Parallell interacting MCMC for learning of topologies of graphical models

Automated statistical learning of graphical models from data has attained a considerable degree of interest in the machine learning and related literature. Many authors have discussed and/or demonstrated the need for consistent stochastic search methods that would not be as prone to yield locally optimal model structures as simple greedy methods. However, at the same time most of the stochastic search methods are based on a standard Metropolis–Hastings theory that necessitates the use of relatively simple random proposals and prevents the utilization of intelligent and efficient search operators. Here we derive an algorithm for learning topologies of graphical models from samples of a finite set of discrete variables by utilizing and further enhancing a recently introduced theory for non-reversible parallel interacting Markov chain Monte Carlo-style computation. In particular, we illustrate how the non-reversible approach allows for novel type of creativity in the design of search operators. Also, the parallel aspect of our method illustrates well the advantages of the adaptive nature of search operators to avoid trapping states in the vicinity of locally optimal network topologies.     

Development of the tracer-pulse method for adsorption studies of analyte mixtures in liquid chromatography utilizing mass spectrometric detection

The tracer-pulse method provides the real adsorption data points directly from simple, straightforward calculations and is therefore a superior method for multicomponent adsorption isotherm determination in HPLC. Only one important problem has restricted its use so far: the tracer peaks are invisible using any conventional detection principle. We present a solution to this problem with an approach with a firm base in analytical chemistry, utilizing stable isotopes and mass spectrometric detection. The new approach was used for the determination of binary adsorption isotherms, and a systematic investigation was made of its main sources of error. With this modification, the tracer method can be a prime choice for future characterizations of multicomponent separation systems and of competitive drug binding studies.     

Adhesion characteristics of nitrifying bacteria in activated sludge

Nitrifiers are known to form relatively dense and strong microcolonies in activated sludge flocs, but little is known about their adhesion characteristics and how these are relative to other floc components. The size distribution of ammonia and nitrite-oxidizing bacteria (Nitrosomonas oligotropha and Nitrospira spp.) in activated sludge from a nutrient removal plant showed that the majority of N. oligotropha cells formed microcolonies with a diameter of 13-22.5 microm, and most Nitrospira spp. cells formed microcolonies with a diameter of 9-22.5 microm. By applying high shear forces (2200 s(-1)), the largest microcolonies of N. oligotropha fragmented to a level well below the Kolmogorov microscale (approx. 15-25 microm). Only very little erosion of single cells took place. Nitrospira spp. microcolonies were generally slightly stronger than N. oligotropha. In order to characterize the adhesion/binding mechanisms for the individual microcolonies, a number of different physico-chemical treatments were combined with shear, and even though this did not lead to any explicit characterization of the species-specific adhesion mechanisms, entanglement of extracellular polymers was proposed as a plausible important adhesion mechanism. When compared to other floc components, the deflocculated fractions of N. oligotropha and Nitrospira spp. were much lower than those of cells in general (total cell count, DAPI) or the organic matter. Deflocculation of N. oligotropha ranged from 3% to 11% of the total N. oligotropha biovolume, Nitrospira spp. from 1% to 4% of the total Nitrospira spp. biovolume, whereas the number of deflocculated cells was 9-54% of the total cell count, and the deflocculated organic matter constituted 8-43% of the total amount of organic matter. These results clearly showed that activated sludge contained a large pool of loosely attached cells and extracellular polymeric substances, and that the nitrifiers and some other microcolony formers were very strong and remained almost intact even under extreme physical and chemical conditions.     

The role of interparticle and external forces in nanoparticle assembly

The past 20 years have witnessed simultaneous multidisciplinary explosions in experimental techniques for synthesizing new materials, measuring and manipulating nanoscale structures, understanding biological processes at the nanoscale, and carrying out large-scale computations of many-atom and complex macromolecular systems. These advances have led to the new disciplines of nanoscience and nanoengineering. For reasons that are discussed here, most nanoparticles do not 'self-assemble' into their thermodynamically lowest energy state, and require an input of energy or external forces to 'direct' them into particular structures or assemblies. We discuss why and how a combination of self- and directed-assembly processes, involving interparticle and externally applied forces, can be applied to produce desired nanostructured materials.     

Transport of C-labelled PCB compounds from sediment to water and from water to air in laboratory model systems

Tetra-, hexa- and octachlorobiphenyls were added to aquatic model systems composed of undisturbed sediment cores with an overlying water phase. Using impactor plates transport of the compounds from sediment to air was observed. About one per cent of the sediment-bound PCBs recovered in the systems left the water by jet drops from bursting bubbles. The transport of PCBs from the sediment to the air was nearly constant over time, with a transport rate of 0.62 μg · dm⁻² week⁻¹ for tetrachlorobiphenyl. Tetrachlorobiphenyl was mobile in systems with and without macroinvertebrates and in those fixed with HgCl₂. Hexa- and octachlorobiphenyls were transported from sediment to water mainly by bioturbation processes. The two latter substances had a higher adsorption to particles than tetrachlorobiphenyl. Compared to tetrachlorobiphenyl, more hexa- and octachlorobiphenyls accumulated in chironomids and tubificids.     

Morphology effects in MgH2 anode for lithium ion batteries

In order to acquire reproducible electrodes relevant for Li ion batteries new MgH₂ electrodes are successfully obtained from homogenous slurries in N-Methyl-2-Pyrrolidone “NMP” solvent. The electrodes are aimed to elucidate the contribution of the cell components to the electrochemical cycling, in terms of morphology and composition. Various electrode preparations were tested and compared regarding their interaction with Li in a half-cell. The obtained electrochemical cycling curves are discussed according to the ball-milling-induced structural morphology changes and presence of carbon additives, along with the effect of the kinetic rate on the conversion reaction mechanism.     

Determination of hydrophobicity of NOM by RP-HPLC,and the effect of pH and ionic strength

A reverse phase high performance liquid chromatography method for the determination of the hydrophobicity of natural organic matter (NOM) was developed by modification of an existing method based on elution with acetic acid/acetonitrile. The modified method separates the NOM into two well-defined groups according to their retention on a C18 column. The recovery was 101±11% for samples containing between 4 and 40 mg/L dissolved organic carbon (DOC). The method was tested for effects of ionic strength up to 0.8 eq. /L. Testing the method on eleven well-described NOM samples at pH 4.7 and 7.0 revealed the strong pH dependency of the hydrophobicity. The effect of pH on the hydrophobicity could be predicted from the proton binding properties of the NOM. Application of the method to NOM samples fractionated by HPSEC demonstrated considerable variations in hydrophobicity among the fractions, particularly at low pH.