Browsing by Title

Sort by: Order: Results:

Now showing items 41-60 of 888
  • Parikka, Kirsti (Helsingin yliopisto, 2007)
    The first synthesis of long chain 5-n-alkylresorcinols (C15-C25) in whole grains and whole grain products by a novel modification of Wittig reaction is described. 5-n-Alkylresorcinols are phenolic lipids that have various effects on biological systems, such as antioxidant activity and interaction with biological membranes. These compounds are considered as biomarkers of whole grain intake, which is connected with reduced risk of cardiovascular diseases and certain cancers. Novel hapten derivatives of 5-n-alkylresorcinols, potential compounds for immunoanalytical techniques, are prepared by the same procedure utilizing microwave catalysed aqueous Wittig reaction as the key step. The synthesised analogues are required by various analytical, metabolism and bioactivity investigations. Four alternative strategies for producing deuterium polylabelled 5-n-alkylresorcinols are explored. Ring-labelled D3-alkylresorcinols were synthesized by acidic H/D exchange. Side chain -labelled D4-derivative was prepared by a total synthesis approach utilizing D2 deuterogenation of a D2-alkene derivative, and deuterogenation of alkynes was investigated in another total synthesis approach. An -D3-labelled alkylresorcinol is isotopically pure and completely stable under all relevant conditions encountered during analytical work. The labelling of another phenolic component of whole grains was explored. The preparation of D3-ferulic acid and related compounds by way of selective methylation of the precursors is described. The deuterated compounds are useful as standards in the quantification of these natural products in various substances, such as food and human fluids. The pure 5-n-alkylresorcinol analogues prepared were used in in vitro experiments on alkylresorcinol antioxidant activity and antigenotoxicity. The in vitro experiments show that alkylresorcinols act as antioxidants, especially when incorporated into biological systems, but possess lower activity in chemical tests (FRAP and DPPH assay). Whole grain alkylresorcinols are shown for the first time to have a protective effect against copper induced oxidation of LDL, and H2O2 or genotoxic faecal water induced damage on HT29 cells.
  • Laitinen, Totti (Helsingin yliopisto, 2013)
    This thesis is based on the construction of a two-step laser desorption-ionization aerosol time-of-flight mass spectrometer (laser AMS), which is capable of measuring 10 to 50 nm aerosol particles collected from urban and rural air at-site and in near real time. The operation and applicability of the instrument was tested with various laboratory measurements, including parallel measurements with filter collection/chromatographic analysis, and then in field experiments in urban environment and boreal forest. Ambient ultrafine aerosol particles are collected on a metal surface by electrostatic precipitation and introduced to the time-of-flight mass spectrometer (TOF-MS) with a sampling valve. Before MS analysis particles are desorbed from the sampling surface with an infrared laser and ionized with a UV laser. The formed ions are guided to the TOF-MS by ion transfer optics, separated according to their m/z ratios, and detected with a micro channel plate detector. The laser AMS was used in urban air studies to quantify the carbon cluster content in 50 nm aerosol particles. Standards for the study were produced from 50 nm graphite particles, suspended in toluene, with 72 hours of high power sonication. The results showed the average amount of carbon clusters (winter 2012, Helsinki, Finland) in 50 nm particles to be 7.2% per sample. Several fullerenes/fullerene fragments were detected during the measurements. In boreal forest measurements, the laser AMS was capable of detecting several different organic species in 10 to 50 nm particles. These included nitrogen-containing compounds, carbon clusters, aromatics, aliphatic hydrocarbons, and oxygenated hydrocarbons. A most interesting event occurred during the boreal forest measurements in spring 2011 when the chemistry of the atmosphere clearly changed during snow melt. On that time concentrations of laser AMS ions m/z 143 and 185 (10 nm particles) increased dramatically. Exactly at the same time, quinoline concentrations in molecular clusters measurements (APi-TOFMS) decreased markedly. With the help of simultaneously collected 30 nm filter samples, laser AMS ions m/z 143 and 185 were later identified as originating from 1-(X-methylquinolin-X-yl)ethanone.
  • Pesonen, Janne (Helsingin yliopisto, 2001)
  • Bućko, Michał (Helsingin yliopisto, 2012)
    Road traffic is at present one of the major sources of environmental pollution in urban areas. Magnetic particles, heavy metals and others compounds generated by traffic can greatly affect ambient air quality and have direct implications for human health. The general aim of this research was to identify and characterize magnetic vehicle-derived particulates using magnetic, geochemical and micro-morphological methods. A combination of three different methods was used to discriminate sources of particular anthropogenic particles. Special emphasis was placed on the application of various collectors (roadside soil, snow, lichens and moss bags) to monitor spatial and temporal distribution of traffic pollution on roadsides. The spatial distribution of magnetic parameters of road dust accumulated in roadside soil, snow, lichens and moss bags indicates that the highest concentration of magnetic particles is in the sampling points situated closest to the road edge. The concentration of magnetic particles decreases with increasing distance from the road indicating vehicle traffic as a major source of emission. Significant differences in horizontal distribution of magnetic susceptibility were observed between soil and snow. Magnetic particles derived from road traffic deposit on soil within a few meters from the road, but on snow up to 60 m from the road. The values of magnetic susceptibility of road dust deposited near busy urban motorway are significantly higher than in the case of low traffic road. These differences are attributed to traffic volume, which is 30 times higher on motorway than on local road. Moss bags placed at the edge of urban parks situated near major roads show higher values of magnetic susceptibility than moss bags from parks located near minor routes. Enhanced concentrations of heavy metals (e.g. Fe, Mn, Zn, Cu, Cr, Ni and Co) were observed in the studied samples. This may be associated with specific sources of vehicle emissions (e.g. exhaust and non-exhaust emissions) and/or grain size of the accumulated particles (large active surface of ultrafine particles). Significant correlations were found between magnetic susceptibility and the concentration of selected heavy metals in the case of moss bags exposed to road traffic. Low-coercivity magnetite was identified as a major magnetic phase in all studied roadside collectors (soil, snow, moss bags and lichens). However, magnetic minerals such as titanomagnetite, ilmenite, pyrite and pyrrhotite were also observed in the studied samples. The identified magnetite particles are mostly pseudo-single-domain (PSD) with a predominant MD fraction (>10 μm). The ultrafine iron oxides (>10 nm) were found in road dust extracted from roadside snow. Large magnetic particles mostly originate from non-exhaust emissions, while ultrafine particles originate from exhaust emissions. The examined road dust contains two types of anthropogenic particles: (1) angular/aggregate particles composed of various elements (diameter ~1-300 µm); (2) spherules (~1-100 µm) mostly composed of iron. The first type of particles originates from non-exhaust emissions such as the abrasion of vehicle components, road surface and winter road maintenance. The spherule-shaped particles are products of combustion processes e.g. combustion of coal in nearby power plants and/or fuel in vehicle engines. This thesis demonstrates that snow is an efficient collector of anthropogenic particles, since it can accumulate and preserve the pollutants for several months (until the late stages of melting). Furthermore, it provides more information about spatial and temporal distribution of traffic-generated magnetic particles than soil. Since the interpretation of data obtained from magnetic measurements of soil is problematic (due to its complexity), this suggests the application of alternative collectors of anthropogenic magnetic particulates (e.g. snow and moss bags). Moss bags and lichens are well suited for magnetic biomonitoring studies, since they effectively accumulate atmospheric pollution and can thus be applied to monitor the spatio-temporal distribution of pollution effects.
  • Välimäki, Niko (Helsingin yliopisto, 2012)
    Recent advancements in the field of compressed data structures create interesting opportunities for interdisciplinary research and applications. Compressed data structures provide essentially a time--space tradeoff for solving a problem; while traditional data structures use extra space in addition to the input, compressed data structures replace the input and require space proportional to the compressed size of the input. The amount of available memory is often fixed, thus, the user might be willing to spend more time if it allows the use of larger inputs. However, despite the potential behind compressed data structures, they have not quite reached the audience of other disciplines. We study how to take advantage of compressed data structures in the fields of bioinformatics, data analysis and information retrieval. We present several novel applications for compressed data structures and include an experimental evaluation of the time--space tradeoffs achieved. More precisely, we propose (i) a space-efficient string mining algorithm to recognise substrings that admit the given frequency constraints, (ii) both theoretical and practical methods for computing approximate overlaps between all string pairs, (iii) a practical path-based graph kernel for predicting the function of unknown enzymatic reactions, and (iv) a compressed XML index that supports efficient XPath queries on both the tree-structure and textual content of XML documents. Problem (i) is motivated by knowledge discovery in databases, where the goal is to extract emerging substrings that discriminate two (or more) databases. Problem (ii) is one of the first phases in a sequence assembly pipeline and requires efficient algorithms due to the new high-throughput sequencing systems. Problem (iii) is motivated by machine learning, where kernels are used to measure the similarity of complex objects. Problem (iv) has its background in information retrieval. The proposed methods achieve theoretical and practical improvements over the earlier state of the art. To raise the overall awareness of compressed data structures, our results have been published in interdisciplinary forums, including conferences and journals from the fields of bioinformatics, data engineering and data mining.
  • Zahabi, Seyedali (Helsingin yliopisto, 2013)
    This thesis investigates different aspects of conformal field theory and string theory and their applications in statistical properties of systems. First, we study the free fermions in planar Ising model and its scaling limit at criticality. On the one hand, we examine the relation between the transfer matrix formalism and discrete holomorphicity. We show that the fermion operators of the Ising model satisfy a complexification of the defining relations of s-holomorphicity, a strong notion of discrete holomorphicity, and examples of fermion correlation functions are shown to reproduce s-holomorphic parafermionic observables. On the other hand, we study the relation between fermionic conformal field theory and Schramm Loewner evolution by focusing on the interfaces and fermionic correlation functions of the Ising model. We demonstrate an explicit, rigorous realization of the CFT/SLE correspondence in the case of Ising model. Second, we develop a statistical framework for bosonic string theory in order to study transport properties of black holes in the context of membrane paradigm. We find that the shear viscosity of a highly excited bosonic string is equal to that of black hole horizon up to a numerical factor.
  • Markkanen, Tommi (Helsingin yliopisto, 2014)
    Cosmic inflation is a phase of accelerating, nearly exponential expansion of the spacetime fabric of the Universe, which is assumed to have taken place almost immediately after the Big Bang. Inflation possesses the appealing property that it provides solutions to deep cosmological problems, such as the flatness and horizon problems, and also gives a natural origin for the formation of the large scale structures we observe today. In this thesis we set out to investigate the role quantum corrections play for some simple models where inflation is driven by a single scalar field. It is essential that here the quantum corrections are calculated via curved space field theory. In this technique one quantizes only the matter fields, the dynamics of which take place on a curved classical background. This approach is rarely used in mainstream cosmology and it has the benefit that it allows the quantum fluctuations to back-react on classical Einsteinian gravity. The curved space quantum corrections are studied first in the effective action formalism via the Schwinger-DeWitt expansion and then by constructing effective equations of motion by using the slow-roll technique. We also focus on consistent renormalization and show how to renormalize the effective equations of motion without any reference to an effective action for an interacting theory in curved spacetime. Due to a potential infrared enhancement in effective equations in quasi-de Sitter space, we also perform a resummation of Feynman diagrams in curved non-static space and observe that it regulates the infrared effects. Concerning implications for actual inflationary models, we focus on chaotic type models and observe the quantum corrections to be insignificant, but nevertheless to have theoretically a non-trivial structure.
  • Vepsäläinen, Mikko (Helsingin yliopisto, 2007)
    When heated to high temperatures, the behavior of matter changes dramatically. The standard model fields go through phase transitions, where the strongly interacting quarks and gluons are liberated from their confinement to hadrons, and the Higgs field condensate melts, restoring the electroweak symmetry. The theoretical framework for describing matter at these extreme conditions is thermal field theory, combining relativistic field theory and quantum statistical mechanics. For static observables the physics is simplified at very high temperatures, and an effective three-dimensional theory can be used instead of the full four-dimensional one via a method called dimensional reduction. In this thesis dimensional reduction is applied to two distinct problems, the pressure of electroweak theory and the screening masses of mesonic operators in quantum chromodynamics (QCD). The introductory part contains a brief review of finite-temperature field theory, dimensional reduction and the central results, while the details of the computations are contained in the original research papers. The electroweak pressure is shown to converge well to a value slightly below the ideal gas result, whereas the pressure of the full standard model is dominated by the QCD pressure with worse convergence properties. For the mesonic screening masses a small positive perturbative correction is found, and the interpretation of dimensional reduction on the fermionic sector is discussed.
  • Tahkokallio, Touko (Helsingin yliopisto, 2008)
    The description of quarks and gluons, using the theory of quantum chromodynamics (QCD), has been known for a long time. Nevertheless, many fundamental questions in QCD remain unanswered. This is mainly due to problems in solving the theory at low energies, where the theory is strongly interacting. AdS/CFT is a duality between a specific string theory and a conformal field theory. Duality provides new tools to solve the conformal field theory in the strong coupling regime. There is also some evidence that using the duality, one can get at least qualitative understanding of how QCD behaves at strong coupling. In this thesis, we try to address some issues related to QCD and heavy ion collisions, applying the duality in various ways.
  • Mykkänen, Anne-Mari (Helsingin yliopisto, 2012)
    In this thesis we use lattice field theory to study different frontier problems in strongly coupled non-Abelian gauge theories, focusing on large-N models and walking technicolor theories. Implementing lattice studies of technicolor theories, we consider the SU(2) gauge theory with two fermions transforming under the adjoint representation, which constitutes one of the candidate theories for technicolor. The early lattice Monte Carlo studies of this model have used an unimproved Wilson fermion formulation. However, large lattice cutoff effects can be expected with the unimproved formulation, and so we present the calculation of the O(a) improved lattice Wilson-clover action. In addition to the adjoint representation fermions, we also determine the improvement coefficients for SU(2) gauge theory with two fundamental representation fermions. In another work, we study the deconfined phase of strongly interacting matter, investigating Casimir scaling and renormalization properties of Polyakov loops in different irreducible representations, in SU(N) gauge theories at finite temperature. We study the approach to the large-N limit by performing lattice simulations of Yang-Mills theories with gauge groups from SU(2) to SU(6), taking the twelve lowest irreducible representations for each gauge group into consideration. We find clear evidence of Casimir scaling and identify the temperature dependence of the renormalized Polyakov loops. The third study I present is related to the long-standing idea of non-Abelian gauge theories having a close relation to some kind of string theory. In the confining regime of SU(N) gauge theories, the flux lines between well separated color sources are expected to be squeezed in a thin, stringlike tube, and the interaction between the sources can be described by an effective string theory. One of the consequences of the effective string description at zero temperature is the presence of the Luescher term - a Casimir effect due to the finiteness of the interquark distance - in the long distance interquark potential. To study the validity of this effective model, we compute the static quark potential in SU(3) and SU(4) Yang-Mills theories through lattice simulations, generalizing an efficient `multilevel' algorithm proposed by Luescher and Weisz to an improved lattice action.
  • Collin, Anssi (Helsingin yliopisto, 2006)
  • Niinimäki, Teppo (Helsingin yliopisto, 2015)
    Bayesian networks are probabilistic graphical models, which can compactly represent complex probabilistic dependencies between a set of variables. Once learned from data or constructed by some other means, they can both give insight into the modeled domain and be used for probabilistic reasoning tasks, such as prediction of future data points. Learning a Bayesian network consists of two tasks: discovering a graphical dependency structure on variables, and finding the numerical parameters of a conditional distribution for each variable. Structure discovery has attracted considerable interest in the recent decades. Attention has mostly been paid to finding a structure that best fits the data under certain criterion. The optimization approach can lead to noisy and partly arbitrary results due to the uncertainty caused by a small amount of data. The so-called full Bayesian approach addresses this shortcoming by learning the posterior distribution of structures. In practice, the posterior distribution is summarized by constructing a representative sample of structures, or by computing marginal posterior probabilities of individual arcs or other substructures. This thesis presents algorithms for the full Bayesian approach to structure learning in Bayesian networks. Because the existing exact algorithms only scale to small networks of up to about 25 variables, we investigate sampling based, Monte Carlo methods. The state-of-the-art sampling algorithms draw orderings of variables along a Markov chain. We propose several improvements to this algorithm. First, we show that sampling partial orders instead of linear orders can lead to radically improved mixing of the Markov chain and consequently better estimates. Second, we suggest replacing Markov chain Monte Carlo by annealed importance sampling. This can further improve the accuracy of estimates and has also other advantages such as independent samples and easy parallelization. Third, we propose a way to correct the bias that is caused by sampling orderings of variables instead of structures. Fourth, we present an algorithm that can significantly speed up per-sample computations via approximation. In addition, the thesis proposes a new algorithm for so-called local learning of the Bayesian network structure. In local learning the task is to discover the neighborhood of a given target variable. In contrast to previous algorithms that are based on conditional independence tests between variables, our algorithm gives scores to larger substructures. This approach often leads to more accurate results.
  • Köster, Urs (Helsingin yliopisto, 2009)
    What can the statistical structure of natural images teach us about the human brain? Even though the visual cortex is one of the most studied parts of the brain, surprisingly little is known about how exactly images are processed to leave us with a coherent percept of the world around us, so we can recognize a friend or drive on a crowded street without any effort. By constructing probabilistic models of natural images, the goal of this thesis is to understand the structure of the stimulus that is the raison d etre for the visual system. Following the hypothesis that the optimal processing has to be matched to the structure of that stimulus, we attempt to derive computational principles, features that the visual system should compute, and properties that cells in the visual system should have. Starting from machine learning techniques such as principal component analysis and independent component analysis we construct a variety of sta- tistical models to discover structure in natural images that can be linked to receptive field properties of neurons in primary visual cortex such as simple and complex cells. We show that by representing images with phase invariant, complex cell-like units, a better statistical description of the vi- sual environment is obtained than with linear simple cell units, and that complex cell pooling can be learned by estimating both layers of a two-layer model of natural images. We investigate how a simplified model of the processing in the retina, where adaptation and contrast normalization take place, is connected to the nat- ural stimulus statistics. Analyzing the effect that retinal gain control has on later cortical processing, we propose a novel method to perform gain control in a data-driven way. Finally we show how models like those pre- sented here can be extended to capture whole visual scenes rather than just small image patches. By using a Markov random field approach we can model images of arbitrary size, while still being able to estimate the model parameters from the data.
  • Johansson, Mikael (Helsingin yliopisto, 2007)
    Quantum effects are often of key importance for the function of biological systems at molecular level. Cellular respiration, where energy is extracted from the reduction of molecular oxygen to water, is no exception. In this work, the end station of the electron transport chain in mitochondria, cytochrome c oxidase, is investigated using quantum chemical methodology. Cytochrome c oxidase contains two haems, haem a and haem a3. Haem a3, with its copper companion, CuB, is involved in the final reduction of oxygen into water. This binuclear centre receives the necessary electrons from haem a. Haem a, in turn, receives its electrons from a copper ion pair in the vicinity, called CuA. Density functional theory (DFT) has been used to clarify the charge and spin distributions of haem a, as well as changes in these during redox activity. Upon reduction, the added electron is shown to be evenly distributed over the entire haem structure, important for the accommodation of the prosthetic group within the protein. At the same time, the spin distribution of the open-shell oxidised state is more localised to the central iron. The exact spin density distribution has been disputed in the literature, however, different experiments indicating different distributions of the unpaired electron. The apparent contradiction is shown to be due to the false assumption of a unit amount of unpaired electron density; in fact, the oxidised state has about 1.3 unpaired electrons. The validity of the DFT results have been corroborated by wave function based coupled cluster calculations. Point charges, for use in classical force field based simulations, have been parameterised for the four metal centres, using a newly developed methodology. In the procedure, the subsystem for which point charges are to be obtained, is surrounded by an outer region, with the purpose of stabilising the inner region, both electronically and structurally. Finally, the possibility of vibrational promotion of the electron transfer step between haem a and a3 has been investigated. Calculating the full vibrational spectra, at DFT level, of a combined model of the two haems, revealed several normal modes that do shift electron density between the haems. The magnitude of the shift was found to be moderate, at most. The proposed mechanism could have an assisting role in the electron transfer, which still seems to be dominated by electron tunnelling.