Weighted geometric mean of large-scale matrices: numerical analysis and algorithms

Computing the weighted geometric mean of large sparse matrices is an operation that tends to become rapidly intractable, when the size of the matrices involved grows. However, if we are not interested in the computation of the matrix function itself, but just in that of its product times a vector, the problem turns simpler and there is a chance to solve it even when the matrix mean would actually be impossible to compute. Our interest is motivated by the fact that this calculation has some practical applications, related to the preconditioning of some operators arising in domain decomposition of elliptic problems. In this thesis, we explore how such a computation can be efficiently performed. First, we exploit the properties of the weighted geometric mean and find several equivalent ways to express it through real powers of a matrix. Hence, we focus our attention on matrix powers and examine how well-known techniques can be adapted to the solution of the problem at hand. In particular, we consider two broad families of approaches for the computation of f(A) v, namely quadrature formulae and Krylov subspace methods, and generalize them to the pencil case f(A\B) v. Finally, we provide an extensive experimental evaluation of the proposed algorithms and also try to assess how convergence speed and execution time are influenced by some characteristics of the input matrices. Our results suggest that a few elements have some bearing on the performance and that, although there is no best choice in general, knowing the conditioning and the sparsity of the arguments beforehand can considerably help in choosing the best strategy to tackle the problem.

Application of numerical methods to earthquake stability assessment of slopes

Questo documento descrive gran parte del lavoro svolto durante un periodo di studio di sei mesi all’International Centre for Geohazards (ICG) di Oslo. Seguendo la linea guida dettata nel titolo, sono stati affrontati diversi aspetti riguardanti la modellazione numerica dei pendii quali l’influenza delle condizioni al contorno e delle proporzioni del modello, la back-analysis di eventi di scivolamento e l’applicazione delle analisi di stabilità monodimensionali. La realizzazione di semplici modelli con il programma agli elementi finiti PLAXIS (Brinkgreve et al., 2008) ha consentito di analizzare le prestazioni dei modelli numerici riguardo all’influenza delle condizioni al contorno confrontandoli con un calcolo teorico del fattore di amplificazione. Questa serie di test ha consentito di stabilire alcune linee guida per la realizzazione di test con un buon livello di affidabilità. Alcuni case-history, in particolare quello di Las Colinas (El Salvador), sono stati modellati allo scopo di applicare e verificare i risultati ottenuti con i semplici modelli sopracitati. Inoltre sono state svolte analisi di sensitività alla dimensione della mesh e ai parametri di smorzamento e di elasticità. I risultati hanno evidenziato una forte dipendenza dei risultati dai parametri di smorzamento, rilevando l’importanza di una corretta valutazione di questa grandezza. In ultima battuta ci si è occupati dell’accuratezza e dell’applicabilità dei modelli monodimensionali. I risultati di alcuni modelli monodimensionali realizzati con il software Quiver (Kaynia, 2009) sono stati confrontati con quelli ottenuti da modelli bidimensionali. Dal confronto è risultato un buon grado di approssimazione accompagnato da un margine di sicurezza costante. Le analisi monodimensionali sono poi state utilizzate per la verifica di sensitività. I risultati di questo lavoro sono qui presentati e accompagnati da suggerimenti qualitativi e quantitativi per la realizzazione di modelli bidimensionali affidabili. Inoltre si descrive la possibilità di utilizzare modelli monodimensionali in caso d’incertezze sui parametri. Dai risultati osservati emerge la possibilità di ottenere un risparmio di tempo nella realizzazione di importanti indagini di sensitività.

A new insight in stress recovery tecniques and hessian reconstruction methods

Stress recovery techniques have been an active research topic in the last few years since, in 1987, Zienkiewicz and Zhu proposed a procedure called Superconvergent Patch Recovery (SPR). This procedure is a last-squares fit of stresses at super-convergent points over patches of elements and it leads to enhanced stress fields that can be used for evaluating finite element discretization errors. In subsequent years, numerous improved forms of this procedure have been proposed attempting to add equilibrium constraints to improve its performances. Later, another superconvergent technique, called Recovery by Equilibrium in Patches (REP), has been proposed. In this case the idea is to impose equilibrium in a weak form over patches and solve the resultant equations by a last-square scheme. In recent years another procedure, based on minimization of complementary energy, called Recovery by Compatibility in Patches (RCP) has been proposed in. This procedure, in many ways, can be seen as the dual form of REP as it substantially imposes compatibility in a weak form among a set of self-equilibrated stress fields. In this thesis a new insight in RCP is presented and the procedure is improved aiming at obtaining convergent second order derivatives of the stress resultants. In order to achieve this result, two different strategies and their combination have been tested. The first one is to consider larger patches in the spirit of what proposed in [4] and the second one is to perform a second recovery on the recovered stresses. Some numerical tests in plane stress conditions are presented, showing the effectiveness of these procedures. Afterwards, a new recovery technique called Last Square Displacements (LSD) is introduced. This new procedure is based on last square interpolation of nodal displacements resulting from the finite element solution. In fact, it has been observed that the major part of the error affecting stress resultants is introduced when shape functions are derived in order to obtain strains components from displacements. This procedure shows to be ultraconvergent and is extremely cost effective, as it needs in input only nodal displacements directly coming from finite element solution, avoiding any other post-processing in order to obtain stress resultants using the traditional method. Numerical tests in plane stress conditions are than presented showing that the procedure is ultraconvergent and leads to convergent first and second order derivatives of stress resultants. In the end, transverse stress profiles reconstruction using First-order Shear Deformation Theory for laminated plates and three dimensional equilibrium equations is presented. It can be seen that accuracy of this reconstruction depends on accuracy of first and second derivatives of stress resultants, which is not guaranteed by most of available low order plate finite elements. RCP and LSD procedures are than used to compute convergent first and second order derivatives of stress resultants ensuring convergence of reconstructed transverse shear and normal stress profiles respectively. Numerical tests are presented and discussed showing the effectiveness of both procedures.

A numerical study of temperature effects on hot wire measurements inside turbulent channel flows

A way to investigate turbulence is through experiments where hot wire measurements are performed. Analysis of the in turbulence of a temperature gradient on hot wire measurements is the aim of this thesis work. Actually - to author's knowledge - this investigation is the first attempt to document, understand and ultimately correct the effect of temperature gradients on turbulence statistics. However a numerical approach is used since instantaneous temperature and streamwise velocity fields are required to evaluate this effect. A channel flow simulation at Re_tau = 180 is analyzed to make a first evaluation of the amount of error introduced by temperature gradient inside the domain. Hot wire data field is obtained processing the numerical flow field through the application of a proper version of the King's law, which connect voltage, velocity and temperature. A drift in mean streamwise velocity profile and rms is observed when temperature correction is performed by means of centerline temperature. A correct mean velocity pro�le is achieved correcting temperature through its mean value at each wall normal position, but a not negligible error is still present into rms. The key point to correct properly the sensed velocity from the hot wire is the knowledge of the instantaneous temperature field. For this purpose three correction methods are proposed. At the end a numerical simulation at Re_tau =590 is also evaluated in order to confirm the results discussed earlier.

Numerical simulation of interdigitated back contact hetero-junction solar cells

The goal of this thesis is the application of an opto-electronic numerical simulation to heterojunction silicon solar cells featuring an all back contact architecture (Interdigitated Back Contact Hetero-Junction IBC-HJ). The studied structure exhibits both metal contacts, emitter and base, at the back surface of the cell with the objective to reduce the optical losses due to the shadowing by front contact of conventional photovoltaic devices. Overall, IBC-HJ are promising low-cost alternatives to monocrystalline wafer-based solar cells featuring front and back contact schemes, in fact, for IBC-HJ the high concentration doping diffusions are replaced by low-temperature deposition processes of thin amorphous silicon layers. Furthermore, another advantage of IBC solar cells with reference to conventional architectures is the possibility to enable a low-cost assembling of photovoltaic modules, being all contacts on the same side. A preliminary extensive literature survey has been helpful to highlight the specific critical aspects of IBC-HJ solar cells as well as the state-of-the-art of their modeling, processing and performance of practical devices. In order to perform the analysis of IBC-HJ devices, a two-dimensional (2-D) numerical simulation flow has been set up. A commercial device simulator based on finite-difference method to solve numerically the whole set of equations governing the electrical transport in semiconductor materials (Sentuarus Device by Synopsys) has been adopted. The first activity carried out during this work has been the definition of a 2-D geometry corresponding to the simulation domain and the specification of the electrical and optical properties of materials. In order to calculate the main figures of merit of the investigated solar cells, the spatially resolved photon absorption rate map has been calculated by means of an optical simulator. Optical simulations have been performed by using two different methods depending upon the geometrical features of the front interface of the solar cell: the transfer matrix method (TMM) and the raytracing (RT). The first method allows to model light prop-agation by plane waves within one-dimensional spatial domains under the assumption of devices exhibiting stacks of parallel layers with planar interfaces. In addition, TMM is suitable for the simulation of thin multi-layer anti reflection coating layers for the reduction of the amount of reflected light at the front interface. Raytracing is required for three-dimensional optical simulations of upright pyramidal textured surfaces which are widely adopted to significantly reduce the reflection at the front surface. The optical generation profiles are interpolated onto the electrical grid adopted by the device simulator which solves the carriers transport equations coupled with Poisson and continuity equations in a self-consistent way. The main figures of merit are calculated by means of a postprocessing of the output data from device simulation. After the validation of the simulation methodology by means of comparison of the simulation result with literature data, the ultimate efficiency of the IBC-HJ architecture has been calculated. By accounting for all optical losses, IBC-HJ solar cells result in a theoretical maximum efficiency above 23.5% (without texturing at front interface) higher than that of both standard homojunction crystalline silicon (Homogeneous Emitter HE) and front contact heterojuction (Heterojunction with Intrinsic Thin layer HIT) solar cells. However it is clear that the criticalities of this structure are mainly due to the defects density and to the poor carriers transport mobility in the amorphous silicon layers. Lastly, the influence of the most critical geometrical and physical parameters on the main figures of merit have been investigated by applying the numerical simulation tool set-up during the first part of the present thesis. Simulations have highlighted that carrier mobility and defects level in amorphous silicon may lead to a potentially significant reduction of the conversion efficiency.

Numerical and semi-analytical models of sliding masses: application to the 1783 Scilla tsunamigenic landslide

The Scilla rock avalanche occurred on 6 February 1783 along the coast of the Calabria region (southern Italy), close to the Messina Strait. It was triggered by a mainshock of the Terremoto delle Calabrie seismic sequence, and it induced a tsunami wave responsible for more than 1500 casualties along the neighboring Marina Grande beach. The main goal of this work is the application of semi-analtycal and numerical models to simulate this event. The first one is a MATLAB code expressly created for this work that solves the equations of motion for sliding particles on a two-dimensional surface through a fourth-order Runge-Kutta method. The second one is a code developed by the Tsunami Research Team of the Department of Physics and Astronomy (DIFA) of the Bologna University that describes a slide as a chain of blocks able to interact while sliding down over a slope and adopts a Lagrangian point of view. A wide description of landslide phenomena and in particular of landslides induced by earthquakes and with tsunamigenic potential is proposed in the first part of the work. Subsequently, the physical and mathematical background is presented; in particular, a detailed study on derivatives discratization is provided. Later on, a description of the dynamics of a point-mass sliding on a surface is proposed together with several applications of numerical and analytical models over ideal topographies. In the last part, the dynamics of points sliding on a surface and interacting with each other is proposed. Similarly, different application on an ideal topography are shown. Finally, the applications on the 1783 Scilla event are shown and discussed.

Numerical and experimental investigation of structural stiffness influence on ratcheting convection cell in granular soils under cyclic loading

Lateral cyclic loaded structures in granular soils can lead to an accumulation of irreversible strains by changing their mechanical response (densification) and forming a closed convective cell in the upper layer of the bedding. In the present thesis the convective cell dimension, formation and grain migration inside this closed volume have been studied and presented in relation to structural stiffness and different loads. This relation was experimentally investigated by applying a cyclic lateral force to a scaled flexible vertical element embedded in dry granular soil. The model was monitored with a camera in order to derive the displacement field by means of the PIV technique. Modelling large soil deformation turns out to be difficult, using mesh-based methods. Consequently, a mesh-free approach (DEM) was chosen in order to investigate the granular flow with the aim of extracting interesting micromechanical information. In both the numerical and experimental analyses the effect of different loading magnitudes and different dimensions of the vertical element were considered. The main results regarded the different development, shape and dimensions of the convection cell and the surface settlements. Moreover, the Discrete Element Method has proven to give satisfactory results in the modelling of large deformation phenomena such as the ratcheting convective cell.

From the traditional verification approach to the spatial methods: a study of applicability of the SAL metric in the framework of the WMO project MesoVICT

Increasing in resolution of numerical weather prediction models has allowed more and more realistic forecasts of atmospheric parameters. Due to the growing variability into predicted fields the traditional verification methods are not always able to describe the model ability because they are based on a grid-point-by-grid-point matching between observation and prediction. Recently, new spatial verification methods have been developed with the aim of show the benefit associated to the high resolution forecast. Nested in among of the MesoVICT international project, the initially aim of this work is to compare the newly tecniques remarking advantages and disadvantages. First of all, the MesoVICT basic examples, represented by synthetic precipitation fields, have been examined. Giving an error evaluation in terms of structure, amplitude and localization of the precipitation fields, the SAL method has been studied more thoroughly respect to the others approaches with its implementation in the core cases of the project. The verification procedure has concerned precipitation fields over central Europe: comparisons between the forecasts performed by the 00z COSMO-2 model and the VERA (Vienna Enhanced Resolution Analysis) have been done. The study of these cases has shown some weaknesses of the methodology examined; in particular has been highlighted the presence of a correlation between the optimal domain size and the extention of the precipitation systems. In order to increase ability of SAL, a subdivision of the original domain in three subdomains has been done and the method has been applied again. Some limits have been found in cases in which at least one of the two domains does not show precipitation. The overall results for the subdomains have been summarized on scatter plots. With the aim to identify systematic errors of the model the variability of the three parameters has been studied for each subdomain.