Whilst a large body of plot and field-scale research exists on the sources, behaviour and mitigation of diffuse water pollution from agriculture, putting this evidence into a practical, context at large spatial scales to inform policy remains challenging. Understanding the behaviour of pollutants (nutrients, sediment, microbes and pesticides) and the effectiveness of mitigation strategies over whole catchments and long timeframes requires new, interdisciplinary approaches to organise and undertake research. This paper provides an introduction to the demonstration test catchments (DTC) programme, which was established in 2009 to gather empirical evidence on the cost-effectiveness of combinations of diffuse pollution mitigation measures at catchment scales. DTC firstly provides a physical platform of instrumented study catchments in which approaches for the mitigation of diffuse agricultural water pollution can be experimentally tested and iteratively improved. Secondly, it has established national and local knowledge exchange networks between researchers and stakeholders through which research has been co-designed. These have provided a vehicle to disseminate emerging findings to inform policy and land management practice. The role of DTC is that of an outdoor laboratory to develop knowledge and approaches that can be applied in less well studied locations. The research platform approach developed through DTC has brought together disparate research groups from different disciplines and institutions through nationally coordinated activities. It offers a model that can be adopted to organise research on other complex, interdisciplinary problems to inform policy and operational decision-making.

The water quality of the River Frome, Dorset, southern England, was monitored at weekly intervals from 1965 until 2009. Determinands included phosphorus, nitrogen, silicon, potassium, calcium, sodium, magnesium, pH, alkalinity and temperature. Nitrate-N concentrations increased from an annual average of 2.4 mg l⁻¹ in the mid to late 1960s to 6.0 mg l⁻¹ in 2008-2009, but the rate of increase was beginning to slow. Annual soluble reactive phosphorus (SRP) concentrations increased from 101 μg l⁻¹ in the mid 1960s to a maximum of 190 μg l⁻¹ in 1989. In 2002, there was a step reduction in SRP concentration (average=88 μg l⁻¹ in 2002-2005), with further improvement in 2007-2009 (average=49 μg l⁻¹), due to the introduction of phosphorus stripping at sewage treatment works. Phosphorus and nitrate concentrations showed clear annual cycles, related to the timing of inputs from the catchment, and within-stream bioaccumulation and release. Annual depressions in silicon concentration each spring (due to diatom proliferation) reached a maximum between 1980 and 1991, (the period of maximum SRP concentration) indicating that algal biomass had increased within the river. The timing of these silicon depressions was closely related to temperature. Excess carbon dioxide partial pressures (EpCO₂) of 60 times atmospheric CO₂ were also observed through the winter periods from 1980 to 1992, when phosphorus concentration was greatest, indicating very high respiration rates due to microbial decomposition of this enhanced biomass. Declining phosphorus concentrations since 2002 reduced productivity and algal biomass in the summer, and EpCO₂ through the winter, indicating that sewage treatment improvements had improved riverine ecology. Algal blooms were limited by phosphorus, rather than silicon concentration. The value of long-term water quality data sets is discussed. The data from this monitoring programme are made freely available to the wider science community through the CEH data portal (http://gateway.ceh.ac.uk/).

Single–particle spectra of Λ and Σ hypernuclei are calculated within a relativistic mean–field theory. The hyperon couplings used are compatible with the Λ binding in saturated nuclear matter, neutron-star masses and experimental data on Λ levels in hypernuclei. Special attention is devoted to the spin-orbit potential for the hyperons and the influence of the ρ-meson field (isospin dependent interaction). PACS number: 21.80.+a

A generative or latent variable model corresponds to a Bayesian network where arcs point from (presumed) hidden sources to observed variables. In this paper we introduce a particular generative model with binary valued hidden sources (i. e. each source can be either ‘on’ or ‘off’) but continuous observable variables. The purpose of this model is to learn a binary, distributed code for continuous data, for example to learn a bit code for gray value image data. For inference we rely on a mean field approximation. A novel and surprisingly simple derivation of general mean field equations is given. The structure of our model is chosen such that it is optimally suited for the structure of mean field inference. Hence, the mean field equations can be solved efficiently even with a few hundred hidden nodes, thus allowing one to learn highly distributed codes. For learning the parameters in the generative model from data, an appropriate EM-procedure is derived. In the second part of the paper we employ our approach to learning a sparse representation of natural images which is applied to code and to denoise images. The image compression rate is comparable to JPEG-coding and image denoising clearly outperforms standard methods such as Wiener filtering. In the outlook we present potential further directions of research in particular with respect to a more complex hidden topography.

Customer Relationship Management has become mandatory for large and medium scale enterprises. Many companies have already made significant investments to collect data in a customer centric way such that all information regarding a customer is accessible in one place independent of channel (visit, telephone, web,... ), action (buy, complaint, information gathering,... ), department or region. This customer centric organization of data is mostly exploited in sales and services where it allows the agent to quickly obtain all information regarding a customer,for example when the customer calls

We present a systematic approach to mean-field theory (MFT) in a general probabilistic setting without assuming a particular model. The mean-field equations derived here may serve as a local, and thus very simple, method for approximate inference in probabilistic models such as Boltzmann machines or Bayesian networks. Our approach is 'model-independent' in the sense that we do not assume a particular type of dependences; in a Bayesian network, for example, we allow arbitrary tables to specify conditional dependences. In general, there are multiple solutions to the mean-field equations. We show that improved estimates can be obtained by forming a weighted mixture of the multiple mean-field solutions. Simple approximate expressions for the mixture weights are given. The general formalism derived so far is evaluated for the special case of Bayesian networks. The benefits of taking into account multiple solutions are demonstrated by using MFT for inference in a small and in a very large Bayesian network. The results are compared with the exact results.