma
(QGP) produced in relativistic heavy-ion collisions at the LHC. Measurements
performed in pp and p-Pb collisions have recently revealed unexpected features
not in line with the expectations based on previous measurements from
$\rm{e^+e^-}$ and ep collisions, showing that charm fragmentation fractions are
not universal. The investigation of initial-state effects such as shadowing in
the collision of a proton with a heavy nucleus is also performed. Measurements
of open heavy-flavour and quarkonia production in Pb-Pb collisions allow for
testing the mechanisms of heavy-quark transport, energy loss, and coalescence
effects during the hadronisation in the presence of a QCD medium. In this
contribution, the most recent results on open heavy-flavour and quarkonia
production in pp, p-Pb, and Pb-Pb collisions obtained by the ALICE
Collaboration are discussed.
|
By partially substituting the tri-valence element La with di-valence element
Sr in $LaOFeAs$, we introduced holes into the system. For the first time, we
successfully synthesized the hole doped new superconductors
$(La_{1-x}Sr_x)OFeAs$. The maximum superconducting transition temperature at
about 25 K was observed at a doping level of x = 0.13. It is evidenced by Hall
effect measurements that the conduction in this type of material is dominated
by hole-like charge carriers, rather than electron-like ones. Together with the
data of the electron doped system $La(O_{1-x}F_x)FeAs$, a generic phase diagram
is depicted and is revealed to be similar to that of the cuprate
superconductors.
|
Identifying phase transitions and classifying phases of matter is central to
understanding the properties and behavior of a broad range of material systems.
In recent years, machine-learning (ML) techniques have been successfully
applied to perform such tasks in a data-driven manner. However, the success of
this approach notwithstanding, we still lack a clear understanding of ML
methods for detecting phase transitions, particularly of those that utilize
neural networks (NNs). In this work, we derive analytical expressions for the
optimal output of three widely used NN-based methods for detecting phase
transitions. These optimal predictions correspond to the results obtained in
the limit of high model capacity. Therefore, in practice they can, for example,
be recovered using sufficiently large, well-trained NNs. The inner workings of
the considered methods are revealed through the explicit dependence of the
optimal output on the input data. By evaluating the analytical expressions, we
can identify phase transitions directly from experimentally accessible data
without training NNs, which makes this procedure favorable in terms of
computation time. Our theoretical results are supported by extensive numerical
simulations covering, e.g., topological, quantum, and many-body localization
phase transitions. We expect similar analyses to provide a deeper understanding
of other classification tasks in condensed matter physics.
|
In this paper we prove that, given s> 0, if E is a subset of R^m with
positive and bounded s-dimensional Hausdorff measure H^s and the principal
values of the s-dimensional signed Riesz transform of H^s|E exist H^s-almost
everywhere in E, then s is integer. Other more general variants of this result
are also proven.
|
Over a century of research into the origin of turbulence in wallbounded shear
flows has resulted in a puzzling picture in which turbulence appears in a
variety of different states competing with laminar background flow. At slightly
higher speeds the situation changes distinctly and the entire flow is
turbulent. Neither the origin of the different states encountered during
transition, nor their front dynamics, let alone the transformation to full
turbulence could be explained to date. Combining experiments, theory and
computer simulations here we uncover the bifurcation scenario organising the
route to fully turbulent pipe flow and explain the front dynamics of the
different states encountered in the process. Key to resolving this problem is
the interpretation of the flow as a bistable system with nonlinear propagation
(advection) of turbulent fronts. These findings bridge the gap between our
understanding of the onset of turbulence and fully turbulent flows.
|
We experimentally demonstrate that the spin-orbit interaction can be utilized
for direct electric-field tuning of the propagation of spin waves in a
single-crystal yttrium iron garnet magnonic waveguide. Magnetoelectric coupling
not due to the spin-orbit interaction, and hence an order of magnitude weaker,
leads to electric-field modification of the spin-wave velocity for waveguide
geometries where the spin-orbit interaction will not contribute. A theory of
the phase shift, validated by the experiment data, shows that, in the exchange
spin wave regime, this electric tuning can have high efficiency. Our findings
point to an important avenue for manipulating spin waves and developing
electrically tunable magnonic devices.
|
There are two types of non(anti-)commutative deformation of D=4, N=1
supersymmetric field theories and D=2, N=2 theories. One is based on the
non-supersymmetric star product and the other is based on the supersymmetric
star product . These deformations cause partial breaking of supersymmetry in
general. In case of supersymmetric star product, the chirality is broken by the
effect of the supersymmetric star product, then it is not clear that lagrangian
or observables including F-terms preserve part of supersymmetry. In this
article, we investigate the ring structure whose product is defined by the
supersymmetric star product. We find the ring whose elements correspond to 1/2
SUSY F-terms. Using this, the 1/2 SUSY invariance of the Wess-Zumino model is
shown easily and directly.
|
It is known that the Perron--Frobenius operators of piecewise expanding
$\mathcal{C}^2$ transformations possess an asymptotic periodicity of densities.
On the other hand, external noise or measurement errors are unavoidable in
practical systems; therefore, all realistic mathematical models should be
regarded as random iterations of transformations. This paper aims to discuss
the effects of randomization on the asymptotic periodicity of densities.
|
Quantum Federated Learning (QFL) is an emerging interdisciplinary field that
merges the principles of Quantum Computing (QC) and Federated Learning (FL),
with the goal of leveraging quantum technologies to enhance privacy, security,
and efficiency in the learning process. Currently, there is no comprehensive
survey for this interdisciplinary field. This review offers a thorough,
holistic examination of QFL. We aim to provide a comprehensive understanding of
the principles, techniques, and emerging applications of QFL. We discuss the
current state of research in this rapidly evolving field, identify challenges
and opportunities associated with integrating these technologies, and outline
future directions and open research questions. We propose a unique taxonomy of
QFL techniques, categorized according to their characteristics and the quantum
techniques employed. As the field of QFL continues to progress, we can
anticipate further breakthroughs and applications across various industries,
driving innovation and addressing challenges related to data privacy, security,
and resource optimization. This review serves as a first-of-its-kind
comprehensive guide for researchers and practitioners interested in
understanding and advancing the field of QFL.
|
An $N$ ${L} \choose {L/2}$-dimensional representation of the periodic
Temperley-Lieb algebra $TL_L(x)$ is presented. It is also a representation of
the cyclic group $Z_N$. We choose $x = 1$ and define a Hamiltonian as a sum of
the generators of the algebra acting in this representation. This Hamiltonian
gives the time evolution operator of a stochastic process. In the finite-size
scaling limit, the spectrum of the Hamiltonian contains representations of the
Virasoro algebra with complex highest weights. The $N = 3$ case is discussed in
detail. One discusses shortly the consequences of the existence of complex
Virasoro representations on the physical properties of the systems.
|
This study investigates the role of topography-induced turbulence, generated
by an idealized urban region, in the transport of firebrands and risk of
spotting. Flight dispersion, deposition, and smoldering state of tens of
thousands of individual mass and size-changing firebrands were investigated in
the atmospheric boundary layer turbulence, which was obtained using Large-eddy
simulations. Firebrands were assumed to be smoldering spherical particles of
Stokes numbers ranging from 30 to 175. Results indicate that the presence of
urban topography significantly affects the firebrand flight behavior, landing
distribution, and risk of spotting. Compared to a case with flat topography,
horizontal dispersions of the smallest size firebrands were significantly
enhanced when urban topography was presented, while the largest firebrands
landed closer to each other and closer to the release point. Consequently, a
notably different and more compact spotting risk map was achieved. Within the
urban boundary layer turbulence, firebrands had shorter flight and smoldering
times in comparison with the flat case. As a result, firebrands landed with
larger temperatures, which contributed to a higher risk of spotting in the
presence of urban topography.
|
I will argue, pace a great many of my contemporaries, that there's something
right about Boltzmann's attempt to ground the second law of thermodynamics in a
suitably amended deterministic time-reversal invariant classical dynamics, and
that in order to appreciate what's right about (what was at least at one time)
Boltzmann's explanatory project, one has to fully apprehend the nature of
microphysical causal structure, time-reversal invariance, and the relationship
between Boltzmann entropy and the work of Rudolf Clausius.
|
We discuss topological rigidity of vector bundles with asymptotically conical
(AC) total spaces of rank greater than 1 with a sufficiently connected link;
our focus will mainly be on ALE (asymptotically locally Euclidean) bundles.
Within the smooth category, we topologically classify all ALE tangent bundles
by showing only 2-sphere, projective plane and open contractible manifolds
admit ALE tangent bundles. We also discuss other interesting topological and
geometric rigidities of ALE vector bundles.
|
We derive an expression for effective gravitational mass for any closed
spacelike 2-surface. This effective gravitational energy is defined directly
through the geometrical quantity of the freely falling 2-surface and thus is
well adapted to intuitive expectation that the gravitational mass should be
determined by the motion of test body moving freely in gravitational field. We
find that this effective gravitational mass has reasonable positive value for a
small sphere in the non-vacuum space-times and can be negative for vacuum case.
Further, this effective gravitational energy is compared with the quasi-local
energy based on the $(2+2)$ formalism of the General Relativity. Although some
gauge freedoms exist, analytic expressions of the quasi-local energy for vacuum
cases are same as the effective gravitational mass. Especially, we see that the
contribution from the cosmological constant is the same in general cases.
|
Based on the detection loophole-free photon key distribution (PKD) compatible
with classical optical systems, an optical key distribution (OKD) protocol is
presented for unconditionally secured cryptography in fiber-optic
communications networks using addressable continuous phase basis, where each
communication channel is composed of paired transmission lines. The
unconditional security in OKD lies in quantum superposition between the paired
lines of each channel. The continuous phase basis in OKD can be applied for
one-time-pad optical cryptography in networks, whose network address capacity
is dependent upon the robustness of OKD to channel noises.
|
The self-trapped state (STS) of interlayer exciton (IX) has been aroused
enormous interesting owing to their significant impact on the fundamental
properties of the van derWaals heterostructures (vdWHs). Nevertheless, the
microscopic mechanisms of STS are still controversial. Herein, we study the
corrections of the binding energies of the IXs due to the exciton-interface
optical phonon coupling in four kinds of vdWHs and find that these IXs are in
the STS for the appropriate ratio of the electron and hole effective masses. We
show that these STSs could be classified into the type I with the increasing
binding energy in the tens of meV range, which are very agreement with the
red-shift of the IXs spectra in experiments, and the type II with the
decreasing binding energy, which provides a possible explanation for the
blue-shift and broad linewidth of the IXs spectra in the low temperature.
Moreover, these two types of self-trapped IXs could be transformed into each
other by adjusting the structural parameters of vdWHs. These results not only
provide an in-depth understanding for the self-trapped mechanism of IX, but
also shed light on the modulations of IXs in vdWHs.
|
Data on hadron multiplicities from inelastic proton-proton interactions in
the energy range of the NICA collider have been compiled. The compilation
includes recent results from the NA61/SHINE and NA49 experiments at the CERN
SPS accelerator. New parameterizations for excitation functions of mean
multiplicities $\left<\pi^{\pm}\right>$, $\left<K^{\pm}\right>$,
$\left<K^{0}_S\right>$, $\left<\Lambda\right>$, $\left<p\right>$,
$\left<\bar{p}\right>$ are obtained in the region of collision energies
$3<\sqrt{s_{NN}}<31$ GeV. The energy dependence of the particle yields, as well
as variation of rapidity and transverse momentum distributions are discussed. A
standalone algorithm for hadron phase space generation in pp collisions is
suggested and compared to model predictions using an example of the PHQMD
generator.
|
Relativistic plasma with radiation at thermodynamic equilibrium is ageneral
system of interest in astrophysics and high energy physics. We develop a new
self-consistent quasi-particle model for such a system to take account of
collective behaviour of plasma andthermodynamic properties are derived. It is
applied to electrodynamic plasma and quark gluon plasma and compared with
existing results.
|
The treatment of cloud structure in numerical weather and climate models is
often greatly simplified to make them computationally affordable. Here we
propose to correct the European Centre for Medium-Range Weather Forecasts 1D
radiation scheme ecRad for 3D cloud effects using computationally cheap neural
networks. 3D cloud effects are learned as the difference between ecRad's fast
1D Tripleclouds solver that neglects them and its 3D SPARTACUS (SPeedy
Algorithm for Radiative TrAnsfer through CloUd Sides) solver that includes them
but is about five times more computationally expensive. With typical errors
between 20 % and 30 % of the 3D signal, neural networks improve Tripleclouds'
accuracy for about 1 % increase in runtime. Thus, rather than emulating the
whole of SPARTACUS, we keep Tripleclouds unchanged for cloud-free parts of the
atmosphere and 3D-correct it elsewhere. The focus on the comparably small 3D
correction instead of the entire signal allows us to improve predictions
significantly if we assume a similar signal-to-noise ratio for both.
|
Effective non-Hermitian Hamiltonians describing decaying systems are derived
and analyzed in connection with the occurrence of possible Hilbert space
partitioning, resulting in a confinement of the dynamics. In some cases, this
fact can be interpreted properly as Zeno effect or Zeno dynamics, according to
the dimension of the subspace one focuses on; in some other cases, the
interpretation is more complicated and traceable back to a mix of Zeno
phenomena and lack of resonance. Depending on the complex phases of the
diagonal terms of the Hamiltonian, the system reacts in different ways,
requiring larger moduli for the dynamical confinement to occur when the complex
phase is close to $\pi/2$.
|
Type II WLFs have weak Balmer line emission and no Balmer jump. We carried
out a set of radiative hydrodynamic simulations to understand how the hydrogen
radiative losses vary with the electron beam parameters and more specifically
with the low energy cutoff. Our results have revealed that for low energy
beams, the excess flare Lyman emission diminishes with increasing low energy
cutoff as the energy deposited into the top chromosphere is low compared to the
energy deposited into the deeper layers. Some Balmer excess emission is always
present and is driven primarily by direct heating from the beam with a minor
contribution from Lyman continuum backwarming. The absence of Lyman excess
emission in electron beam models with high low energy cutoff is a prominent
spectral signature of type II WLFs.
|
Network modeling is a key enabler to achieve efficient network operation in
future self-driving Software-Defined Networks. However, we still lack
functional network models able to produce accurate predictions of Key
Performance Indicators (KPI) such as delay, jitter or loss at limited cost. In
this paper we propose RouteNet, a novel network model based on Graph Neural
Network (GNN) that is able to understand the complex relationship between
topology, routing, and input traffic to produce accurate estimates of the
per-source/destination per-packet delay distribution and loss. RouteNet
leverages the ability of GNNs to learn and model graph-structured information
and as a result, our model is able to generalize over arbitrary topologies,
routing schemes and traffic intensity. In our evaluation, we show that RouteNet
is able to predict accurately the delay distribution (mean delay and jitter)
and loss even in topologies, routing and traffic unseen in the training (worst
case MRE=15.4%). Also, we present several use cases where we leverage the KPI
predictions of our GNN model to achieve efficient routing optimization and
network planning.
|
This paper studies differential graded modules and representations up to
homotopy of Lie $n$-algebroids, for general $n\in\mathbb{N}$. The adjoint and
coadjoint modules are described, and the corresponding split versions of the
adjoint and coadjoint representations up to homotopy are explained. In
particular, the case of Lie 2-algebroids is analysed in detail. The
compatibility of a Poisson bracket with the homological vector field of a Lie
$n$-algebroid is shown to be equivalent to a morphism from the coadjoint module
to the adjoint module, leading to an alternative characterisation of
non-degeneracy of higher Poisson structures. Moreover, the Weil algebra of a
Lie $n$-algebroid is computed explicitly in terms of splittings, and
representations up to homotopy of Lie $n$-algebroids are used to encode
decomposed VB-Lie $n$-algebroid structures on double vector bundles.
|
The electronic implications of strain in graphene can be captured at low
energies by means of pseudovector potentials which can give rise to
pseudomagnetic fields. These strain-induced vector potentials arise from the
local perturbation to the electronic hopping amplitudes in a tight-binding
framework. Here we complete the standard description of the strain-induced
vector potential, which accounts only for the hopping perturbation, with the
explicit inclusion of the lattice deformations or, equivalently, the
deformation of the Brillouin zone. These corrections are linear in strain and
are different at each of the strained, inequivalent Dirac points, and hence are
equally necessary to identify the precise magnitude of the vector potential.
This effect can be relevant in scenarios of inhomogeneous strain profiles,
where electronic motion depends on the amount of overlap among the local Fermi
surfaces. In particular, it affects the pseudomagnetic field distribution
induced by inhomogeneous strain configurations, and can lead to new
opportunities in tailoring the optimal strain fields for certain desired
functionalities.
|
Symmetric minima of surface potential energy of a nanocatalyst act as
nucleation sites for chirally selective initial growth of single walled carbon
tubes at low temperatures. The nucleation sites are sites of maximum
coordination number of the adsorbed carbon. We show this using the five fold
symmetry of a pentagonal pyramid of an icosahedron. Initial zigzag structure
from nucleation sites results in formation of hexagons and pentagons that
result in anomalous cap formation. Possible cap lift off mechanism is
discussed.
|
Ceci est un rapport sur l'article "A finiteness theorem for zero-cycles over
p-adic fields" (arXiv:math/0605165) de Shuji Saito et Kanetomo Sato.
-----
This is a survey on the paper "A finiteness theorem for zero-cycles over
p-adic fields" (arXiv:math/0605165) by Shuji Saito and Kanetomo Sato.
|
We show that the black hole in the x-ray binary Cygnus X-1 was formed in situ
and did not receive an energetic trigger from a nearby supernova. The
progenitor of the black hole had an initial mass greater than 40 solar masses
and during the collapse to form the ~10 solar mass black hole of Cygnus X-1,
the upper limit for the mass that could have been suddenly ejected is ~1 solar
mass, much less than the mass ejected in a supernova. The observations suggest
that high-mass stellar black holes may form promptly, when massive stars
disappear silently.
|
It has recently been noted that the relative prevalence of the various kinds
of epistasis varies along an adaptive walk. This has been explained as a result
of mean regression in NK model fitness landscapes. Here we show that this
phenomenon occurs quite generally in fitness landscapes. We propose a simple
and general explanation for this phenomemon, confirming the role of mean
regression. We provide support for this explanation with simulations, and
discuss the empirical relevance of our findings.
|
Majorana fermions are predicted to play a crucial role in condensed matter
realizations of topological quantum computation. These heretofore undiscovered
quasiparticles have been predicted to exist at the cores of vortex excitations
in topological superconductors and in heterostructures of superconductors and
materials with strong spin-orbit coupling. In this work we examine topological
insulators with bulk s-wave superconductivity in the presence of a
vortex-lattice generated by a perpendicular magnetic field. Using
self-consistent Bogoliubov-de Gennes, calculations we confirm that beyond the
semi-classical, weak-pairing limit that the Majorana vortex states appear as
the chemical potential is tuned from either side of the band edge so long as
the density of states is sufficient for superconductivity to form. Further, we
demonstrate that the previously predicted vortex phase transition survives
beyond the semi-classical limit. At chemical potential values smaller than the
critical chemical potential, the vortex lattice modes hybridize within the top
and bottom surfaces giving rise to a dispersive low-energy mid-gap band. As the
chemical potential is increased, the Majorana states become more localized
within a single surface but spread into the bulk toward the opposite surface.
Eventually, when the chemical potential is sufficiently high in the bulk bands,
the Majorana modes can tunnel between surfaces and eventually a critical point
is reached at which modes on opposite surfaces can freely tunnel and annihilate
leading to the topological phase transition previously studied in the work of
Hosur et al.
|
We propose a simple interpolation-based method for the efficient
approximation of gradients in neural ODE models. We compare it with the reverse
dynamic method (known in the literature as "adjoint method") to train neural
ODEs on classification, density estimation, and inference approximation tasks.
We also propose a theoretical justification of our approach using logarithmic
norm formalism. As a result, our method allows faster model training than the
reverse dynamic method that was confirmed and validated by extensive numerical
experiments for several standard benchmarks.
|
Coral reefs are under increasing threat from the impacts of climate change.
Whilst current restoration approaches are effective, they require significant
human involvement and equipment, and have limited deployment scale. Harvesting
wild coral spawn from mass spawning events, rearing them to the larval stage
and releasing the larvae onto degraded reefs is an emerging solution for reef
restoration known as coral reseeding. This paper presents a reconfigurable
autonomous surface vehicle system that can eliminate risky diving, cover
greater areas with coral larvae, has a sensory suite for additional data
measurement, and requires minimal non-technical expert training. A key feature
is an on-board real-time benthic substrate classification model that predicts
when to release larvae to increase settlement rate and ultimately,
survivability. The presented robot design is reconfigurable, light weight,
scalable, and easy to transport. Results from restoration deployments at Lizard
Island demonstrate improved coral larvae release onto appropriate coral
substrate, while also achieving 21.8 times more area coverage compared to
manual methods.
|
Mobile Agents (MAs) represent a distributed computing technology that
promises to address the scalability problems of centralized network management.
A critical issue that will affect the wider adoption of MA paradigm in
management applications is the development of MA Platforms (MAPs) expressly
oriented to distributed management. However, most of available platforms impose
considerable burden on network and system resources and also lack of essential
functionality. In this paper, we discuss the design considerations and
implementation details of a complete MAP research prototype that sufficiently
addresses all the aforementioned issues. Our MAP has been implemented in Java
and tailored for network and systems management applications.
|
We extend our model for the pion, which we used previously to calculate its
diagonal structure function, to the off-forward case. The imaginary part of the
off-forward gamma* pi -> gamma* pi scattering amplitude is evaluated in the
chiral limit (m_pi=0) and related to the twist-two and twist-three generalised
parton distributions, H, H3, H3tilde. Non-perturbative effects, linked to the
size of the pion and still preserving gauge invariance, are included.
Remarkable new relations between H, H3 and H3tilde are obtained and discussed.
|
Recent breakthroughs in Deep Learning (DL) applications have made DL models a
key component in almost every modern computing system. The increased popularity
of DL applications deployed on a wide-spectrum of platforms have resulted in a
plethora of design challenges related to the constraints introduced by the
hardware itself. What is the latency or energy cost for an inference made by a
Deep Neural Network (DNN)? Is it possible to predict this latency or energy
consumption before a model is trained? If yes, how can machine learners take
advantage of these models to design the hardware-optimal DNN for deployment?
From lengthening battery life of mobile devices to reducing the runtime
requirements of DL models executing in the cloud, the answers to these
questions have drawn significant attention.
One cannot optimize what isn't properly modeled. Therefore, it is important
to understand the hardware efficiency of DL models during serving for making an
inference, before even training the model. This key observation has motivated
the use of predictive models to capture the hardware performance or energy
efficiency of DL applications. Furthermore, DL practitioners are challenged
with the task of designing the DNN model, i.e., of tuning the hyper-parameters
of the DNN architecture, while optimizing for both accuracy of the DL model and
its hardware efficiency. Therefore, state-of-the-art methodologies have
proposed hardware-aware hyper-parameter optimization techniques. In this paper,
we provide a comprehensive assessment of state-of-the-art work and selected
results on the hardware-aware modeling and optimization for DL applications. We
also highlight several open questions that are poised to give rise to novel
hardware-aware designs in the next few years, as DL applications continue to
significantly impact associated hardware systems and platforms.
|
An important question about resonance extraction is how much resonance poles
and residues extracted from data depend on a model used for the extraction, and
on the precision of data. We address this question with the dynamical
coupled-channel (DCC) model developed in Excited Baryon Analysis Center (EBAC)
at JLab. We focus on the P11 pi-N scattering. We examine the model-dependence
of the poles by varying parameters to a large extent within the EBAC-DCC model.
We find that two poles associated with the Roper resonance are fairly stable
against the variation. We also develop a model with a bare nucleon, thereby
examining the stability of the Roper poles against different analytic structure
of the P11 amplitude below pi-N threshold. We again find a good stability of
the Roper poles.
|
We study a theoretical model for the magnetothermal conductivity of a
spin-1/2 ladder with low exchange coupling ($J\ll\Theta_D$) subject to a strong
magnetic field $B$. Our theory for the thermal transport accounts for the
contribution of spinons coupled to lattice phonon modes in the one-dimensional
lattice. We employ a mapping of the ladder Hamiltonian onto an XXZ spin-chain
in a weaker effective field B_{eff}=B-B_{0}$, where $B_{0}=(B_{c1}+B_{c2})/2$
corresponds to half-filling of the spinon band. This provides a low-energy
theory for the spinon excitations and their coupling to the phonons. The
coupling of acoustic longitudinal phonons to spinons gives rise to
hybridization of spinons and phonons, and provides an enhanced $B$-dependant
scattering of phonons on spinons. Using a memory matrix approach, we show that
the interplay between several scattering mechanisms, namely: umklapp, disorder
and phonon-spinon collisions, dominates the relaxation of heat current. This
yields magnetothermal effects that are qualitatively consistent with the
thermal conductivity measurements in the spin-1/2 ladder compound ${\rm
Br_4(C_5H_{12}N)_2}$ (BPCB).
|
Automated plot generation for games enhances the player's experience by
providing rich and immersive narrative experience that adapts to the player's
actions. Traditional approaches adopt a symbolic narrative planning method
which limits the scale and complexity of the generated plot by requiring
extensive knowledge engineering work. Recent advancements use Large Language
Models (LLMs) to drive the behavior of virtual characters, allowing plots to
emerge from interactions between characters and their environments. However,
the emergent nature of such decentralized plot generation makes it difficult
for authors to direct plot progression. We propose a novel plot creation
workflow that mediates between a writer's authorial intent and the emergent
behaviors from LLM-driven character simulation, through a novel authorial
structure called "abstract acts". The writers define high-level plot outlines
that are later transformed into concrete character action sequences via an
LLM-based narrative planning process, based on the game world state. The
process creates "living stories" that dynamically adapt to various game world
states, resulting in narratives co-created by the author, character simulation,
and player. We present StoryVerse as a proof-of-concept system to demonstrate
this plot creation workflow. We showcase the versatility of our approach with
examples in different stories and game environments.
|
We present a unifying view on various statistical estimation techniques
including penalization, variational and thresholding methods. These estimators
will be analyzed in the context of statistical linear inverse problems
including nonparametric and change point regression, and high dimensional
linear models as examples. Our approach reveals many seemingly unrelated
estimation schemes as special instances of a general class of variational
multiscale estimators, named MIND (MultIscale Nemirovskii--Dantzig). These
estimators result from minimizing certain regularization functionals under
convex constraints that can be seen as multiple statistical tests for local
hypotheses.
For computational purposes, we recast MIND in terms of simpler unconstraint
optimization problems via Lagrangian penalization as well as Fenchel duality.
Performance of several MINDs is demonstrated on numerical examples.
|
We present updated results on the search for a neutrino signal from the core
of the Earth and of the Sun induced by Weakly Interacting Massive Particles
(WIMPs). In this paper we concentrate on neutralinos as WIMP candidates. The
971 and 642 events used respectively for the search from the Sun and from the
Earth are compatible with the background of atmospheric neutrinos. Consequently
we calculate flux limits for various search cones around these sources. Limits
as a function of the neutralino mass are given and compared to the
supersymmetric (SUSY) models.
|
Objective: A numerical 3D model of the human trunk was developed to study the
biomechanical effects of lumbar belts used to treat low back pain. Methods:
This model was taken from trunk radiographies of a person and simplified so as
to make a parametric study by varying morphological parameters of the patient,
characteristic parameters of the lumbar belt and mechanical parameters of body
and finally to determine the parameters influencing the effects of low back
pain when of wearing the lumbar belt. The loading of lumbar belt is modelled by
Laplace's law. These results were compared with clinical study. Results: All
the results of this parametric study showed that the choice of belt is very
important depending on the patient's morphology. Surprisingly, the therapeutic
treatment is not influenced by the mechanical characteristics of the body
structures except the mechanical properties of intervertebral discs.
Discussion: The numerical model can serve as a basis for more in-depth studies
concerning the analysis of efficiency of lumbar belts in low back pain. In
order to study the impact of the belt's architecture, the pressure applied to
the trunk modelled by Laplace's law could be improved. This model could also be
used as the basis for a study of the impact of the belt over a period of
wearing time. Indeed, the clinical study shows that movement has an important
impact on the distribution of pressure applied by the belt.
|
Two emerging areas of research, attosecond and nanoscale physics, have
recently started to merge. Attosecond physics deals with phenomena occurring
when ultrashort laser pulses, with duration on the femto- and sub-femtosecond
time scales, interact with atoms, molecules or solids. The laser-induced
electron dynamics occurs natively on a timescale down to a few hundred or even
tens of attoseconds (1 attosecond=1 as=10$^{-18}$ s), which is of the order of
the optical field cycle. For comparison, the revolution of an electron on a
$1s$ orbital of a hydrogen atom is $\sim152$ as. On the other hand, the second
topic involves the manipulation and engineering of mesoscopic systems, such as
solids, metals, and dielectrics, with nanometric precision. Although
nano-engineering is a vast and well-established research field on its own, the
combination with intense laser physics is relatively recent. We present a
comprehensive theoretical overview of the tools to tackle and understand the
physics that takes place when short and intense laser pulses interact with
nanosystems, such as metallic and dielectric nanostructures. In particular, we
elucidate how the spatially inhomogeneous laser-induced fields at a nanometer
scale modify the laser-driven electron dynamics. Consequently, this has an
important impact on pivotal processes such as above-threshold ionization and
high-order harmonic generation. The deep understanding of the coupled dynamics
between these spatially inhomogeneous fields and matter configures a promising
way to new avenues of research and applications. Thanks to the maturity that
attosecond physics has reached, together with the tremendous advance in
material engineering and manipulation techniques, the age of atto-nano physics
has begun, but it is still in an incipient stage.
|
Despite the increasing adoption of Field-Programmable Gate Arrays (FPGAs) in
compute clouds, there remains a significant gap in programming tools and
abstractions which can leverage network-connected, cloud-scale, multi-die FPGAs
to generate accelerators with high frequency and throughput. To this end, we
propose TAPA-CS, a task-parallel dataflow programming framework which
automatically partitions and compiles a large design across a cluster of FPGAs
with no additional user effort while achieving high frequency and throughput.
TAPA-CS has three main contributions. First, it is an open-source framework
which allows users to leverage virtually "unlimited" accelerator fabric,
high-bandwidth memory (HBM), and on-chip memory, by abstracting away the
underlying hardware. This reduces the user's programming burden to a logical
one, enabling software developers and researchers with limited FPGA domain
knowledge to deploy larger designs than possible earlier. Second, given as
input a large design, TAPA-CS automatically partitions the design to map to
multiple FPGAs, while ensuring congestion control, resource balancing, and
overlapping of communication and computation. Third, TAPA-CS couples
coarse-grained floorplanning with automated interconnect pipelining at the
inter- and intra-FPGA levels to ensure high frequency. We have tested TAPA-CS
on our multi-FPGA testbed where the FPGAs communicate through a high-speed
100Gbps Ethernet infrastructure. We have evaluated the performance and
scalability of our tool on designs, including systolic-array based
convolutional neural networks (CNNs), graph processing workloads such as page
rank, stencil applications like the Dilate kernel, and K-nearest neighbors
(KNN). TAPA-CS has the potential to accelerate development of increasingly
complex and large designs on the low power and reconfigurable FPGAs.
|
We investigate cosmological inflationary scenarios from a
gravitoelectromagnetic theory. Our work is formulated in the light of a
recently introduced geometrical Weyl-Invariant scalar-tensor theory of gravity,
where the nature of both the electromagnetic potential and the inflaton field
is attributed to the space-time geometry. We obtain a Harrison-Zeldovich power
spectrum for quantum fluctuations of the inflaton field. In our model the
electromagnetic fields have also a nearly scale invariant power spectrum for a
power-law inflation. We found that the the seed magnetic fields have a nearly
scale invariant power spectrum and generate in the present times cosmic
magnetic fields of the order $\lesssim 10^{9}$ gauss, in good agreement with
CMB observations.
|
A famous result due to Grothendieck asserts that every continuous linear
operator from $\ell_{1}$ to $\ell_{2}$ is absolutely $(1,1)$-summing. If
$n\geq2,$ however, it is very simple to prove that every continuous $n$-linear
operator from $\ell_{1}\times...\times\ell_{1}$ to $\ell_{2}$ is absolutely
$(1;1,...,1) $-summing, and even absolutely $(\frac{2}% {n};1,...,1)
$-summing$.$ In this note we deal with the following problem:
Given a positive integer $n\geq2$, what is the best constant $g_{n}>0$ so
that every $n$-linear operator from $\ell_{1}\times...\times\ell_{1}$ to
$\ell_{2}$ is absolutely $(g_{n};1,...,1) $-summing?
We prove that $g_{n}\leq\frac{2}{n+1}$ and also obtain an optimal improvement
of previous recent results (due to Heinz Juenk $\mathit{et}$ $\mathit{al}$,
Geraldo Botelho $\mathit{et}$ $\mathit{al}$ and Dumitru Popa) on inclusion
theorems for absolutely summing multilinear operators.
|
This paper introduces a novel perception framework that has the ability to
identify and track objects in autonomous vehicle's field of view. The proposed
algorithms don't require any training for achieving this goal. The framework
makes use of ego-vehicle's pose estimation and a KD-Tree-based segmentation
algorithm to generate object clusters. In turn, using a VFH technique, the
geometry of each identified object cluster is translated into a multi-modal PDF
and a motion model is initiated with every new object cluster for the purpose
of robust spatio-temporal tracking. The methodology further uses statistical
properties of high-dimensional probability density functions and Bayesian
motion model estimates to identify and track objects from frame to frame. The
effectiveness of the methodology is tested on a KITTI dataset. The results show
that the median tracking accuracy is around 91% with an end-to-end
computational time of 153 milliseconds
|
Knife safety in the kitchen is essential for preventing accidents or injuries
with an emphasis on proper handling, maintenance, and storage methods. This
research presents a comparative analysis of three YOLO models, YOLOv5, YOLOv8,
and YOLOv10, to detect the hazards involved in handling knife, concentrating
mainly on ensuring fingers are curled while holding items to be cut and that
hands should only be in contact with knife handle avoiding the blade.
Precision, recall, F-score, and normalized confusion matrix are used to
evaluate the performance of the models. The results indicate that YOLOv5
performed better than the other two models in identifying the hazard of
ensuring hands only touch the blade, while YOLOv8 excelled in detecting the
hazard of curled fingers while holding items. YOLOv5 and YOLOv8 performed
almost identically in recognizing classes such as hand, knife, and vegetable,
whereas YOLOv5, YOLOv8, and YOLOv10 accurately identified the cutting board.
This paper provides insights into the advantages and shortcomings of these
models in real-world settings. Moreover, by detailing the optimization of YOLO
architectures for safe knife handling, this study promotes the development of
increased accuracy and efficiency in safety surveillance systems.
|
The structure of the observable algebra ${\mathfrak O}_{\Lambda}$ of lattice
QCD in the Hamiltonian approach is investigated. As was shown earlier,
${\mathfrak O}_{\Lambda}$ is isomorphic to the tensor product of a gluonic
$C^{*}$-subalgebra, built from gauge fields and a hadronic subalgebra
constructed from gauge invariant combinations of quark fields. The gluonic
component is isomorphic to a standard CCR algebra over the group manifold
SU(3). The structure of the hadronic part, as presented in terms of a number of
generators and relations, is studied in detail. It is shown that its
irreducible representations are classified by triality. Using this, it is
proved that the hadronic algebra is isomorphic to the commutant of the triality
operator in the enveloping algebra of the Lie super algebra ${\rm sl(1/n)}$
(factorized by a certain ideal).
|
We use nearly two decades of helioseismic data obtained from the GONG
(2002-2020) and HMI (2010-2020) ring-diagram pipelines to examine the temporal
variations of the properties of individual equatorial Rossby modes with
azimuthal orders in the range $6 \le m \le 10$. We find that the mode
parameters obtained from GONG and HMI are consistent during the data
overlapping period of 2010-2020. The power and the frequency of each mode
exhibit significant temporal variations over the full observing period. Using
the GONG data during solar cycles 23 and 24, we find that the mode power
averaged over $6 \le m \le 10$ shows a positive correlation with the sunspot
number ($0.42$), while the averaged frequency shift is anti-correlated
($-0.91$). The anti-correlation between the average mode power and frequency
shift is $-0.44$.
|
Transverse instability of a bunched beam is investigated with synchrotron
oscillations, space charge, and resistive wall wakefield taken into account.
Boxcar model is used for all-round analysis, and Gaussian distribution is
invoked for details. The beam spectrum, instability growth rate and effects of
chromaticity are studied in a wide range of parameters, both with head-tail and
collective bunch interactions included. Effects of the internal bunch
oscillations on the of collective instabilities is investigated thoroughly.
Landau damping caused by the space charge tune spread is discussed, and the
instability thresholds of different modes of Gaussian bunch are estimated.
|
The image matching field has been witnessing a continuous emergence of novel
learnable feature matching techniques, with ever-improving performance on
conventional benchmarks. However, our investigation shows that despite these
gains, their potential for real-world applications is restricted by their
limited generalization capabilities to novel image domains. In this paper, we
introduce OmniGlue, the first learnable image matcher that is designed with
generalization as a core principle. OmniGlue leverages broad knowledge from a
vision foundation model to guide the feature matching process, boosting
generalization to domains not seen at training time. Additionally, we propose a
novel keypoint position-guided attention mechanism which disentangles spatial
and appearance information, leading to enhanced matching descriptors. We
perform comprehensive experiments on a suite of $7$ datasets with varied image
domains, including scene-level, object-centric and aerial images. OmniGlue's
novel components lead to relative gains on unseen domains of $20.9\%$ with
respect to a directly comparable reference model, while also outperforming the
recent LightGlue method by $9.5\%$ relatively.Code and model can be found at
https://hwjiang1510.github.io/OmniGlue
|
Personalized recommendations are one of the most widely deployed machine
learning (ML) workload serviced from cloud datacenters. As such, architectural
solutions for high-performance recommendation inference have recently been the
target of several prior literatures. Unfortunately, little have been explored
and understood regarding the training side of this emerging ML workload. In
this paper, we first perform a detailed workload characterization study on
training recommendations, root-causing sparse embedding layer training as one
of the most significant performance bottlenecks. We then propose our
algorithm-architecture co-design called Tensor Casting, which enables the
development of a generic accelerator architecture for tensor gather-scatter
that encompasses all the key primitives of training embedding layers. When
prototyped on a real CPU-GPU system, Tensor Casting provides 1.9-21x
improvements in training throughput compared to state-of-the-art approaches.
|
An ortho-polygon visibility representation of an $n$-vertex embedded graph
$G$ (OPVR of $G$) is an embedding-preserving drawing of $G$ that maps every
vertex to a distinct orthogonal polygon and each edge to a vertical or
horizontal visibility between its end-vertices. The vertex complexity of an
OPVR of $G$ is the minimum $k$ such that every polygon has at most $k$ reflex
corners. We present polynomial time algorithms that test whether $G$ has an
OPVR and, if so, compute one of minimum vertex complexity. We argue that the
existence and the vertex complexity of an OPVR of $G$ are related to its number
of crossings per edge and to its connectivity. More precisely, we prove that if
$G$ has at most one crossing per edge (i.e., $G$ is a 1-plane graph), an OPVR
of $G$ always exists while this may not be the case if two crossings per edge
are allowed. Also, if $G$ is a 3-connected 1-plane graph, we can compute an
OPVR of $G$ whose vertex complexity is bounded by a constant in $O(n)$ time.
However, if $G$ is a 2-connected 1-plane graph, the vertex complexity of any
OPVR of $G$ may be $\Omega(n)$. In contrast, we describe a family of
2-connected 1-plane graphs for which an embedding that guarantees constant
vertex complexity can be computed in $O(n)$ time. Finally, we present the
results of an experimental study on the vertex complexity of ortho-polygon
visibility representations of 1-plane graphs.
|
Active region 11029 was a small, highly flare-productive solar active region
observed at a time of extremely low solar activity. The region produced only
small flares: the largest of the $>70$ Geostationary Observational
Environmental Satellite (GOES) events for the region has a peak 1--$8{\AA}$
flux of $2.2\times 10^{-6} {\rm W} {\rm m}^{-2}$ (GOES C2.2). The
background-subtracted GOES peak-flux distribution suggests departure from
power-law behavior above $10^{-6} {\rm W} {\rm m}^{-2}$, and a Bayesian model
comparison strongly favors a power-law plus rollover model for the distribution
over a simple power-law model. The departure from the power law is attributed
to this small active region having a finite amount of energy. The rate of
flaring in the region varies with time, becoming very high for two days
coinciding with the onset of an increase in complexity of the photospheric
magnetic field. The observed waiting-time distribution for events is consistent
with a piecewise-constant Poisson model. These results present challenges for
models of flare statistics and of energy balance in solar active regions.
|
Many black holes (BHs) detected by the Laser Interferometer
Gravitational-wave Observatory (LIGO) and the Virgo detectors are multiple
times more massive than those in X-ray binaries. One possibility is that some
BBHs merge within a few Schwarzschild radii of a supermassive black hole
(SMBH), such that the gravitational waves (GWs) are highly redshifted, causing
the mass inferred from GW signals to appear higher than the real mass. The
difficulty of this scenario lies in the delivery of BBH to such a small
distance to a SMBH. Here we revisit the theoretical models for the migration of
compact objects (COs) in the accretion discs of active galactic nuclei (AGNs).
We find that when the accretion rate is high so that the disc is best described
by the slim disc model, the COs in the disc could migrate to a radius close to
the innermost stable circular orbit (ISCO) and be trapped there for the
remaining lifetime of the AGN. The exact trapping radius coincides with the
transition region between the sub- and super-Keplerian rotation of the slim
disc. We call this region "the last migration trap" because inside it COs can
no longer be trapped for a long time. We pinpoint the parameter space which
could induce such a trap and we estimate that the last migration trap
contributes a few per cent of the LIGO/Virgo events. Our result implies that a
couple of BBHs discovered by LIGO/Virgo could have smaller intrinsic masses.
|
We give estimates for the torsion in the Postnikov sections $\tau_{[1, n]}
S^0$ of the sphere spectrum, and show that the $p$-localization is annihilated
by $p^{n/(2p-2) + O(1)}$. This leads to explicit bounds on the exponents of the
kernel and cokernel of the Hurewicz map $\pi_*(X) \to H_*(X; \mathbb{Z})$ for a
connective spectrum $X$. Such bounds were first considered by Arlettaz,
although our estimates are tighter and we prove that they are the best possible
up to a constant factor. As applications, we sharpen existing bounds on the
orders of $k$-invariants in a connective spectrum, sharpen bounds on the
unstable Hurewicz of an infinite loop space, and prove an exponent theorem for
the equivariant stable stems.
|
We present an analysis of survey observations targeting the leading L4
Jupiter Trojan cloud near opposition using the wide-field Suprime-Cam CCD
camera on the 8.2 m Subaru Telescope. The survey covered about 38 deg$^2$ of
sky and imaged 147 fields spread across a wide region of the L4 cloud. Each
field was imaged in both the $g'$ and the $i'$ band, allowing for the
measurement of $g-i$ color. We detected 557 Trojans in the observed fields,
ranging in absolute magnitude from $H=10.0$ to $H = 20.3$. We fit the total
magnitude distribution to a broken power law and show that the power-law slope
rolls over from $0.45\pm 0.05$ to $0.36^{+0.05}_{-0.09}$ at a break magnitude
of $H_{b}=14.93^{+0.73}_{-0.88}$. Combining the best-fit magnitude distribution
of faint objects from our survey with an analysis of the magnitude distribution
of bright objects listed in the Minor Planet Center catalog, we obtain the
absolute magnitude distribution of Trojans over the entire range from $H=7.2$
to $H=16.4$. We show that the $g-i$ color of Trojans decreases with increasing
magnitude. In the context of the less-red and red color populations, as
classified in Wong et al. 2014 using photometric and spectroscopic data, we
demonstrate that the observed trend in color for the faint Trojans is
consistent with the expected trend derived from extrapolation of the best-fit
color population magnitude distributions for bright catalogued Trojans. This
indicates a steady increase in the relative number of less-red objects with
decreasing size. Finally, we interpret our results using collisional modeling
and propose several hypotheses for the color evolution of the Jupiter Trojan
population.
|
The maximum sustainable amplitude, so-called wave breaking limit, of a
nonlinear plasma wave in arbitrary mass ratio warm plasmas is obtained in the
non-relativistic regime. Using the method of Sagdeev potential a general wave
breaking formula is derived by taking into account the dynamics of both the
species having finite temperature. It is found, that the maximum amplitude of
the plasma wave decreases monotonically with the increase in temperature and
mildly increases with increase in mass ratio.
|
We consider the problem of estimating covariance and precision matrices, and
their associated discriminant coefficients, from normal data when the rank of
the covariance matrix is strictly smaller than its dimension and the available
sample size. Using unbiased risk estimation, we construct novel estimators by
minimizing upper bounds on the difference in risk over several classes. Our
proposal estimates are empirically demonstrated to offer substantial
improvement over classical approaches.
|
Knapik et al. introduced the safety restriction which constrains both the
types and syntax of the production rules defining a higher-order recursion
scheme. This restriction gives rise to an equi-expressivity result between
order-n pushdown automata and order-n safe recursion schemes, when such devices
are used as tree generators. We show that the typing constraint of safety,
called homogeneity, is unnecessary in the sense that imposing the syntactic
restriction alone is sufficient to prove the equi-expressivity result for
trees.
|
A nonperturbative, purely numerical, solution of the reduced Rayleigh
equation for the scattering of p- and s-polarized light from a dielectric film
with a two-dimensional randomly rough surface deposited on a planar metallic
substrate, has been carried out. It is found that satellite peaks are present
in the angular dependence of the elements of the mean differential reflection
coefficient in addition to an enhanced backscattering peak. This result
resolves a conflict between the results of earlier approximate theoretical
studies of scattering from this system.
|
We study the nonlinear Schr\"odinger equation for systems of $N$ orthonormal
functions. We prove the existence of ground states for all $N$ when the
exponent $p$ of the non linearity is not too large, and for an infinite
sequence $N_j$ tending to infinity in the whole range of possible $p$'s, in
dimensions $d\geq1$. This allows us to prove that translational symmetry is
broken for a quantum crystal in the Kohn-Sham model with a large Dirac exchange
constant.
|
We study the renormalization problem for the Hartree--Fock approximation of
the $O(N)-$invariant $\phi^4$ model in the symmetric phase and show how to
systematically improve the corresponding diagrammatic resummation to achieve
the correct renormalization properties of the effective field equations,
including Renormalization--Group invariance with the one--loop beta function.
These new Hartree--Fock dynamics is still of mean field type but includes
memory effects which are generically nonlocal also in space.
|
Common observations of the unpredictability of human behavior and the
influence of one question on the answer to another suggest social science
experiments are probabilistic and may be mutually incompatible with one
another, characteristics attributed to quantum mechanics (as distinguished from
classical mechanics). This paper examines this superficial similarity in depth
using the Foulis-Randall Operational Statistics language. In contradistinction
to physics, social science deals with complex, open systems for which the set
of possible experiments is unknowable and outcome interference is a graded
phenomenon resulting from the ways the human brain processes information. It is
concluded that social science is, in some ways, "less classical" than quantum
mechanics, but that generalized "quantum" structures may provide appropriate
descriptions of social science experiments. Specific challenges to extending
"quantum" structures to social science are identified.
|
Machine learning and artificial neural networks (ANNs) have increasingly
become integral to data analysis research in astrophysics due to the growing
demand for fast calculations resulting from the abundance of observational
data. Simultaneously, neutron stars and black holes have been extensively
examined within modified theories of gravity since they enable the exploration
of the strong field regime of gravity. In this study, we employ ANNs to develop
a surrogate model for a numerical iterative method to solve the structure
equations of NSs within a specific 4D Einstein-Gauss-Bonnet gravity framework.
We have trained highly accurate surrogate models, each corresponding to one of
twenty realistic EoSs. The resulting ANN models predict the mass and radius of
individual NS models between 10 and 100 times faster than the numerical solver.
In the case of batch processing, we demonstrated that the speed up is several
orders of magnitude higher. We have trained additional models where the radius
is predicted for specific masses. Here, the speed up is considerably higher
since the original numerical code that constructs the equilibrium models would
have to do additional iterations to find a model with a specific mass. Our ANN
models can be used to speed up Bayesian inference, where the mass and radius of
equilibrium models in this theory of gravity are required.
|
The learning speed of feed-forward neural networks is notoriously slow and
has presented a bottleneck in deep learning applications for several decades.
For instance, gradient-based learning algorithms, which are used extensively to
train neural networks, tend to work slowly when all of the network parameters
must be iteratively tuned. To counter this, both researchers and practitioners
have tried introducing randomness to reduce the learning requirement. Based on
the original construction of Igelnik and Pao, single layer neural-networks with
random input-to-hidden layer weights and biases have seen success in practice,
but the necessary theoretical justification is lacking. In this paper, we begin
to fill this theoretical gap. We provide a (corrected) rigorous proof that the
Igelnik and Pao construction is a universal approximator for continuous
functions on compact domains, with approximation error decaying asymptotically
like $O(1/\sqrt{n})$ for the number $n$ of network nodes. We then extend this
result to the non-asymptotic setting, proving that one can achieve any desired
approximation error with high probability provided $n$ is sufficiently large.
We further adapt this randomized neural network architecture to approximate
functions on smooth, compact submanifolds of Euclidean space, providing
theoretical guarantees in both the asymptotic and non-asymptotic forms.
Finally, we illustrate our results on manifolds with numerical experiments.
|
Let $n\ge2$ and $\mathcal{L}=-\mathrm{div}(A\nabla\cdot)$ be an elliptic
operator on $\mathbb{R}^n$. Given an exterior Lipschitz domain $\Omega$, let
$\mathcal{L}_D$ and $\mathcal{L}_N$ be the elliptic operators $\mathcal{L}$ on
$\Omega$ subject to the Dirichlet and the Neumann boundary {conditions},
respectively. For the Neumann operator, we show that the reverse inequality
$\|\mathcal{L}_N^{1/2}f\|_{L^p(\Omega)} \le C\|\nabla f\|_{L^p(\Omega)}$ holds
true for any $p\in(1,\infty)$. For the Dirichlet operator, it was known that
the Riesz operator $\nabla \mathcal{L}_D^{-1/2}$ is not bounded for $p>2$ and
$p\ge n$, even if $\mathcal{L}=-\Delta$ being the Laplace operator. Suppose
that $A$ are CMO coefficients or VMO coefficients satisfying certain
perturbation property, and $\partial\Omega$ is $C^1$, we prove that for $p>2$
and $p\in [n,\infty)$, it holds $$
\inf_{\phi\in\mathcal{A}^p_0(\Omega)}\left\|\nabla
f-\nabla\phi\right\|_{L^p(\Omega)}\le C\left\|\mathcal{L}^{1/2}_D
f\right\|_{L^p(\Omega)} $$ for $f\in \dot{W}^{1,p}_0(\Omega)$. Here
$\mathcal{A}^p_0(\Omega)=\{f\in \dot{W}^{1,p}_0(\Omega):\,\mathcal{L}_Df=0\}$
is a non-trivial subspace generated by harmonic function in $\Omega$ with zero
boundary value.
|
The mass radius is a fundamental property of the proton that so far has not
been determined from experiment. Here we show that the mass radius of the
proton can be rigorously defined through the formfactor of the trace of the
energy-momentum tensor (EMT) of QCD in the weak gravitational field
approximation, as appropriate for this problem. We then demonstrate that the
scale anomaly of QCD enables the extraction of the formfactor of the trace of
the EMT from the data on threshold photoproduction of $J/\psi$ and $\Upsilon$
quarkonia, and use the recent GlueX Collaboration data to extract the r.m.s.
mass radius of the proton ${\rm R_m = 0.55 \pm 0.03 \ fm}$. The extracted mass
radius is significantly smaller than the charge radius of the proton ${\rm R_C
= 0.8409 \pm 0.0004 \ fm}$. We attribute this difference to the interplay of
asymptotic freedom and spontaneous breaking of chiral symmetry in QCD, and
outline future measurements needed to determine the mass radius more precisely.
|
Self-assembly of ordered nanometer-scale patterns is interesting in itself,
but its practical value depends on the ability to predict and control pattern
formation. In this paper we demonstrate theoretically and numerically that
engineering of extrinsic as well as intrinsic substrate geometry may provide
such a controllable ordering mechanism for block copolymers films. We develop
an effective two-dimensional model of thin films of striped-phase diblock
copolymers on general curved substrates. The model is obtained as an expansion
in the film thickness and thus takes the third dimension into account, which
crucially allows us to predict the preferred orientations even in the absence
of intrinsic curvature. We determine the minimum-energy textures on several
curved surfaces and arrive at a general principle for using substrate curvature
as an ordering field, namely that the stripes will tend to align along
directions of maximal curvature.
|
Inspired by precision tests of the Standard Model in future lepton colliders,
the numerical analysis of the following scattering processes, $e^+e^-
\rightarrow Z h^0\gamma$ and $e^+e^- \rightarrow Z H^0 \gamma$, are carried at
the tree level including all possible diagrams in the two-Higgs-doublet model
(2HDM). This model has many free parameters, but the parameters which take part
in the scattering amplitudes of these two processes are primarily the mixing
angle parameter, $s_{\beta-\alpha}$, and the masses of the neutral Higgs
bosons, ($h^0, H^0$). Therefore, measuring the production rates of $Z
h^0\gamma$ and $Z H^0 \gamma$ final states open another test for the scalar
sectors of the 2HDM. The numerical analysis is performed under the current
experimental constraints. The production rates and the asymmetry in the
forward-backward direction are presented as a function of the center-of-mass
energy covering the future lepton colliders. The unpolarized cross-section gets
up to 6.19 (4.86) fb at $\sqrt{s} = 350$ (500) GeV and 0.164 (0.157) fb at
$\sqrt{s} = 350$ (500) GeV for $e^+e^- \rightarrow Z h^0\gamma$ and $e^+e^-
\rightarrow Z H^0 \gamma$, respectively. The polarization of the incoming
$e^+e^-$ beams are studied for various configurations, and it enhances the
cross-section by a factor of 1.78 in both processes for
$P_{e^+,e^-}=(+0.6,-0.8)$.
|
We report the detection of 239 trans-Neptunian Objects discovered through the
on-going New Horizons survey for distant minor bodies being performed with the
Hyper Suprime-Cam mosaic imager on the Subaru Telescope. These objects were
discovered in images acquired with either the r2 or the recently commissioned
EB-gri filter using shift and stack routines. Due to the extremely high stellar
density of the search region down stream of the spacecraft, new machine
learning techniques had to be developed to manage the extremely high false
positive rate of bogus candidates produced from the shift and stack routines.
We report discoveries as faint as r2$\sim26.5$. We highlight an overabundance
of objects found at heliocentric distances $R\gtrsim70$~au compared to
expectations from modelling of the known outer Solar System. If confirmed,
these objects betray the presence of a heretofore unrecognized abundance of
distant objects that can help explain a number of other observations that
otherwise remain at odds with the known Kuiper Belt, including detections of
serendipitous stellar occultations, and recent results from the Student Dust
Counter on-board the New Horizons spacecraft.
|
By H\"ormander's $L^2$-m\'ethode, we study some operators in the Hilbert
space of weight $L^2(\mathbb{C}, \mathrm{e}^{-|z|^2})$. We prove in each case
of operator the existence of its inverse which is also a bounded operator.
|
Subsets and Splits
Filtered Text Samples
Retrieves 100 samples of text containing the specific phrase "You are a helpful assistant", providing limited insight into the dataset.
Helpful Assistant Text Samples
Returns a limited set of rows containing the phrase 'helpful assistant' in the text, providing basic filtering of relevant entries.