Machine learning for predictive condensed-phase simulation

We show how machine learning techniques based on Bayesian inference can be used to reach new levels of realism in the computer simulation of molecular materials, focusing here on water. We train our machine-learning algorithm using accurate, correlated quantum chemistry, and predict energies and forces in molecular aggregates ranging from clusters to solid and liquid phases. The widely used electronic-structure methods based on density-functional theory (DFT) give poor accuracy for molecular materials like water, and we show how our techniques can be used to generate systematically improvable corrections to DFT. The resulting corrected DFT scheme gives remarkably accurate predictions for the relative energies of small water clusters and of different ice structures, and greatly improves the description of the structure and dynamics of liquid water.

Engineering development of the Lancaster flexible bag.

Describes progress in the last 12 months which has established bag designs and fabrication techniques, giving greater confidence in the life and cost of these components. A quarter scale bag is under construction. Extensive tank testing has also established life time bending moment and mooring load envelopes, enabling hull and mooring design to proceed. A computer simulation programme has been used to check tank model results and to establish turbine and generator operating conditions. This has allowed generation and transmission component design to proceed, and suggests a high operating efficiency can be maintained with a simple control regime. Simple solutions in minor areas such as valve design and damage stability control add to the picture of steady progress in establishing the Lancaster Flexible Bag 's feasibility.

Structural and thermodynamic properties of different phases of supercooled liquid water.

Computer simulation results are reported for a realistic polarizable potential model of water in the supercooled region. Three states, corresponding to the low density amorphous ice, high density amorphous ice, and very high density amorphous ice phases are chosen for the analyses. These states are located close to the liquid-liquid coexistence lines already shown to exist for the considered model. Thermodynamic and structural quantities are calculated, in order to characterize the properties of the three phases. The results point out the increasing relevance of the interstitial neighbors, which clearly appear in going from the low to the very high density amorphous phases. The interstitial neighbors are found to be, at the same time, also distant neighbors along the hydrogen bonded network of the molecules. The role of these interstitial neighbors has been discussed in connection with the interpretation of recent neutron scattering measurements. The structural properties of the systems are characterized by looking at the angular distribution of neighboring molecules, volume and face area distribution of the Voronoi polyhedra, and order parameters. The cumulative analysis of all the corresponding results confirms the assumption that a close similarity between the structural arrangement of molecules in the three explored amorphous phases and that of the ice polymorphs I(h), III, and VI exists.

Structure of coexisting liquid phases of supercooled water: analogy with ice polymorphs.

The structural changes occurring in supercooled liquid water upon moving from one coexisting liquid phase to the other have been investigated by computer simulation using a polarizable interaction potential model. The obtained results favorably compare with recent neutron scattering data of high and low density water. In order to assess the physical origin of the observed structural changes, computer simulation of several ice polymorphs has also been carried out. Our results show that there is a strict analogy between the structure of various disordered (supercooled) and ordered (ice) phases of water, suggesting that the occurrence of several different phases of supercooled water is rooted in the same physical origin that is responsible for ice polymorphism.

State estimation schemes for independent component coupled hidden Markov models

Conventional Hidden Markov models generally consist of a Markov chain observed through a linear map corrupted by additive noise. This general class of model has enjoyed a huge and diverse range of applications, for example, speech processing, biomedical signal processing and more recently quantitative finance. However, a lesser known extension of this general class of model is the so-called Factorial Hidden Markov Model (FHMM). FHMMs also have diverse applications, notably in machine learning, artificial intelligence and speech recognition [13, 17]. FHMMs extend the usual class of HMMs, by supposing the partially observed state process is a finite collection of distinct Markov chains, either statistically independent or dependent. There is also considerable current activity in applying collections of partially observed Markov chains to complex action recognition problems, see, for example, [6]. In this article we consider the Maximum Likelihood (ML) parameter estimation problem for FHMMs. Much of the extant literature concerning this problem presents parameter estimation schemes based on full data log-likelihood EM algorithms. This approach can be slow to converge and often imposes heavy demands on computer memory. The latter point is particularly relevant for the class of FHMMs where state space dimensions are relatively large. The contribution in this article is to develop new recursive formulae for a filter-based EM algorithm that can be implemented online. Our new formulae are equivalent ML estimators, however, these formulae are purely recursive and so, significantly reduce numerical complexity and memory requirements. A computer simulation is included to demonstrate the performance of our results. © Taylor & Francis Group, LLC.

Complex study of transport AC loss in various 2G HTS racetrack coils

HTS racetrack coils are becoming important elements of an emerging number of superconducting devices such as generators or motors. In these devices the issue of AC loss is crucial, as performance and cooling power are derived from this quantity. This paper presents a comparative study of transport AC loss in two different types of 2G HTS racetrack coils. In this study, both experimental measurements and computer simulation approaches were employed. All the experiments were performed using classical AC electrical method. The finite-element computer model was used to estimate electromagnetic properties and calculate transport AC loss. The main difference between the characterized coils is covered inside tape architectures. While one coil uses tape based on RABITS magnetic substrate, the second coil uses a non-magnetic tape. Ferromagnetic loss caused by a magnetic substrate is an important issue involved in the total AC loss. As a result, the coil with the magnetic substrate surprised with high AC loss and rather low performance. © 2013 Elsevier B.V. All rights reserved.

Complex study of transport AC loss in various 2G HTS racetrack coils

HTS racetrack coils are becoming important elements of an emerging number of superconducting devices such as generators or motors. In these devices the issue of AC loss is crucial, as performance and cooling power are derived from this quantity. This paper presents a comparative study of transport AC loss in two different types of 2G HTS racetrack coils. In this study, both experimental measurements and computer simulation approaches were employed. All the experiments were performed using classical AC electrical method. The finite-element computer model was used to estimate electromagnetic properties and calculate transport AC loss. The main difference between the characterized coils is covered inside tape architectures. While one coil uses tape based on RABITS magnetic substrate, the second coil uses a non-magnetic tape. Ferromagnetic loss caused by a magnetic substrate is an important issue involved in the total AC loss. As a result, the coil with the magnetic substrate surprised with high AC loss and rather low performance. © 2013 Elsevier B.V. All rights reserved.

Implementation of multi-group cross-section methodology in BGCore MC-depletion code

BGCore is a software package for comprehensive computer simulation of nuclear reactor systems and their fuel cycles. The BGCore interfaces Monte Carlo particles transport code MCNP4C with a SARAF module - an independently developed code for calculating in-core fuel composition and spent fuel emissions following discharge. In BGCore system, depletion coupling methodology is based on the multi-group approach that significantly reduces computation time and allows tracking of large number of nuclides during calculations. In this study, burnup calculation capabilities of BGCore system were validated against well established and verified, computer codes for thermal and fast spectrum lattices. Very good agreement in k eigenvalue and nuclide densities prediction was observed for all cases under consideration. In addition, decay heat prediction capabilities of the BGCore system were benchmarked against the most recent edition of ANS Standard methodology for UO2 fuel decay power prediction in LWRs. It was found that the difference between ANS standard data and that predicted by the BGCore does not exceed 5%.

Bayesian Learning of Noisy Markov Decision Processes

We consider the inverse reinforcement learning problem, that is, the problem of learning from, and then predicting or mimicking a controller based on state/action data. We propose a statistical model for such data, derived from the structure of a Markov decision process. Adopting a Bayesian approach to inference, we show how latent variables of the model can be estimated, and how predictions about actions can be made, in a unified framework. A new Markov chain Monte Carlo (MCMC) sampler is devised for simulation from the posterior distribution. This step includes a parameter expansion step, which is shown to be essential for good convergence properties of the MCMC sampler. As an illustration, the method is applied to learning a human controller.

Modeling mode i fracture of bitumen films

The fracture behavior of thin films of bitumen in double cantilever beam (DCB) specimens was investigated over a wide range of temperature and loading rate conditions using finite-element analysis. The model includes a phenomenological model for the mechanical behavior of bitumen, implemented into a special-purpose finite-element user material subroutine, combined with a cohesive zone model (CZM) for simulating the fracture process. The finite-element model is validated against experimental results from laboratory tests of DCB specimens by comparing measured and predicted load-line deflection histories and fracture energy release rates. Computer simulation results agreed well with experimental data of DCB joints containing bitumen films in terms of peak stress, fracture toughness, and stress-strain history response. The predicted "normalized toughness," G=2h, was found to increase in a power-law manner with effective temperaturecompensated strain rate in the ductile region as previously observed experimentally. In the brittle regime, G=2h is virtually constant. The model successfully captured the ductile and brittle failure behavior of bitumen films in opening mode (tension) for stable crack growth conditions. © 2013 American Society of Civil Engineers.

Behavior learning of a face robot based on the characteristics of human instruction

Computer simulation experiments were performed to examine the effectiveness of OR- and comparative-reinforcement learning algorithms. In the simulation, human rewards were given as +1 and -1. Two models of human instruction that determine which reward is to be given in every step of a human instruction were used. Results show that human instruction may have a possibility of including both model-A and model-B characteristics, and it can be expected that the comparative-reinforcement learning algorithm is more effective for learning by human instructions.