Automatic Global Data Partitioning for Distributed Memory Machines (DMMs) is a difficult problem. Distributed memory machines are scalable, but since the memory is distributed across processors, the scheme of placement of data (arrays) onto local memories of different processors become crucial since any communication between processors for non-local data access is an order of magnitude costlier than access to local memory. Researchers have given varied solutions to this problem, most of which work for uniform dependences in loops and they suggest HPF-like distributions only. For non-uniform dependences the loop was made to run sequentially. In this work, we present a partitioning strategy called Hyperplane Partitioning which works well with loops with non-uniform dependences also. In this method of partitioning, the iteration space is partitioned into as many number of partitions as there are number of logical processors, in such a way that the overall inter-processor communication will be minimum. The idea is to localize as many as dependences as possible so that overall communication both beacuse of non-local data as well as inter-processor synchronizations are reduced. These partitions are then induced into data spaces of the arrays referenced in the loop. Each processor then runs its part of iteration space keeping the data partition that it owns locally. Any non-local data access is implemented by inter-processor communication at run-time.The Hyperplane Partitioning is also extended to a sequence of loops. This is done by first finding Best Local Distribution (BLD) for every loop first and then finding the best way of grouping different adjacent loops (just for finding the data partition) which gives best global data partition. This sequence of distributions/redistributions is found by constructing a data structure called Data Distribution Tree (DDT) and finding the least cost path from the source to any of the leaf nodes in the DDT. The costs for the edges come from the communication cost incurred while running a loop with a particular distribution and redistribution to suit the requirement at the next loop. For this a communication cost estimator is developed which works well for fewer dimensions. To handle complete programs we use some heuristic to find the best global distribution for the entire program.Some optimizations like message optimization to reduce the number of messages sent across processors, time optimization which is done by uniform scheduling across processors, and space optimization to keep only the part of array space that any processor owns onto its local memory, are studied. Hyperplane Partitioning is also implemented using an algorithm for synchronization to handle non-local memory access as well as obeying data dependence constraints. The algorithm is also proved to be correct. The target machine is IBM-SP2 using PVM for the message passing library. The performance of the tool on some standard benchmarks (ADI and RHS) and also on some programs designed by us to show the specific merits of the tool. The results show that the loops which have non-uniform dependences also can be run on DMM with good speed-ups.

Thesis (Ph. D.)--University of Washington, 1997. / Vita. Includes bibliographical references (leaves [100]-113).

No description available.

The operational profile-based testing is currently not a well-studied topic and there are no specific instructions for writing test cases for testing the program. In our thesis, we presented our idea about on the basis of what data test cases can be written. Also, in order to show the effectiveness of operational profile-based testing, we additionally described the equivalent partitioning testing technique. The software for this experiment was taken from the open-source SIR repository. We have selected software: “Account”, that was tested by two different testing methods. The test results of both techniques were compared and it was determined that the operational profile-based testing technique is more effective.

No description available.

Ein Großteil der Masse des troposphärischen Aerosols wird vielerorts durch organische Aerosolbestandteile gebildet. Ferner bestehen feine Aerosolpartikel bis zur Hälfte aus sekundär gebildetem organischen Aerosol (SOA). Jedoch wird die SOA-Massenkonzentration derzeit häufig durch numerische Modelle unterschätzt. Deshalb war das Ziel der vorliegenden Arbeit, die Beschreibung der SOA-Bildung durch Gas-zu-Partikel-Konversion in atmosphärischen Modellen zu verbessern, um ein besseres Verständnis der ablaufenden Prozesse und bestimmenden Parameter zu erlangen. Im hierfür verwendeten Boxmodell SPACCIM (SPectral Aerosol Cloud Chemistry Interaction Model) wurde bisher der Massentransfer zwischen der Gasphase und der organischen Partikelphase nicht beschrieben. In der vorliegenden Arbeit wurde deshalb das Modell sowohl um einen absorptiven als auch einen kinetischen Partitionierungsansatz erweitert. Wesentliche Vorteile des kinetischen Partitionierungsansatzes sind zum einen die Berücksichtigung von Eigenschaften der Aerosolphase (wie Größe und Phasenzustand der Partikel) und zum anderen Reaktionen in der Partikelphase mit einzubeziehen. Innerhalb der Arbeit wurde dieser Partitionierungsansatz weiterentwickelt, um die von der Zusammensetzung abhängige Diffusivität der Partikelphase und reversible Partikelreaktionen abbilden zu können. Weiterhin wurde das 3-D Chemie-Transport-Modell COSMO-MUSCAT (COnsortium for Small-scale MOdeling und MUlti-Scale Chemistry Aerosol Transport) mit dem kinetischen Partitionierungsansatz erweitert, um eine verbesserte Beschreibung der SOA-Bildung zu erreichen. Die Bildung von hochoxidierten, wenig flüchtigen organischen Reaktionsprodukten wurde in den jeweiligen Gasphasen-Chemiemechanismen beider Modelle implementiert. Unter Verwendung von SPACCIM mit dem absorptiven Partitionierungsansatz wurden (i) Sensitivitäten und Limitierungen der SOA-Bildung untersucht, (ii) Aerosolkammerstudien der Ozonolyse von α- und β-Pinen simuliert und (iii) der Einfluss von Wandverlusten bestimmt. Detaillierte Sensitivitätsstudien für den kinetischen Partitionierungsansatz machen deutlich, dass die Partikelreaktivität ein Schlüsselparameter ist, mit dem die SOA-Masse flüssiger Aerosolpartikel um den Faktor vier erhöht werden konnte. Weiterhin konnte gezeigt werden, dass die Partikeldiffusivität und, für halbfeste Partikel, die Partikelgröße ebenfalls bestimmende Modellparameter sind. Für Simulationen von Kammerexperimenten der Ozonolyse von α-Pinen mit mäßig schnellen Reaktionen in der Partikelphase konnte eine gute Übereinstimmung mit den gemessenen SOA-Konzentrationen erzielt werden. Dies stellt eine Verbesserung im Vergleich zum absorptiven Partitionierungsansatz dar. Des Weiteren wurde ein Experiment der AIDA-Aerosolkammer unter Verwendung einer monodispersen und einer unimodalen Aerosolverteilung simuliert. Dabei wurde, unter Berücksichtigung von mäßig schnellen oder reversiblen Reaktionen in der Partikelphase, eine gute Übereinstimmung mit den Messergebnissen für beide Verteilungen erzielt. Die Sensitivitätsstudien für den kinetischen Partitionierungsansatz haben jedoch deutlich gezeigt, dass die wichtigsten Modellparameter noch genauer aus Labormessungen charakterisiert werden müssen. Die Modellergebnisse der COSMO-MUSCAT Simulationen mit dem absorptiven Partitionierungsansatz zeigten, dass die Berücksichtigung von hochoxidierten, organischen wenig flüchtigen Reaktionsprodukten zu erhöhten SOA-Massenkonzentrationen führt und die Vaporisierungsenthalpie die SOA-Bildung deutlich beeinflusst. Außerdem wurde die Anwendbarkeit des Modellsystems mit dem kinetischen Partitionierungsansatz gezeigt und mittels einer nicht-reaktiven Partikelphase vergleichbare Ergebnisse zum absorptiven Partitionierungsansatz erzielt. Der Vergleich mit Messungen von 6 Messstationen zeigte, dass COSMO-MUSCAT in der Lage ist den zeitlichen Trend der beobachteten organischen Massenkonzentration abzubilden. Eine detailliertere Analyse für den Messort Melpitz offenbarte jedoch die zeitweise Unterschätzung der Konzentrationen von monoterpenoiden Vorläufersubstanzen insbesondere in den Nachtstunden. Dies könnte die gleichzeitig auftretende Unterschätzung der SOA-Massenkonzentration durch das Modell erklären. Die zusätzliche Berücksichtigung von Partikelphasenreaktionen erhöht zwar die SOA-Massenkonzentration in diesen Fällen, jedoch werden zuverlässige kinetische Labordaten zur Spezifizierung des jetzigen Modells und für weiterführende Studien benötigt. / In many locations, organic matter represents the largest fraction of the aerosol mass, where the aerosol fine fraction consists of up to about one-half of secondary organic aerosol (SOA). However, the SOA mass is still often underpredicted by models. Therefore, this thesis was aimed at an advanced description of SOA formation from gas-to-particle conversion in a parcel model and a 3-D chemistry transport model (CTM) to gain a better process understanding. Since the utilized parcel model SPACCIM (SPectral Aerosol Cloud Chemistry Interaction Model) has not originally comprised gas-to-particle mass transfer, this framework was further developed by both an absorptive and a kinetic partitioning approach. The latter one has the advantages to be based on aerosol bulk properties, so that particle size and phase state are considered, and to include particle reactivity. Moreover, this approach has been enhanced by a composition-dependent particle-phase diffusivity and reversible particle-phase reactions. The 3-D CTM COSMO-MUSCAT (COnsortium for Small-scale MOdeling and MUlti-Scale Chemistry Aerosol Transport) has been improved by the kinetic partitioning approach to achieve an advanced treatment of SOA formation for process studies. For both models, the respective gas-phase chemistry mechanisms were modified to consider the formation of HOMs (highly oxygenated molecules). For SPACCIM using the absorptive partitioning approach, sensitivities and limitations have been explored, aerosol chamber studies for α- and β-pinene ozonolysis have been simulated, and the impact of wall losses have been examined. Using the kinetic partitioning approach, extensive sensitivity studies have revealed the particle-phase reactivity as key parameter, with a 4-times increased SOA mass for liquid particles, followed by the particle-phase bulk diffusivity and, for semi-solid particles, the particle size. Consideration of moderate fast particle-phase reactions for the simulation of α-pinene ozonolysis exhibits an improvement to the absorptive partitioning approach because the formed SOA mass is similar to the measurements. Moreover, an AIDA chamber experiment was simulated for a monodisperse and a unimodal treatment of the aerosol distribution. For both distributions, high similarity to the measurement results have been achieved for moderate fast or reversible particle-phase reactions. Nevertheless, the sensitivity studies have been shown that the kinetic partitioning approach requires more input from laboratory studies for an appropriate characterization of the key model parameters. The model results of COSMO-MUSCAT using the absorptive partitioning approach have shown that the consideration of HOMs increases the simulated SOA mass and the enthalpy of vaporization markedly influences the formation of SOA. Moreover, the applicability of the kinetic partitioning framework has been established and, for a non-reactive particle phase, almost similar results as for the absorptive partitioning approach are achieved. Comparison with measurements from six field sites has revealed that COSMO-MUSCAT can capture the temporal course of the observed organic mass concentrations. A more detailed analysis for the field site at Melpitz has indicated a partly underestimation of measured monoterpenoid precursor concentrations particularly at nighttime, which might caused the coincident underprediction of the SOA mass concentration. The consideration of particle-phase reactions increase the SOA formation in such cases, but for a final conclusion reliable kinetic measurement data are needed as model input.

City-wide travel time prediction in real-time is an important enabler for efficient use of the road network. It can be used in traveler information to enable more efficient routing of individual vehicles as well as decision support for traffic management applications such as directed information campaigns or incident management. 3D speed maps have been shown to be a promising methodology for revealing day-to-day regularities of city-level travel times and possibly also for short-term prediction. In this paper, we aim to further evaluate and benchmark the use of 3D speed maps for short-term travel time prediction and to enable scenario-based evaluation of traffic management actions we also evaluate the framework for traffic flow prediction. The 3D speed map methodology is adapted to short-term prediction and benchmarked against historical mean as well as against Probabilistic Principal Component Analysis (PPCA). The benchmarking and analysis are made using one year of travel time and traffic flow data for the city of Stockholm, Sweden. The result of the case study shows very promising results of the 3D speed map methodology for short-term prediction of both travel times and traffic flows. The modified version of the 3D speed map prediction outperforms the historical mean prediction as well as the PPCA method. Further work includes an extended evaluation of the method for different conditions in terms of underlying sensor infrastructure, preprocessing and spatio-temporal aggregation as well as benchmarking against other prediction methods. / <p>QC 20190531</p>

Integer programming can be used to provide solutionsto complex decision and planning problems occurring in a wide varietyof situations. Applying integer programming to a real life problembasically involves a first phase where a mathematical model isconstructed, and a second phase where the problem described by themodel is solved. While the nature of the challenges involved in therespective two phases differ, the strong relationship between theproperties of models, and which methods that are appropriate for theirsolution, links the two phases. This thesis constitutes of threepapers, of which the third one considers the modeling phase, while thefirst and second one consider the solution phase.   Many applications of column generation yield master problems of setpartitioning type, and the first and second papers presentmethodologies for solving such problems. The characteristics of themethodologies presented are that all successively found solutions arefeasible and integral, where the retention of integrality is a majordistinction from other column generation methods presented in theliterature.   The third paper concerns nurse scheduling and describes the results ofa pilot implementation of a scheduling tool at a Swedish nursing ward.This paper focuses on the practical aspects of modeling and thechallenges of providing a solution to a complex real life problem.

Polyethylene glycol-stabilized lipodisks have emerged as a novel type of lipid-based nanoparticles with high potential as both drug carriers and biomimetic membranes. In this thesis we assess both of these applications, and show how the properties of the lipodisks can be further developed and optimized. Initially, we show that the antimicrobial peptides melittin, alamethicin and magainin 2, in spite of their very different physico-chemical properties and suggested modes of action on membranes, all have high affinity to lipodisks. Using melittin as a model peptide, we confirm a maintained antimicrobial effect of disk-formulated peptides. We also show that melittin dissociates slowly from the disks, resulting in extended drug release and prolonged antibacterial effect. Additionally, we present evidence that the peptide is protected against enzymatic degradation when formulated in the disks. Further, we develop a stable HPLC-MS system with immobilized lipodisks as model membranes. The stability of the system is confirmed by drug partitioning analysis using 15 different drug compounds. We also show how the lipodisk column can be supplemented with cyclooxygenase by in situ incorporation of the protein in the lipodisks. The specific binding of the protein to the disks is confirmed using QCM-D. Finally, by changing the polymer length and applying a new preparation protocol, we have optimized the lipodisks for use as drug carriers and biomimetic membranes. Previous lipodisk studies have been conducted on systems containing PEG-lipids with polymer molecular weights of 2000 or 5000 Da. Also, conventional protocols for the preparation of lipodisks typically require a PEG-lipid concentration of 15 mol% or more. Here we show that stable lipodisks can also be produced using PEG-lipids with a 1000 Da molecular weight polymer and that the use of shorter PEG-lipids dramatically improve the amount of lipodisks that can be immobilized on silica surfaces. Moreover, through the development of a method in which lipid mixtures are sonicated at low temperatures, we produce lipodisks containing as little as 2 mol% PEG-lipid. We present data verifying that these disks are superior to disks with higher PEG-lipid content in terms of their ability to incorporate externally added PEG-lipids functionalized with targeting agents.

The chemical and optical properties of particulate organic compounds remain unclear, which leaves large uncertainties in the estimation of global radiative transfer balance. Gas and find particle (PM2.5) phase formic acid concentrations were measured with online instrumentation during separate one-month studies in the summer of 2010 in Los Angeles (LA), CA, and Atlanta, GA, and the gas-particle partitioning behavior was investigated and compared with that of water-soluble organic compounds (WSOC). The diurnal profiles clearly indicated that the photochemistry production serves as a strong source for the formation of organics, while the correlation between the gas and particle phase suggested that another partitioning route, the aqueous reactions, is also very important. Later, the optical properties of light-absorbing organic compounds were examined. Little is known about the optical importance of light absorbing particulate organic compounds (brown carbon), especially its extent and absorption relative to black carbon throughout the tropospheric column. Mie theory was applied to size-resolved spectrophotometric absorption measurements of methanol and water-extracts from cascade impactor substrates collected at three surface sites around Atlanta, GA, including both urban and rural. These results were applied to similar measurements of brown carbon in extracts from aircraft bulk filter samples collected over central USA. At the surface sites predicted light absorption by brown carbon relative to total absorption (brown carbon plus pure black carbon) was about 10% and 30% at 350 nm, versus 1 and 11% at 450 nm, for water and methanol extracts, respectively. The relative contribution of brown carbon was greater in the free troposphere and significantly increased with altitude. Although this approach has limitations, it demonstrates the ubiquity and significant potential contribution of brown carbon.