Search for: All records

We see the external world as consisting not only of objects and their parts, but also of relations that hold between them. Visual analogy, which depends on similarities between relations, provides a clear example of how perception supports reasoning. Here we report an experiment in which we quantitatively measured the human ability to find analogical mappings between parts of different objects, where the objects to be compared were drawn either from the same category (e.g., images of two mammals, such as a dog and a horse), or from two dissimilar categories (e.g., a chair image mapped to a cat image). Humans showed systematic mapping patterns, but with greater variability in mapping responses when objects were drawn from dissimilar categories. We simulated the human response of analogical mapping using a computational model of mapping between 3D objects, visiPAM (visual Probabilistic Analogical Mapping). VisiPAM takes point-cloud representations of two 3D objects as inputs, and outputs the mapping between analogous parts of the two objects. VisiPAM consists of a visual module that constructs structural representations of individual objects, and a reasoning module that identifies a probabilistic mapping between parts of the two 3D objects. Model simulations not only capture the qualitative pattern of human mapping performance cross conditions, but also approach human-level reliability in solving visual analogy problems. 

Is analogical reasoning a task that must be learned to solve from scratch by applying deep learning models to massive numbers of reasoning problems? Or are analogies solved by computing similarities between structured representations of analogs? We address this question by comparing human performance on visual analogies created using images of familiar three-dimensional objects (cars and their subregions) with the performance of alternative computational models. Human reasoners achieved above-chance accuracy for all problem types, but made more errors in several conditions (e.g., when relevant subregions were occluded). We compared human performance to that of two recent deep learning models (Siamese Network and Relation Network) directly trained to solve these analogy problems, as well as to that of a compositional model that assesses relational similarity between part-based representations. The compositional model based on part representations, but not the deep learning models, generated qualitative performance similar to that of human reasoners. 

Acoustic compressional and shear wave velocities (VP, VS) of anhydrous (AHRG) and hydrous rhyolitic glasses (HRG) containing 3.28 wt% (HRG-3) and 5.90 wt% (HRG-6) total water concentra- tion (H2Ot) have been measured using Brillouin light scattering (BLS) spectroscopy up to 3 GPa in a diamond-anvil cell at ambient temperature. In addition, Fourier-transform infrared (FTIR) spectroscopy was used to measure the speciation of H2O in the glasses up to 3 GPa. At ambient pressure, HRG-3 contains 1.58 (6) wt% hydroxyl groups (OH–) and 1.70 (7) wt% molecular water (H2Om) while HRG-6 contains 1.67 (10) wt% OH– and 4.23 (17) wt% H2Om where the numbers in parentheses are ±1σ. With increasing pressure, very little H2Om, if any, converts to OH– within uncertainties in hydrous rhyolitic glasses such that HRG-6 contains much more H2Om than HRG-3 at all experimental pressures. We observe a nonlinear relationship between high-pressure sound velocities and H2Ot, which is attributed to the distinct effects of each water species on acoustic velocities and elastic moduli of hydrous glasses. Near ambient pressure, depolymerization due to OH– reduces VS and G more than VP and KS. VP and KS in both anhydrous and hydrous glasses decrease with increasing pressure up to ~1–2 GPa before increasing with pressure. Above ~1–2 GPa, VP and KS in both hydrous glasses converge with those in AHRG. In particular, VP in HRG-6 crosses over and becomes higher than VP in AHRG. HRG-6 displays lower VS and G than HRG-3 near ambient pressure, but VS and G in these glasses converge above ~2 GPa. Our results show that hydrous rhyolitic glasses with ~2–4 wt% H2Om can be as incompressible as their anhydrous counterpart above ~1.5 GPa. The nonlinear effects of hydration on high-pressure acoustic velocities and elastic moduli of rhyolitic glasses observed here may provide some insight into the behavior of hydrous silicate melts in felsic magma chambers at depth. 

Abstract An array of twelve 0.28 kg lithium molybdate (LMO) low-temperature bolometers equipped with 16 bolometric Ge light detectors, aiming at optimization of detector structure for CROSS and CUPID double-beta decay experiments, was constructed and tested in a low-background pulse-tube-based cryostat at the Canfranc underground laboratory in Spain. Performance of the scintillating bolometers was studied depending on the size of phonon NTD-Ge sensors glued to both LMO and Ge absorbers, shape of the Ge light detectors (circular vs. square, from two suppliers), in different light collection conditions (with and without reflector, with aluminum coated LMO crystal surface). The scintillating bolometer array was operated over 8 months in the low-background conditions that allowed to probe a very low, μBq/kg, level of the LMO crystals radioactive contamination by 228 Th and 226 Ra. 

Abstract CUPID is a next-generation bolometric experiment aiming at searching for neutrinoless double-beta decay with ∼250 kg of isotopic mass of 100 Mo. It will operate at ∼10 mK in a cryostat currently hosting a similar-scale bolometric array for the CUORE experiment at the Gran Sasso National Laboratory (Italy). CUPID will be based on large-volume scintillating bolometers consisting of 100 Mo-enriched Li 2 MoO 4 crystals, facing thin Ge-wafer-based bolometric light detectors. In the CUPID design, the detector structure is novel and needs to be validated. In particular, the CUORE cryostat presents a high level of mechanical vibrations due to the use of pulse tubes and the effect of vibrations on the detector performance must be investigated. In this paper we report the first test of the CUPID-design bolometric light detectors with NTD-Ge sensors in a dilution refrigerator equipped with a pulse tube in an above-ground lab. Light detectors are characterized in terms of sensitivity, energy resolution, pulse time constants, and noise power spectrum. Despite the challenging noisy environment due to pulse-tube-induced vibrations, we demonstrate that all the four tested light detectors comply with the CUPID goal in terms of intrinsic energy resolution of 100 eV RMS baseline noise. Indeed, we have measured 70–90 eV RMS for the four devices, which show an excellent reproducibility. We have also obtained high energy resolutions at the 356 keV line from a 133 Ba source, as good as Ge semiconductor γ detectors in this energy range.