Second-order quantile methods for experts and combinatorial games

We aim to design strategies for sequential decision making that adjust to the difficulty of the learning problem. We study this question both in the setting of prediction with expert advice, and for more general combinatorial decision tasks. We are not satisfied with just guaranteeing minimax regret rates, but we want our algorithms to perform significantly better on easy data. Two popular ways to formalize such adaptivity are second-order regret bounds and quantile bounds. The underlying notions of 'easy data', which may be paraphrased as "the learning problem has small variance" and "multiple decisions are useful", are synergetic. But even though there are sophisticated algorithms that exploit one of the two, no existing algorithm is able to adapt to both. In this paper we outline a new method for obtaining such adaptive algorithms, based on a potential function that aggregates a range of learning rates (which are essential tuning parameters). By choosing the right prior we construct efficient algorithms and show that they reap both benefits by proving the first bounds that are both second-order and incorporate quantiles.

Process variability of pollutant build-up on urban road surfaces

Knowledge of the pollutant build-up process is a key requirement for developing stormwater pollution mitigation strategies. In this context, process variability is a concept which needs to be understood in-depth. Analysis of particulate build-up on three road surfaces in an urban catchment confirmed that particles <150µm and >150µm have characteristically different build-up patterns, and these patterns are consistent over different field conditions. Three theoretical build-up patterns were developed based on the size-fractionated particulate build-up patterns, and these patterns explain the variability in particle behavior and the variation in particle-bound pollutant load and composition over the antecedent dry period. Behavioral variability of particles <150µm was found to exert the most significant influence on the build-up process variability. As characterization of process variability is particularly important in stormwater quality modeling, it is recommended that the influence of behavioral variability of particles <150µm on pollutant build-up should be specifically addressed. This would eliminate model deficiencies in the replication of the build-up process and facilitate the accounting of the inherent process uncertainty, and thereby enhance the water quality predictions.

Quantifying uncertainty in parameter estimates for stochastic models of collective cell spreading using approximate Bayesian computation

Wound healing and tumour growth involve collective cell spreading, which is driven by individual motility and proliferation events within a population of cells. Mathematical models are often used to interpret experimental data and to estimate the parameters so that predictions can be made. Existing methods for parameter estimation typically assume that these parameters are constants and often ignore any uncertainty in the estimated values. We use approximate Bayesian computation (ABC) to estimate the cell diffusivity, D, and the cell proliferation rate, λ, from a discrete model of collective cell spreading, and we quantify the uncertainty associated with these estimates using Bayesian inference. We use a detailed experimental data set describing the collective cell spreading of 3T3 fibroblast cells. The ABC analysis is conducted for different combinations of initial cell densities and experimental times in two separate scenarios: (i) where collective cell spreading is driven by cell motility alone, and (ii) where collective cell spreading is driven by combined cell motility and cell proliferation. We find that D can be estimated precisely, with a small coefficient of variation (CV) of 2–6%. Our results indicate that D appears to depend on the experimental time, which is a feature that has been previously overlooked. Assuming that the values of D are the same in both experimental scenarios, we use the information about D from the first experimental scenario to obtain reasonably precise estimates of λ, with a CV between 4 and 12%. Our estimates of D and λ are consistent with previously reported values; however, our method is based on a straightforward measurement of the position of the leading edge whereas previous approaches have involved expensive cell counting techniques. Additional insights gained using a fully Bayesian approach justify the computational cost, especially since it allows us to accommodate information from different experiments in a principled way.

A semi-alternating direction method for a 2-D fractional FitzHugh–Nagumo monodomain model on an approximate irregular domain

A FitzHugh-Nagumo monodomain model has been used to describe the propagation of the electrical potential in heterogeneous cardiac tissue. In this paper, we consider a two-dimensional fractional FitzHugh-Nagumo monodomain model on an irregular domain. The model consists of a coupled Riesz space fractional nonlinear reaction-diffusion model and an ordinary differential equation, describing the ionic fluxes as a function of the membrane potential. Secondly, we use a decoupling technique and focus on solving the Riesz space fractional nonlinear reaction-diffusion model. A novel spatially second-order accurate semi-implicit alternating direction method (SIADM) for this model on an approximate irregular domain is proposed. Thirdly, stability and convergence of the SIADM are proved. Finally, some numerical examples are given to support our theoretical analysis and these numerical techniques are employed to simulate a two-dimensional fractional Fitzhugh-Nagumo model on both an approximate circular and an approximate irregular domain.

High order unconditionally stable difference schemes for the Riesz space-fractional telegraph equation

In this paper, a class of unconditionally stable difference schemes based on the Pad´e approximation is presented for the Riesz space-fractional telegraph equation. Firstly, we introduce a new variable to transform the original dfferential equation to an equivalent differential equation system. Then, we apply a second order fractional central difference scheme to discretise the Riesz space-fractional operator. Finally, we use (1, 1), (2, 2) and (3, 3) Pad´e approximations to give a fully discrete difference scheme for the resulting linear system of ordinary differential equations. Matrix analysis is used to show the unconditional stability of the proposed algorithms. Two examples with known exact solutions are chosen to assess the proposed difference schemes. Numerical results demonstrate that these schemes provide accurate and efficient methods for solving a space-fractional hyperbolic equation.

A fast semi-implicit difference method for a nonlinear two-sided space-fractional diffusion equation with variable diffusivity coefficients

In this paper, we derive a new nonlinear two-sided space-fractional diffusion equation with variable coefficients from the fractional Fick’s law. A semi-implicit difference method (SIDM) for this equation is proposed. The stability and convergence of the SIDM are discussed. For the implementation, we develop a fast accurate iterative method for the SIDM by decomposing the dense coefficient matrix into a combination of Toeplitz-like matrices. This fast iterative method significantly reduces the storage requirement of O(n2)O(n2) and computational cost of O(n3)O(n3) down to n and O(nlogn)O(nlogn), where n is the number of grid points. The method retains the same accuracy as the underlying SIDM solved with Gaussian elimination. Finally, some numerical results are shown to verify the accuracy and efficiency of the new method.

Maximum principle and numerical method for the multi-term time–space Riesz–Caputo fractional differential equations

The maximum principle for the space and time–space fractional partial differential equations is still an open problem. In this paper, we consider a multi-term time–space Riesz–Caputo fractional differential equations over an open bounded domain. A maximum principle for the equation is proved. The uniqueness and continuous dependence of the solution are derived. Using a fractional predictor–corrector method combining the L1 and L2 discrete schemes, we present a numerical method for the specified equation. Two examples are given to illustrate the obtained results.

The mathematical modelling of cheese ripening

A mathematical model is developed for the ripening of cheese. Such models may assist predicting final cheese quality using measured initial composition. The main constituent chemical reactions are described with ordinary differential equations. Numerical solutions to the model equations are found using Matlab. Unknown parameter values have been fitted using experimental data available in the literature. The results from the numerical fitting are in good agreement with the data. Statistical analysis is performed on near infrared data provided to the MISG. However, due to the inhomogeneity and limited nature of the data, not many conclusions can be drawn from the analysis. A simple model of the potential changes in acidity of cheese is also considered. The results from this model are consistent with cheese manufacturing knowledge, in that the pH of cheddar cheese does not significantly change during ripening.

Copula-based spatial modelling of geometallurgical variables

The most important aspect of modelling a geological variable, such as metal grade, is the spatial correlation. Spatial correlation describes the relationship between realisations of a geological variable sampled at different locations. Any method for spatially modelling such a variable should be capable of accurately estimating the true spatial correlation. Conventional kriged models are the most commonly used in mining for estimating grade or other variables at unsampled locations, and these models use the variogram or covariance function to model the spatial correlations in the process of estimation. However, this usage assumes the relationships of the observations of the variable of interest at nearby locations are only influenced by the vector distance between the locations. This means that these models assume linear spatial correlation of grade. In reality, the relationship with an observation of grade at a nearby location may be influenced by both distance between the locations and the value of the observations (ie non-linear spatial correlation, such as may exist for variables of interest in geometallurgy). Hence this may lead to inaccurate estimation of the ore reserve if a kriged model is used for estimating grade of unsampled locations when nonlinear spatial correlation is present. Copula-based methods, which are widely used in financial and actuarial modelling to quantify the non-linear dependence structures, may offer a solution. This method was introduced by Bárdossy and Li (2008) to geostatistical modelling to quantify the non-linear spatial dependence structure in a groundwater quality measurement network. Their copula-based spatial modelling is applied in this research paper to estimate the grade of 3D blocks. Furthermore, real-world mining data is used to validate this model. These copula-based grade estimates are compared with the results of conventional ordinary and lognormal kriging to present the reliability of this method.

Influence of pollutant build-up on variability in wash-off from urban road surfaces

Variability in the pollutant wash-off process is a concept which needs to be understood in-depth in order to better assess the outcomes of stormwater quality models, and thereby strengthen stormwater pollution mitigation strategies. Current knowledge about the wash-off process does not extend to a clear understanding of the influence of the initially available pollutant build-up on the variability of the pollutant wash-off load and composition. Consequently, pollutant wash-off process variability is poorly characterised in stormwater quality models, which can result in inaccurate stormwater quality predictions. Mathematical simulation of particulate wash-off from three urban road surfaces confirmed that the wash-off load of particle size fractions <150µm and >150µm after a storm event vary with the build-up of the respective particle size fractions available at the beginning of the storm event. Furthermore, pollutant load and composition associated with the initially available build-up of <150µm particles predominantly influence the variability in washed-off pollutant load and composition. The influence of the build-up of pollutants associated with >150µm particles on wash-off process variability is significant only for relatively shorter duration storm events.

Wake angle for surface gravity waves on a finite depth fluid

Linear water wave theory suggests that wave patterns caused by a steadily moving disturbance are contained within a wedge whose half-angle depends on the depth-based Froude number $F_H$. For the problem of flow past an axisymmetric pressure distribution in a finite-depth channel, we report on the apparent angle of the wake, which is the angle of maximum peaks. For moderately deep channels, the dependence of the apparent wake angle on the Froude number is very different to the wedge angle, and varies smoothly as $F_H$ passes through the critical value $F_H=1$. For shallow water, the two angles tend to follow each other more closely, which leads to very large apparent wake angles for certain regimes.

Are there generalizable trends in the release of airborne synthetic clay nanoparticles from a jet milling process?

Despite recent efforts to assess the release of nanoparticles to the workplace during different nanotechnology activities, the existence of a generalizable trend in the particle release has yet to be identified. This study aimed to characterize the release of synthetic clay nanoparticles from a laboratory-based jet milling process by quantifying the variations arising from primary particle size and surface treatment of the material used, as well as the feed rate of the machine. A broad range of materials were used in this study, and the emitted particles mass (PM2.5) and number concentrations (PNC) were measured at the release source. Analysis of variance, followed by linear mixed-effects modeling, was applied to quantify the variations in PM2.5 and PNC of the released particles caused by the abovementioned factors. The results confirmed that using materials of different primary size and surface treatment affects the release of the particles from the same process by causing statistically-significant variations in PM2.5 and PNC. The interaction of these two factors should also be taken into account as it resulted in variations in the measured particles release properties. Furthermore, the feed rate of the milling machine was confirmed to be another influencing parameter. Although this research does not identify a specific pattern in the release of synthetic clay nanoparticles from the jet milling process generalizable to other similar settings, it emphasizes that each tested case should be handled individually in terms of exposure considerations.

Numerical treatment of a two-dimensional variable-order fractional nonlinear reaction-diffusion model

A two-dimensional variable-order fractional nonlinear reaction-diffusion model is considered. A second-order spatial accurate semi-implicit alternating direction method for a two-dimensional variable-order fractional nonlinear reaction-diffusion model is proposed. Stability and convergence of the semi-implicit alternating direct method are established. Finally, some numerical examples are given to support our theoretical analysis. These numerical techniques can be used to simulate a two-dimensional variable order fractional FitzHugh-Nagumo model in a rectangular domain. This type of model can be used to describe how electrical currents flow through the heart, controlling its contractions, and are used to ascertain the effects of certain drugs designed to treat arrhythmia.

Impaction of spray droplets on leaves: Influence of formulation and leaf character on shatter, bounce and adhesion

This paper combines experimental data with simple mathematical models to investigate the influence of spray formulation type and leaf character (wettability) on shatter, bounce and adhesion of droplets impacting with cotton, rice and wheat leaves. Impaction criteria that allow for different angles of the leaf surface and the droplet impact trajectory are presented; their predictions are based on whether combinations of droplet size and velocity lie above or below bounce and shatter boundaries. In the experimental component, real leaves are used, with all their inherent natural variability. Further, commercial agricultural spray nozzles are employed, resulting in a range of droplet characteristics. Given this natural variability, there is broad agreement between the data and predictions. As predicted, the shatter of droplets was found to increase as droplet size and velocity increased, and the surface became harder to wet. Bouncing of droplets occurred most frequently on hard to wet surfaces with high surface tension mixtures. On the other hand, a number of small droplets with low impact velocity were observed to bounce when predicted to lie well within the adhering regime. We believe this discrepancy between the predictions and experimental data could be due to air layer effects that were not taken into account in the current bounce equations. Other discrepancies between experiment and theory are thought to be due to the current assumption of a dry impact surface, whereas, in practice, the leaf surfaces became increasingly covered with fluid throughout the spray test runs.

A comprehensive mechanistic kinetic model for dilute acid hydrolysis of switchgrass cellulose to glucose, 5-HMF and levulinic acid

Switchgrass was treated by 1% (w/w) H₂SO₄in batch tube reactors at temperatures ranging from 140–220°C for up to 60 minutes. In this study, release patterns of glucose, 5-hydroxymethylfurfural (5-HMF), and levulinic acid from switchgrass cellulose were investigated through a mechanistic kinetic model. The predictions were consistent with the measured products of interest when new parameters reflecting the effects of reaction limitations, such as cellulose crystallinity, acid soluble lignin–glucose complex (ASL–glucose) and humins that cannot be quantitatively analyzed, were included. The new mechanistic kinetic model incorporating these parameters simulated the experimental data with R² above 0.97. Results showed that glucose yield was most sensitive to variations in the parameter regarding the cellulose crystallinity at low temperatures (140–180°C), while the impact of crystallinity on the glucose yield became imperceptible at elevated temperatures (200–220 °C). Parameters related to the undesired products (e.g. ASL–glucose and humins) were the most sensitive factors compared with rate constants and other additional parameters in impacting the levulinic acid yield at elevated temperatures (200–220°C), while their impacts were negligible at 140–180°C. These new findings provide a more rational explanation for the kinetic changes in dilute acid pretreatment performance and suggest that the influences of cellulose crystallinity and undesired products including ASL–glucose and humins play key roles in determining the generation of glucose, 5-HMF and levulinic acid from biomass-derived cellulose.