Search for: All records

Stress fields imparted with an ultrafast laser can correct low spatial frequency surface figure error of mirrors through ultrafast laser stress figuring (ULSF): the formation of nanograting structures within the bulk substrate generates localized stress, creating bending moments that equilibrize via wafer deformation. For ULSF to be used as an optical figuring process, the ultrafast laser generated stress must be effectively permanent or risk unwanted figure drift. Two isochronal annealing experiments were performed to measure ultrafast laser-generated stress stability in fused silica and Corning ultra-low expansion (ULE) wafers. The first experiment tracked changes to induced astigmatism up to 1000 °C on 25.4 mm-diameter wafers. Only small changes were measured after each thermal cycle up to 500 °C for both materials, but significant changes were observed at higher temperatures. The second experiment tracked stress changes in fused silica and ULE up to 500 °C but with 4 to 16× higher signal-to-noise ratio. Change in trefoil on 100 mm-diameter wafers was measured, and the induced stress in fused silica and ULE was found to be stable after thermal cycling up to 300 °C and 200 °C, respectively, with larger changes at higher temperatures. 

Vision-Language Models (VLMs) have made rapid progress in reasoning across visual and textual data. While VLMs perform well on vision tasks that they are trained on, our results highlight key challenges in abstract pattern recognition. We present GlyphPattern, a 954 item dataset that pairs 318 human-written descriptions of visual patterns from 40 writing systems with three visual presentation styles.GlyphPattern evaluates abstract pattern recognition in VLMs, requiring models to understand and judge natural language descriptions of visual patterns. GlyphPattern patterns are drawn from a large-scale cognitive science investigation of human writing systems; as a result, they are rich in spatial reference and compositionality. Our experiments show that GlyphPattern is challenging for state-of-the-art VLMs (GPT-4o achieves only 55% accuracy), with marginal gains from few-shot prompting. Our detailed analysis reveals errors at multiple levels, including visual processing, natural language understanding, and pattern generalization. 

ABSTRACT Marine fish are likely one of the top producers of biogenic carbonate in the oceans. However, nothing is known about the production rate and composition of intestinal carbonate (ichthyocarbonate) excreted by mesopelagic fishes, which are small, fragile and account for up to 94% of global fish biomass. To address this knowledge gap, and associated uncertainty of global ichthyocarbonate production, we identified a model species residing at 350–430 m, depths relevant for mesopelagic fishes. The blackbelly rosefish (Helicolenus dactylopterus) lacks swim bladders and survives capture and transfer to the lab. Freshly collected blackbelly rosefish, maintained at 6°C, contained high amounts of intestinal ichthyocarbonate (0.4 g kg−1) and excreted ∼5 mg kg−1 h−1 ichthyocarbonate, in agreement with expectations based on allometric and thermal relationships for other species. Despite longer intestinal residence time, intestinal and excreted ichthyocarbonates are similar in crystallite morphology, composition and sinking rate, but have a higher dissolution rate than that produced by shallow water species at higher temperatures, ruling out strong effects of pressure and low temperatures on ichthyocarbonate formation and excretion. Considering allometric and thermal relationships, the metabolic rate of blackbelly rosefish is lower than that of other marine fish in general, and mesopelagic fishes in particular. Our observations support assumptions of ichthyocarbonate excretion by mesopelagic fishes, and suggest that thermal and allometric relationships for ichthyocarbonate excretion determined from shallow water species extend to fish populations at depth. 

Gold nanoparticles (AuNPs) are increasingly used in applications across the biomedical domain, yet their long-term biodistribution and biocompatibility remain poorly understood. Conventional brightfield microscopy imaging techniques often fail to detect AuNPs due to optical diffraction limits and lack of chromogenic contrast. Understanding the biodistribution and ultimate fate of these nonbiodegradable NPs is crucial for further development of AuNP-based therapeutics and diagnostics. Here, we present a label-free multiphoton luminescence (MPL) imaging workflow that enables sensitive detection of AuNPs in liver histology sections, even 1 year after intravenous (IV) administration. MPL imaging exploits the intrinsic nonlinear optical properties of AuNPs to generate broadband emission under ultrafast pulsed laser excitation, enabling subcellular localization without exogenous labels while having the ability to rapidly image entire organ sections. The intrinsic, distinct broadband MPL emission produced by gold allows us to study these NPs in their biological context without extrinsic labels while also faithfully representing the surrounding tissue architecture via autofluorescence and second harmonic generation. We demonstrate that MPL imaging detects up to 98% more AuNP-positive regions than brightfield microscopy in challenging low-dose (1 nM) conditions and requires no modification of standard histology workflows. Correlative imaging with SEM–EDS confirms high spatial specificity (AUC = 0.955) of MPL for AuNP localization. Dose-dependent retention patterns were observed across liver tissue, and MPL analysis showed strong correlation with ICP–MS quantification. Importantly, histological and immunohistochemical analyses (Masson’s trichrome, CD3, and TUNEL) revealed no significant fibrosis, immune activation, or apoptosis in liver tissue at either low (1 nM) or high (10 nM) doses at 1 year post IV administration. These findings establish MPL imaging as a robust, label-free tool for long-term tracking of AuNPs in biological tissue and highlight its potential for improving biodistribution and safety assessments. 

To gain support, children use signals to communicate their needs and wants to parents. Infant signals of need, particularly infant cries, have been extensively studied in diverse populations. However, the full range of potential child signals of need, which extend beyond cries, has rarely been investigated in a single study of children of all ages. To help fill this gap, we collected mother and other primary caregiver reports of three common types of child signaling from 131 families with 263 children on Utila, a small island off the coast of Honduras. In exploratory analyses, we found that child signaling was common in both sexes and across all ages, although it decreased with age and neighborhood quality and increased with the frequency of conflict between children and caretakers. Consistent with signaling theory, children who were sad more frequently were perceived as needier within the household and were more likely to receive investment. Caregivers were less likely to respond positively in situations of family conflict or child transgressions, and more likely for injuries and illness. Our results suggest that evolutionary theories of signaling can help explain patterns of child sadness, crying, and temper tantrums. 

Although LLMs are increasing the productivity of professional programmers, existing work shows that beginners struggle to prompt LLMs to solve text-to-code tasks (Nguyen et al., 2024; Prather et al., 2024; Mordechai et al., 2024). Why is this the case? This paper explores two competing hypotheses about the cause of student-LLM miscommunication: (1) students simply lack the technical vocabulary needed to write good prompts, and (2) students do not understand the extent of information that LLMs need to solve code generation tasks. We study (1) with a causal intervention experiment on technical vocabulary and (2) by analyzing graphs that abstract how students edit prompts and the different failures that they encounter. We find that substance beats style: a poor grasp of technical vocabulary is merely correlated with prompt failure; that the information content of prompts predicts success; that students get stuck making trivial edits; and more. Our findings have implications for the use of LLMs in programming education, and for efforts to make computing more accessible with LLMs.