Dilawer Singh, Antoni J. Wojcik, Timothy D. Wilkinson
Phase-only computer-generated holography (CGH) seeks a phase pattern for a spatial light modulator (SLM) whose propagated optical field reproduces a desired intensity distribution. In the far-field (Fraunhofer) regime, optical propagation reduces to a Fourier transform, such that each hologram pixel contributes to the entire reconstructed intensity distribution. When restricted to phase-only modulation, intensity must be shaped through global phase interference effects, making the inverse mapping from target intensity to phase highly non-linear and sensitive to local minima. We present a proof-of-concept physics-in-the-loop approach in which a transformer maps a target intensity image to a phase-only SLM field and is trained end-to-end through exact FFT-based propagation embedded directly within optimization. We further observe that patch tokenization strongly shapes the optimization geometry: coarse tokenization acts as an implicit spectral regularizer that stabilizes training and suppresses checkerboard-like attractors, while finer tokenization increases spatial degrees of freedom but benefits from curriculum or hierarchical refinement. Despite training on limited primitives and a single digit class (only digit 6), the learned generator exhibits out-of-distribution (OOD) generalization to unseen digits and hand-drawn target patterns. These results suggest that transformer architectures, whose self-attention enables global token interactions, are a natural fit for far-field holography and provide a viable foundation for scalable physics-grounded hologram generation.
Weerayut Buaphet, Thanh-Nhi Nguyen, Risa Kondo
et al.
Social media data has been of interest to Natural Language Processing (NLP) practitioners for over a decade, because of its richness in information, but also challenges for automatic processing. Since language use is more informal, spontaneous, and adheres to many different sociolects, the performance of NLP models often deteriorates. One solution to this problem is to transform data to a standard variant before processing it, which is also called lexical normalization. There has been a wide variety of benchmarks and models proposed for this task. The MultiLexNorm benchmark proposed to unify these efforts, but it consists almost solely of languages from the Indo-European language family in the Latin script. Hence, we propose an extension to MultiLexNorm, which covers 5 Asian languages from different language families in 4 different scripts. We show that the previous state-of-the-art model performs worse on the new languages and propose a new architecture based on Large Language Models (LLMs), which shows more robust performance. Finally, we analyze remaining errors, revealing future directions for this task.
Recently, Large Language Models (LLMs) have dominated much of the artificial intelligence scene with their ability to process and generate natural languages. However, the majority of LLM research and development remains English-centric, leaving low-resource languages such as those in the Southeast Asian (SEA) region under-represented. To address this representation gap, we introduce Llama-SEA-LION-v3-8B-IT and Gemma-SEA-LION-v3-9B-IT, two cutting-edge multilingual LLMs designed for SEA languages. The SEA-LION family of LLMs supports 11 SEA languages, namely English, Chinese, Indonesian, Vietnamese, Malay, Thai, Burmese, Lao, Filipino, Tamil, and Khmer. Our work leverages large-scale multilingual continued pre-training with a comprehensive post-training regime involving multiple stages of instruction fine-tuning, alignment, and model merging. Evaluation results on multilingual benchmarks indicate that our models achieve state-of-the-art performance across LLMs supporting SEA languages. We open-source the models to benefit the wider SEA community.
Not all nations on earth have previously been surveyed accurately enough to know for certain which peak is the national highpoint, the highest peak in the country. Knowledge of these peaks is important for understanding the physical geography of these countries in terms of natural resource availability, watershed management, and tourism potential. For this study, ground surveys were conducted between 2018-2025 with modern professional surveying equipment, including differential GPS units and Abney levels, to accurately determine the national highpoints in five African and Asian countries where uncertainty existed. New national highpoints were determined for Saudi Arabia (Jabal Ferwa), Uzbekistan (Alpomish), Gambia (Sare Firasu Hill), Guinea-Bissau (Mt Ronde), and Togo (Mt Atilakoutse). Elevations were measured with sub-meter vertical accuracy for candidate peaks in Saudi Arabia, Gambia, Guinea-Bissau, and Togo. Relative elevations were measured between contender peaks in Uzbekistan with sufficient accuracy to determine the highpoint.
This is an expository note explaining how the geometric notions of local connectedness and properness are related to the $Σ$-type and $Π$-type constructors of dependent type theory.
Arianna Burzacchi, Lidia Rossi, Tommaso Agasisti
et al.
Despite its crucial role in students' daily lives, commuting time remains an underexplored dimension in higher education research. To address this gap, this study focuses on challenges that students face in urban environments and investigates the impact of commuting time on the academic performance of first-year bachelor students of Politecnico di Milano, Italy. This research employs an innovative two-step methodology. In the initial phase, machine learning algorithms trained on privacy-preserving GPS data from anonymous users are used to construct accessibility maps to the university and to obtain an estimate of students' commuting times. In the subsequent phase, authors utilize polynomial linear mixed-effects models and investigate the factors influencing students' academic performance, with a particular emphasis on commuting time. Notably, this investigation incorporates a causal framework, which enables the establishment of causal relationships between commuting time and academic outcomes. The findings underscore the significant impact of travel time on students' performance and may support policies and implications aiming at improving students' educational experience in metropolitan areas. The study's innovation lies both in its exploration of a relatively uncharted factor and the novel methodologies applied in both phases.
Stefan Wagner, Marvin Muñoz Barón, Davide Falessi
et al.
In the short period since the release of ChatGPT, large language models (LLMs) have changed the software engineering research landscape. While there are numerous opportunities to use LLMs for supporting research or software engineering tasks, solid science needs rigorous empirical evaluations. However, so far, there are no specific guidelines for conducting and assessing studies involving LLMs in software engineering research. Our focus is on empirical studies that either use LLMs as part of the research process or studies that evaluate existing or new tools that are based on LLMs. This paper contributes the first set of holistic guidelines for such studies. Our goal is to start a discussion in the software engineering research community to reach a common understanding of our standards for high-quality empirical studies involving LLMs.
Wei Qi Leong, Jian Gang Ngui, Yosephine Susanto
et al.
The rapid development of Large Language Models (LLMs) and the emergence of novel abilities with scale have necessitated the construction of holistic, diverse and challenging benchmarks such as HELM and BIG-bench. However, at the moment, most of these benchmarks focus only on performance in English and evaluations that include Southeast Asian (SEA) languages are few in number. We therefore propose BHASA, a holistic linguistic and cultural evaluation suite for LLMs in SEA languages. It comprises three components: (1) a NLP benchmark covering eight tasks across Natural Language Understanding (NLU), Generation (NLG) and Reasoning (NLR) tasks, (2) LINDSEA, a linguistic diagnostic toolkit that spans the gamut of linguistic phenomena including syntax, semantics and pragmatics, and (3) a cultural diagnostics dataset that probes for both cultural representation and sensitivity. For this preliminary effort, we implement the NLP benchmark only for Indonesian, Vietnamese, Thai and Tamil, and we only include Indonesian and Tamil for LINDSEA and the cultural diagnostics dataset. As GPT-4 is purportedly one of the best-performing multilingual LLMs at the moment, we use it as a yardstick to gauge the capabilities of LLMs in the context of SEA languages. Our initial experiments on GPT-4 with BHASA find it lacking in various aspects of linguistic capabilities, cultural representation and sensitivity in the targeted SEA languages. BHASA is a work in progress and will continue to be improved and expanded in the future. The repository for this paper can be found at: https://github.com/aisingapore/BHASA
Here we present high resolution (15-24 au) observations of CO isotopologue lines from the Molecules with ALMA on Planet-forming Scales (MAPS) ALMA Large Program. Our analysis employs $^{13}$CO and C$^{18}$O ($J$=2-1), (1-0), and C$^{17}$O (1-0) line observations of five protoplanetary disks. We retrieve CO gas density distributions, using three independent methods: (1) a thermo-chemical modeling framework based on the CO data, the broadband spectral energy distribution, and the mm-continuum emission; (2) an empirical temperature distribution based on optically thick CO lines; and (3) a direct fit to the C$^{17}$O hyperfine lines. Results from these methods generally show excellent agreement. The CO gas column density profiles of the five disks show significant variations in the absolute value and the radial shape. Assuming a gas-to-dust mass ratio of 100, all five disks have a global CO-to-H$_2$ abundance of 10-100 times lower than the ISM ratio. The CO gas distributions between 150-400 au match well with models of viscous disks, supporting the long-standing theory. CO gas gaps appear to be correlated with continuum gap locations, but some deep continuum gaps do not have corresponding CO gaps. The relative depths of CO and dust gaps are generally consistent with predictions of planet-disk interactions, but some CO gaps are 5-10 times shallower than predictions based on dust gaps. This paper is part of the MAPS special issue of the Astrophysical Journal Supplement.
Viviana V. Guzmán, Jennifer B. Bergner, Charles J. Law
et al.
Small organic molecules, such as C2H, HCN, and H2CO, are tracers of the C, N, and O budget in protoplanetary disks. We present high angular resolution (10-50 au) observations of C2H, HCN, and H2CO lines in five protoplanetary disks from the Molecules with ALMA at Planet-forming Scales (MAPS) ALMA Large Program. We derive column density and excitation temperature profiles for HCN and C2H, and find that the HCN emission arises in a temperate (20-30 K) layer in the disk, while C2H is present in relatively warmer (20-60 K) layers. In the case of HD 163296, we find a decrease in column density for HCN and C2H inside one of the dust gaps near 83 au, where a planet has been proposed to be located. We derive H2CO column density profiles assuming temperatures between 20 and 50 K, and find slightly higher column densities in the colder disks around T Tauri stars than around Herbig Ae stars. The H2CO column densities rise near the location of the CO snowline and/or millimeter dust edge, suggesting an efficient release of H2CO ices in the outer disk. Finally, we find that the inner 50 au of these disks are rich in organic species, with abundances relative to water that are similar to cometary values. Comets could therefore deliver water and key organics to future planets in these disks, similar to what might have happened here on Earth. This paper is part of the MAPS special issue of the Astrophysical Journal Supplement.
The Asian giant hornet (AGH) appeared in Washington State appears to have a potential danger of bioinvasion. Washington State has collected public photos and videos of detected insects for verification and further investigation. In this paper, we analyze AGH using data analysis,statistics, discrete mathematics, and deep learning techniques to process the data to controlAGH spreading.First, we visualize the geographical distribution of insects in Washington State. Then we investigate insect populations to varying months of the year and different days of a month.Third, we employ wavelet analysis to examine the periodic spread of AGH. Fourth, we apply ordinary differential equations to examine AGH numbers at the different natural growthrate and reaction speed and output the potential propagation coefficient. Next, we leverage cellular automaton combined with the potential propagation coefficient to simulate the geographical spread under changing potential propagation. To update the model, we use delayed differential equations to simulate human intervention. We use the time difference between detection time and submission time to determine the unit of time to delay time. After that, we construct a lightweight CNN called SqueezeNet and assess its classification performance. We then relate several non-reference image quality metrics, including NIQE, image gradient, entropy, contrast, and TOPSIS to judge the cause of misclassification. Furthermore, we build a Random Forest classifier to identify positive and negative samples based on image qualities only. We also display the feature importance and conduct an error analysis. Besides, we present sensitivity analysis to verify the robustness of our models. Finally, we show the strengths and weaknesses of our model and derives the conclusions.
We have used the capability of the MUSE instrument to explore the impact of stellar feedback at large scales in Haro 11, a galaxy under extreme starburst condition and one of the first galaxies where Lyman continuum (LyC) has been detected. Using Ha, [OIII] and [OI] emission lines from deep MUSE observations, we have constructed a sequence of velocity-dependent maps of the Ha emission, the state of the ionised gas and a tracer of fast shocks. These allowed us to investigate the ionisation structure of the galaxy in 50 kms^2 bins over a velocity range of -400 to 350 kms. The ionised gas in Haro 11 is assembled by a rich arrangement of structures, such as superbubbles, filaments, arcs and galactic ionised channels, whose appearances change drastically with velocity. The central star forming knots and the star forming dusty arm are the main engines that power the strong mechanical feedback in this galaxy, although with different impact on the ionisation structure. Haro 11 appears to leak LyC radiation in many directions. We found evidence of a kpc-scale fragmented superbubble, that may have cleared galactic-scale channels in the ISM. Additionally, the southwestern hemisphere is highly ionised in all velocities, hinting at a density bound scenario. A compact kpc-scale structure of lowly ionised gas coincides with the diffuse Lya emission and the presence of fast shocks. Finally, we find evidence that a significant fraction of the ionised gas mass may escape the gravitational potential of the galaxy.
This study describes a detailed analysis of museum visitors' decoding process as they used a visualization designed to support exploration of a large, complex dataset. Quantitative and qualitative analyses revealed that it took, on average, 43 seconds for visitors to decode enough of the visualization to see patterns and relationships in the underlying data represented, and 54 seconds to arrive at their first correct data interpretation. Furthermore, visitors decoded throughout and not only upon initial use of the visualization. The study analyzed think-aloud data to identify issues visitors had mapping the visual representations to their intended referents, examine why they occurred, and consider if and how these decoding issues were resolved. The paper also describes how multiple visual encodings both helped and hindered decoding and concludes with implications on the design and adaptation of visualizations for informal science learning venues.
Fuzzy cognitive maps (FCMs) model feedback causal relations in interwoven webs of causality and policy variables. FCMs are fuzzy signed directed graphs that allow degrees of causal influence and event occurrence. Such causal models can simulate a wide range of policy scenarios and decision processes. Their directed loops or cycles directly model causal feedback. Their nonlinear dynamics permit forward-chaining inference from input causes and policy options to output effects. Users can add detailed dynamics and feedback links directly to the causal model or infer them with statistical learning laws. Users can fuse or combine FCMs from multiple experts by weighting and adding the underlying fuzzy edge matrices and do so recursively if needed. The combined FCM tends to better represent domain knowledge as the expert sample size increases if the expert sample approximates a random sample. Many causal models use more restrictive directed acyclic graphs (DAGs) and Bayesian probabilities. DAGs do not model causal feedback because they do not contain closed loops. Combining DAGs also tends to produce cycles and thus tends not to produce a new DAG. Combining DAGs tends to produce a FCM. FCM causal influence is also transitive whereas probabilistic causal influence is not transitive in general. Overall: FCMs trade the numerical precision of probabilistic DAGs for pattern prediction, faster and scalable computation, ease of combination, and richer feedback representation. We show how FCMs can apply to problems of public support for insurgency and terrorism and to US-China conflict relations in Graham Allison's Thucydides-trap framework. The appendix gives the textual justification of the Thucydides-trap FCM. It also extends our earlier theorem [Osoba-Kosko2017] to a more general result that shows the transitive and total causal influence that upstream concept nodes exert on downstream nodes.
This work is motivated by the question of whether there are spaces $X$ for which the Farber-Grant symmetric topological complexity $TC^S(X)$ differs from the Basabe-González-Rudyak-Tamaki symmetric topological complexity $TC^Σ(X)$. It is known that, for a projective space $RP^m$, $TC^S(RP^m)$ captures, with a few potentially exceptional cases, the Euclidean embedding dimension of $RP^m$. We now show that, for all $m\geq1$, $TC^Σ(RP^m)$ is characterized as the smallest positive integer $n$ for which there is a symmetric $\mathbb{Z}_2$-biequivariant map $S^m\times S^m\to S^n$ with a "monoidal" behavior on the diagonal. This result thus lies at the core of the efforts in the 1970's to characterize the embedding dimension of real projective spaces in terms of the existence of symmetric axial maps. Together with Nakaoka's description of the cohomology ring of symmetric squares, this allows us to compute both $TC$ numbers in the case of $RP^{2^e}$ for $e\geq1$. In particular, this leaves the torus $S^1\times S^1$ as the only closed surface whose symmetric (symmetrized) $TC^S$ ($TC^Σ$) -invariant is currently unknown.
Hisashi Hayakawa, Kiyomi Iwahashi, Harufumi Tamazawa
et al.
The magnetic storm around 1859 September 2, caused by so-called Carrington flare, was the most intense in the history of modern scientific observations, and hence is considered to be the benchmark event for space weather. The magnetic storm caused worldwide observations of auroras even at very low latitudes such as Hawaii, Panama, or Santiago, and the available magnetic field measurement at Bombay, India, showed two peaks: the main was the Carrington event which occurred in day time in East Asia, and a second storm after the Carrington event which occurred at night in East Asia. In this paper, we present a result from surveys of aurora records in East Asia, which provides new information of the aurora activity of this important event. We found some new East Asian records of low latitude aurora observations caused by the storm which occurred after the Carrington event. The size of the aurora belt of the second peak of the Carrington magnetic storm was even wider than usual low-latitude aurora events.
We present a catalogue of nearly 3,000 submillimetre sources detected at 850um over ~5 square degrees surveyed as part of the James Clerk Maxwell Telescope (JCMT) SCUBA-2 Cosmology Legacy Survey (S2CLS). This is the largest survey of its kind at 850um, probing a meaningful cosmic volume at the peak of star formation activity and increasing the sample size of submillimetre galaxies selected at 850um by an order of magnitude. We describe the wide 850um survey component of S2CLS, which covers the key extragalactic survey fields: UKIDSS-UDS, COSMOS, Akari-NEP, Extended Groth Strip, Lockman Hole North, SSA22 and GOODS-North. The average 1-sigma depth of S2CLS is 1.2 mJy/beam, approaching the SCUBA-2 850um confusion limit, which we determine to be ~0.8 mJy/beam. We measure the single dish 850um number counts to unprecedented accuracy, reducing the Poisson errors on the differential counts to approximately 4% at S_850~3mJy. With several independent fields, we investigate field-to-field variance, finding that the number counts on 0.5-1 degree scales are generally within 50% of the S2CLS mean for S_850>3mJy, with scatter consistent with the Poisson and estimated cosmic variance uncertainties, although there is a marginal (2-sigma) density enhancement in the GOODS-North field. The observed number counts are in reasonable agreement with recent phenomenological and semi-analytic models. Finally, the large solid angle of S2CLS allows us to measure the bright-end counts: at S_850>10mJy there are approximately ten sources per square degree, and we detect the distinctive up-turn in the number counts indicative of the detection of local sources of 850um emission and strongly lensed high-redshift galaxies. Here we describe the data collection and reduction procedures and present calibrated maps and a catalogue of sources; these are made publicly available.
Assuming a unitarily invariant norm $|||\cdot|||$ is given on a two-sided ideal of bounded linear operators acting on a separable Hilbert space, it induces some unitarily invariant norms $|||\cdot|||$ on matrix algebras $\mathcal{M}_n$ for all finite values of $n$ via $|||A|||=|||A\oplus 0|||$. We show that if $\mathscr{A}$ is a $C^*$-algebra of finite dimension $k$ and $Φ: \mathscr{A} \to \mathcal{M}_n$ is a unital completely positive map, then \begin{equation*} |||Φ(AB)-Φ(A)Φ(B)||| \leq \frac{1}{4} |||I_{n}|||\,|||I_{kn}||| d_A d_B \end{equation*} for any $A,B \in \mathscr{A}$, where $d_X$ denotes the diameter of the unitary orbit $\{UXU^*: U \mbox{ is unitary}\}$ of $X$ and $I_{m}$ stands for the identity of $\mathcal{M}_{m}$. Further we get an analogous inequality for certain $n$-positive maps in the setting of full matrix algebras by using some matrix tricks. We also give a Grüss operator inequality in the setting of $C^*$-algebras of arbitrary dimension and apply it to some inequalities involving continuous fields of operators.
Shabeh ul Hasson, Valerio Lucarini, Salvatore Pascale
We investigate how the climate models contributing to the PCMDI/CMIP3 dataset describe the hydrological cycle over four major South and Southeast Asian river basins (Indus, Ganges, Brahmaputra and Mekong) for the 20th, 21st (13 models) and 22nd (10 models) centuries. For the 20th century, some models do not seem to conserve water at the river basin scale up to a good degree of approximation. The simulated precipitation minus evaporation (P - E), total runoff (R) and precipitation (P) quantities are neither consistent with the observations nor among the models themselves. Most of the models underestimate P - E for all four river basins, which is mainly associated with the underestimation of precipitation. This is in agreement with the recent results on the biases of the representation of monsoonal dynamics by GCMs. Overall, a modest inter-model agreement is found only for the evaporation and inter-annual variability of P - E. For the 21st and 22nd centuries, models agree on the negative (positive) changes of P - E for the Indus basin (Ganges, Brahmaputra and Mekong basins). Most of the models foresee an increase in the inter-annual variability of P - E for the Ganges and Mekong basins, thus suggesting an increase in large low-frequency dry/wet events. Instead, no considerable future change in the inter-annual variability of P - E is found for the Indus and Brahmaputra basins.