Search

Calorimetric determination of enthalpies of mixing ($\Delta$H$_{\rm mix}$) of multicomponent molten salts often employs empirical models that lack parameters with clear physical interpretation (e.g., coordination numbers, molar volumes, and pair potentials). Although such physics informed models are not always needed, a thermodynamic understanding of the relationships between excess energies of mixing and local to intermediate solvation structures is particularly important for pyrochemical separation, as is the case for lanthanides (Ln), which are common neutron poisons and critical industrial elements found in spent nuclear fuels. Here we implement the molecular interaction volume model (MIVM) to synthesize information from experimentally measured $\Delta$H$_{\rm mix}$ (using high temperature melt drop calorimetry) and the distribution of solvation structures from ab initio molecular dynamics (AIMD) simulations. This was demonstrated by a case study of molten salt system consisted of LaCl$_3$ mixing with a eutectic LiCl-KCl (58mol% to 42mol%) at 873 K and 1133 K. The parameters modelled from MIVM were used to extrapolate excess Gibbs energy ($\Delta$G$_{\rm mix}$), and compositional dependence of La$^{3+}$ activity in the LaCl$_3$-(LiCl-KCl) system. In contrast, by AIMD or polarizable ion model (PIM) simulations, a significant deviation regarding the predicted $\Delta$H$_{\rm mix}$ was seen if computed directly from the molecular dynamic trajectories. The integrated experimental and simulation data within the MIVM formalism are generalizable to a wide variety of molten salts and demonstrate a significant improvement over currently employed methods to study molten salts for nuclear and separations sciences.

Inverse rendering of outdoor scenes from unconstrained image collections is a challenging task, particularly illumination/albedo ambiguities and occlusion of the illumination environment (shadowing) caused by geometry. However, there are many cues in an image that can aid in the disentanglement of geometry, albedo and shadows. Whilst sky is frequently masked out in state-of-the-art methods, we exploit the fact that any sky pixel provides a direct observation of distant lighting in the corresponding direction and, via a neural illumination prior, a statistical cue to derive the remaining illumination environment. The incorporation of our illumination prior is enabled by a novel `outside-in' method for computing differentiable sky visibility based on a neural directional distance function. This is highly efficient and can be trained in parallel with the neural scene representation, allowing gradients from appearance loss to flow from shadows to influence the estimation of illumination and geometry. Our method estimates high-quality albedo, geometry, illumination and sky visibility, achieving state-of-the-art results on the NeRF-OSR relighting benchmark. Our code and models can be found at https://github.com/JADGardner/neusky, Comment: Accepted to ECCV 2024

Inverse rendering is an ill-posed problem. Previous work has sought to resolve this by focussing on priors for object or scene shape or appearance. In this work, we instead focus on a prior for natural illuminations. Current methods rely on spherical harmonic lighting or other generic representations and, at best, a simplistic prior on the parameters. This results in limitations for the inverse setting in terms of the expressivity of the illumination conditions, especially when taking specular reflections into account. We propose a conditional neural field representation based on a variational auto-decoder and a transformer decoder. We extend Vector Neurons to build equivariance directly into our architecture, and leveraging insights from depth estimation through a scale-invariant loss function, we enable the accurate representation of High Dynamic Range (HDR) images. The result is a compact, rotation-equivariant HDR neural illumination model capable of capturing complex, high-frequency features in natural environment maps. Training our model on a curated dataset of 1.6K HDR environment maps of natural scenes, we compare it against traditional representations, demonstrate its applicability for an inverse rendering task and show environment map completion from partial observations. We share our PyTorch implementation, dataset and trained models at https://github.com/JADGardner/ns_reni, Comment: Project Repo - https://github.com/JADGardner/ns_reni. arXiv admin note: substantial text overlap with arXiv:2206.03858

Probabilistic programming languages rely fundamentally on some notion of sampling, and this is doubly true for probabilistic programming languages which perform Bayesian inference using Monte Carlo techniques. Verifying samplers - proving that they generate samples from the correct distribution - is crucial to the use of probabilistic programming languages for statistical modelling and inference. However, the typical denotational semantics of probabilistic programs is incompatible with deterministic notions of sampling. This is problematic, considering that most statistical inference is performed using pseudorandom number generators. We present a higher-order probabilistic programming language centred on the notion of samplers and sampler operations. We give this language an operational and denotational semantics in terms of continuous maps between topological spaces. Our language also supports discontinuous operations, such as comparisons between reals, by using the type system to track discontinuities. This feature might be of independent interest, for example in the context of differentiable programming. Using this language, we develop tools for the formal verification of sampler correctness. We present an equational calculus to reason about equivalence of samplers, and a sound calculus to prove semantic correctness of samplers, i.e. that a sampler correctly targets a given measure by construction., Comment: Extended version of LiCS 2023 paper

Embedding a face image to a descriptor vector using a deep CNN is a widely used technique in face recognition. Via several possible training strategies, such embeddings are supposed to capture only identity information. Information about the environment (such as background and lighting) or changeable aspects of the face (such as pose, expression, presence of glasses, hat etc.) should be discarded since they are not useful for recognition. In this paper, we present a surprising result that this is not the case. We show that non-ID attributes, as well as landmark positions and the image histogram can be recovered from the ID embedding of state-of-the-art face embedding networks (VGGFace2 and ArcFace). In fact, these non-ID attributes can be predicted from ID embeddings with similar accuracy to a prediction from the original image. Going further, we present an optimisation strategy that uses a generative model (specifically StyleGAN2 for faces) to recover images from an ID embedding. We show photorealistic inversion from ID embedding to face image in which not only is the ID realistically reconstructed but the pose, lighting and background/apparel to some extent as well., Comment: SCIA 2023

In this paper we propose a novel method for zero-shot, cross-domain image retrieval in which we make two key contributions. The first is a test-time re-ranking procedure that enables query-gallery pairs, without meaningful shared visual features, to be matched by incorporating gallery-gallery ranks into an iterative re-ranking process. The second is the use of cross-attention at training time and knowledge distillation to encourage cross-attention-like features to be extracted at test time from a single image. When combined with the Vision Transformer architecture and zero-shot retrieval losses, our approach yields state-of-the-art results on the Sketchy and TU-Berlin sketch-based image retrieval benchmarks. However, unlike many previous methods, none of the components in our approach are engineered specifically towards the sketch-based image retrieval task - it can be generally applied to any cross-domain, zero-shot retrieval task. We therefore also show results on zero-shot cartoon-to-photo retrieval using the Office-Home dataset.

The arrival time prediction of Coronal mass ejections (CMEs) is an area of active research. Many methods with varying levels of complexity have been developed to predict CME arrival. However, the mean absolute error (MAE) of predictions remains above 12 hours, even with the increasing complexity of methods. In this work we develop a new method for CME arrival time prediction that uses magnetohydrodynamic simulations involving data-constrained flux-rope-based CMEs, which are introduced in a data-driven solar wind background. We found that, for 6 CMEs studied in this work, the MAE in arrival time was ~8 hours. We further improved our arrival time predictions by using ensemble modeling and comparing the ensemble solutions with STEREO-A&B heliospheric imager data. This was done by using our simulations to create synthetic J-maps. A machine learning (ML) method called the lasso regression was used for this comparison. Using this approach, we could reduce the MAE to ~4 hours. Another ML method based on the neural networks (NNs) made it possible to reduce the MAE to ~5 hours for the cases when HI data from both STEREO-A&B were available. NNs are capable of providing similar MAE when only the STEREO-A data is used. Our methods also resulted in very encouraging values of standard deviation (precision) of arrival time. The methods discussed in this paper demonstrate significant improvements in the CME arrival time predictions. Our work highlights the importance of using ML techniques in combination with data-constrained magnetohydrodynamic modeling to improve space weather predictions.

We consider the problem of finding an accurate representation of neuron shapes, extracting sub-cellular features, and classifying neurons based on neuron shapes. In neuroscience research, the skeleton representation is often used as a compact and abstract representation of neuron shapes. However, existing methods are limited to getting and analyzing "curve" skeletons which can only be applied for tubular shapes. This paper presents a 3D neuron morphology analysis method for more general and complex neuron shapes. First, we introduce the concept of skeleton mesh to represent general neuron shapes and propose a novel method for computing mesh representations from 3D surface point clouds. A skeleton graph is then obtained from skeleton mesh and is used to extract sub-cellular features. Finally, an unsupervised learning method is used to embed the skeleton graph for neuron classification. Extensive experiment results are provided and demonstrate the robustness of our method to analyze neuron morphology.

This paper presents a learning framework to estimate an agent capability and task requirement model for multi-agent task allocation. With a set of team configurations and the corresponding task performances as the training data, linear task constraints can be learned to be embedded in many existing optimization-based task allocation frameworks. Comprehensive computational evaluations are conducted to test the scalability and prediction accuracy of the learning framework with a limited number of team configurations and performance pairs. A ROS and Gazebo-based simulation environment is developed to validate the proposed requirements learning and task allocation framework in practical multi-agent exploration and manipulation tasks. Results show that the learning process for scenarios with 40 tasks and 6 types of agents uses around 12 seconds, ending up with prediction errors in the range of 0.5-2%., Comment: The video and open-source code are at https://brg.engin.umich.edu/publications/learn-multiagent-taskreq/

We propose to tackle the multiview photometric stereo problem using an extension of Neural Radiance Fields (NeRFs), conditioned on light source direction. The geometric part of our neural representation predicts surface normal direction, allowing us to reason about local surface reflectance. The appearance part of our neural representation is decomposed into a neural bidirectional reflectance function (BRDF), learnt as part of the fitting process, and a shadow prediction network (conditioned on light source direction) allowing us to model the apparent BRDF. This balance of learnt components with inductive biases based on physical image formation models allows us to extrapolate far from the light source and viewer directions observed during training. We demonstrate our approach on a multiview photometric stereo benchmark and show that competitive performance can be obtained with the neural density representation of a NeRF., Comment: 9 pages, 6 figures, 1 table

Inverse rendering is an ill-posed problem. Previous work has sought to resolve this by focussing on priors for object or scene shape or appearance. In this work, we instead focus on a prior for natural illuminations. Current methods rely on spherical harmonic lighting or other generic representations and, at best, a simplistic prior on the parameters. We propose a conditional neural field representation based on a variational auto-decoder with a SIREN network and, extending Vector Neurons, build equivariance directly into the network. Using this, we develop a rotation-equivariant, high dynamic range (HDR) neural illumination model that is compact and able to express complex, high-frequency features of natural environment maps. Training our model on a curated dataset of 1.6K HDR environment maps of natural scenes, we compare it against traditional representations, demonstrate its applicability for an inverse rendering task and show environment map completion from partial observations. A PyTorch implementation, our dataset and trained models can be found at jadgardner.github.io/RENI., Comment: NeurIPS 2022 - Project Website: jadgardner.github.io/RENI

Knowledge of the vapor-liquid equilibrium (VLE) properties of molten salts is important in the design of thermal energy storage systems for solar power and nuclear energy production applications. The high temperatures involved make their experimental determination problematic, and the development of both macroscopic thermodynamic correlations and predictive molecular-based methodologies are complicated by the requirement to appropriately incorporate the chemically reacting vapor-phase species. We derive a general thermodynamic-based atomistic simulation framework for molten salt VLE prediction and show its application to NaCl. Its input quantities are temperature-dependent ideal-gas free energy data for the vapor phase reactions, and density and residual chemical potential data for the liquid. If these are not available experimentally, the former may be predicted using standard electronic structure software, and the latter by means of classical atomistic simulation methodology. The framework predicts the temperature dependence of vapor pressure, coexisting phase densities, vapor phase composition, and vaporization enthalpy. It also predicts the concentrations of vapor phase species present in minor amounts (such as the free ions), quantities that are extremely difficult to measure experimentally. We furthermore use the VLE results to obtain approximations to the complete VLE binodal dome and the critical properties. We verify the framework for molten NaCl, for which experimentally based density and chemical potential data are available in the literature. We then apply it to the analysis of NaCl simulation data for two commonly used atomistic force fields. The framework can be readily extended to molten salt mixtures and to ionic liquids., Comment: Submitted to the Journal of Chemical Physics