A standard procedure for P3HT:PCBM bulk-heterojunction solar cells has been developed. Fabrication conditions, such as environment; solution concentration, thickness of active layer or post-treatment methods are systematically optimized. The best device performance is obtained by slow-drying spin-coated P3HT:PCBM (1:0.8) blend layer with DCB as solvent. C60:LiF composite films with up to 80% LiF concentration as hole blocking layer have been developed to significantly increase power conversion efficiencies of OPV devices. The short-circuit current density is greatly enhanced, without sacrificing open-circuit voltage and fill factor. Due to its superior oxygen diffusion blocking effect, the C60:LiF composite layer also can provide a more effective passivation film than a thin LiF layer, resulting in an impressive enhancement in air stability of devices.

A standard procedure for P3HT:PCBM bulk-heterojunction solar cells has been developed. Fabrication conditions, such as environment; solution concentration, thickness of active layer or post-treatment methods are systematically optimized. The best device performance is obtained by slow-drying spin-coated P3HT:PCBM (1:0.8) blend layer with DCB as solvent. C60:LiF composite films with up to 80% LiF concentration as hole blocking layer have been developed to significantly increase power conversion efficiencies of OPV devices. The short-circuit current density is greatly enhanced, without sacrificing open-circuit voltage and fill factor. Due to its superior oxygen diffusion blocking effect, the C60:LiF composite layer also can provide a more effective passivation film than a thin LiF layer, resulting in an impressive enhancement in air stability of devices.

Tor is a network designed for low-latency anonymous communications. Tor clients form circuits through relays that are listed in a public directory, and then relay their encrypted traffic through these circuits. This indirection makes it difficult for a local adversary to determine with whom a particular Tor user is communicating. Tor may also be used to circumvent regional Internet censorship, since the final hop of a user's connection can be in a different country. In response, some local adversaries restrict access to Tor by blocking each of the publicly listed relays. To deal with such an adversary, Tor uses bridges, which are unlisted relays that can be used as alternative entry points into the Tor network. Unfortunately, issues with Tor's bridge implementation make it easy to discover large numbers of bridges. This makes bridges easy to block. Also, an adversary that hoards this information may use it to determine when each bridge is online over time. If a bridge operator also browses with Tor on the same machine, this information may be sufficient to deanonymize him. We present BridgeSPA as a method to mitigate these issues. A client using BridgeSPA relies on innocuous single packet authorization (SPA) to present a time-limited key to a bridge. Before this authorization takes place, the bridge will not reveal whether it is online. We have implemented BridgeSPA as a working proof-of-concept for GNU/Linux systems. The implementation is available under a free licence. We have integrated our implementation to work in an OpenWRT environment. This enables BridgeSPA support for any client behind a deployed BridgeSPA OpenWRT router, no matter which operating system they are running.

The experiments reported in this thesis studied the contributions of opioid and NMDA receptors to predictive fear learning, as measured by freezing in the rat. The first series of experiments (Chapter 2) used a within-subject one-trial blocking design to study whether opioid receptors mediate a direct action of predictive error on Pavlovian association formation. Systemic administrations of the opioid receptor antagonist naloxone or intra-vlPAG administrations of the selective μ-opioid receptor antagonist CTAP prior to Stage II training prevented one-trial blocking. These results show for the first time that opioid receptors mediate the direct actions of predictive error on Pavlovian association formation. The second series of experiments (Chapter 3) then studied temporal-difference prediction errors during Pavlovian fear conditioning. In Stage I rats received CSA ?? shock pairings. In Stage II they received CSA/CSB ?? shock pairings that blocked learning to CSB. In Stage III, a serial overlapping compound, CSB → CSA, was followed by shock. The change in intra-trial durations supported fear learning to CSB but reduced fear of CSA, revealing the selective operation of temporal-difference prediction errors. This bi-directional change in responding was prevented by systemic NMDA receptor antagonism prior to Stage III training. In contrast opioid receptor antagonism differentially affected the learning taking place during Stage III, enhancing learning to CSB while impairing the loss of fear to CSA. The final series of experiments (Chapter 4) then examined potential neuroanatomical loci for the systemic effects reported in Chapter 3. It was observed that intra-BLA infusion of ifenprodil, an antagonist of NMDA receptors containing the NR2B subunit, prevented all learning during Stage III, whereas intra-vlPAG infusion of the μ-opioid receptor antagonist CTAP facilitated learning to CSB but impaired learning to CSA. These results are consistent with the suggestion that opioid receptors in the vlPAG provide an important contribution to learning. Importantly, this contribution of the vlPAG is over and above its role in producing the freezing conditioned response. Furthermore, the findings of this thesis identify complementary but dissociable roles for amygdala NMDA receptors and vlPAG μ-opioid receptors in predictive fear learning.

The present thesis examined the behavioural and brain mechanisms of associative change in the rat during Pavlovian fear conditioning as measured by freezing. The first series of experiments (Chapter 3) used compound test designs to study how learning is distributed among excitatory and neutral conditional stimuli (CSs). More was learned about a neutral CSB than an excitatory CSA when trained in isolation, indicating that fear learning is negatively accelerated. CSA blocked fear learning to CSB when trained in compound. Unblocking of CSB occurred if the AB compound signalled an increase in unconditional stimulus (US) intensity or number. Assessments of associative change during blocking showed that more was learned about CSB than CSA. Such assessments during unblocking revealed that more was learned about CSB than CSA following an increase in US intensity but not US number. These US manipulations had no differential effects on single-cue learning. The results show that variations in US intensity or number produce unblocking of fear learning, but for each there is a different profile of associative change and a potentially different mechanism. The second series of experiments (Chapter 4) demonstrated that these stimulus selection effects are mediated, at least in part, by nucleus accumbens shell (AcbSh). AcbSh lesions augmented overshadowing during compound conditioning and promoted learning about CSA at the expense of CSB during blocking designs. Lesioned rats could learn normally about the novel CSB if it was rendered more informative regarding shock in Stage II. These results identify an important role for AcbSh and ventral striatum in distributing attention and learning among competing predictors of danger.

Diss. (sammanfattning) Stockholm : Karol. inst., 2003. / Härtill 4 uppsatser.

A possibility of a temporary disconnection of database service exists in many computing environments. It is a common need to permit a participating site to lag behind and re-initialize to full recovery. It is also necessary that active transactions view a globally consistent system state for ongoing operations. We present an algorithm for on-the-fly backup and site-initialization. The technique is non-blocking in the sense that failure and recovery procedures do not interfere with ordinary transactions. As a result the system can tolerate disconnection of services and reconnection of disconnected services, without incurring high overheads

No domínio da morfologia derivacional, é difícil encontrar padrões de formação de palavras que possam se aplicar a todas as bases que se encaixam em seus contextos de aplicação. Isso equivale a dizer que a produtividade de padrões derivacionais costuma ser limitada. Entre as causas dessa limitação, vemos que formações potenciais são frequentemente bloqueadas por itens lexicais já existentes; em outros casos, elas são suplantadas por expressões formadas por padrões derivacionais concorrentes. Este trabalho dedica-se a explorar tais situações de competição. Iniciamos pelo exame da ideia de produtividade e de como entender as diferenças quantitativas entre padrões rivais quanto a este aspecto. Fazemos, em seguida, uma discussão mais detida da competição morfológica e da noção central de bloqueio, contrapondo às teorias gramaticais de base lexical uma abordagem pragmática deste fenômeno. Por fim, apresentamos o modelo desenvolvido em Albright e Hayes (1999) e em trabalhos posteriores, que explora a ideia de que o grau de confiabilidade do emprego de padrões morfológicos em diferentes contextos fonológicos é um fator determinante da produtividade desses padrões, bem como da competição entre eles. Testamos este modelo utilizando dados dos sufixos -ção e -mento, que se encontram em competição há bastante tempo no português. Estes dados provêm do Dicionário Houaiss 3.0 e de um levantamento de textos de jornais e blogs, coletados com o auxílio de programas computacionais desenvolvivdos para este trabalho. Nossos resultados sugerem que a manutenção da produtividade de -mento ao longo da história, mesmo após -ção ter se tornado o padrão dominante de nominalização, foi escorada pela existência de contextos fonológicos em que -mento atinge um alto grau de confiabilidade. Dada a produtividade da primeira conjugação, foram particularmente importantes os contextos de aplicação de -mento encontrados entre palavras desta classe verbal. Com base nestas generalizações, mostramos como um modelo estatístico é capaz de prever, na maior parte dos casos, a escolha entre estes dois afixos diante de uma nova base verbal. / In the field of derivational morphology, it is hard to find word formation patterns that may be applied to every base satisfying its context of application. This means that the productivity of derivational patterns is often limited. Among the causes of this limitation, we find that potential words are blocked by existing lexical items in many cases; in other cases, they are preempted by expressions formed by rival derivational patterns. This work devotes itself to exploring these instances of competition. We start by exploring the concept of productivity and by investigating how to understand quantitative differences between rival patterns in this respect. We then proceeed to a more detailed discussion of morphological competition and the fundamental notion of blocking, comparing a pragmatic approach to this phenomenon with lexicalist grammatical theories. Finally, we present the model of Albright e Hayes (1999) and later works, which explores the idea that the reliability of morphological patterns in different phonological contexts is a key determinant of the productivity of these patterns and the competition between them. We test this model on data formed by the suffixes -ção and -mento, which have been in competition for a long time in Portuguese. These data come from Dicionário Houaiss 3.0 and from a corpus created from newspapers and blogs with the help of software developed for this research. Our results suggest that the continued productivity of -mento throughout history, even after -ção had become the dominant nominalization pattern in the language, was supported by the existence of phonological contexts in which -mento reaches a high degree of reliability. Given the productivity of the first conjugation, contexts of application of -mento in words of this verbal class have shown to be especially important. We show that a statistical model equipped with these generalizations is able to predict the choice between these affixes in most cases.

Many problems in nature can be represented as some kind of a satisfiability problem. Several SAT solvers and SMT solvers have been developed in the last decade with the goal of checking the satisfiability of different SAT problems. An all-solution satisfiability modulo theories on top of the Z3 SMT solver is presented that uses the clause blocking algorithm to find all the solution sets of a SAT problem. Then, an incremental All-SMT solver has been presented based on the all-SMT solver which is able to find the satisfiable answers of an incremental SMT problem based on the solution set of the original problem.

No domínio da morfologia derivacional, é difícil encontrar padrões de formação de palavras que possam se aplicar a todas as bases que se encaixam em seus contextos de aplicação. Isso equivale a dizer que a produtividade de padrões derivacionais costuma ser limitada. Entre as causas dessa limitação, vemos que formações potenciais são frequentemente bloqueadas por itens lexicais já existentes; em outros casos, elas são suplantadas por expressões formadas por padrões derivacionais concorrentes. Este trabalho dedica-se a explorar tais situações de competição. Iniciamos pelo exame da ideia de produtividade e de como entender as diferenças quantitativas entre padrões rivais quanto a este aspecto. Fazemos, em seguida, uma discussão mais detida da competição morfológica e da noção central de bloqueio, contrapondo às teorias gramaticais de base lexical uma abordagem pragmática deste fenômeno. Por fim, apresentamos o modelo desenvolvido em Albright e Hayes (1999) e em trabalhos posteriores, que explora a ideia de que o grau de confiabilidade do emprego de padrões morfológicos em diferentes contextos fonológicos é um fator determinante da produtividade desses padrões, bem como da competição entre eles. Testamos este modelo utilizando dados dos sufixos -ção e -mento, que se encontram em competição há bastante tempo no português. Estes dados provêm do Dicionário Houaiss 3.0 e de um levantamento de textos de jornais e blogs, coletados com o auxílio de programas computacionais desenvolvivdos para este trabalho. Nossos resultados sugerem que a manutenção da produtividade de -mento ao longo da história, mesmo após -ção ter se tornado o padrão dominante de nominalização, foi escorada pela existência de contextos fonológicos em que -mento atinge um alto grau de confiabilidade. Dada a produtividade da primeira conjugação, foram particularmente importantes os contextos de aplicação de -mento encontrados entre palavras desta classe verbal. Com base nestas generalizações, mostramos como um modelo estatístico é capaz de prever, na maior parte dos casos, a escolha entre estes dois afixos diante de uma nova base verbal. / In the field of derivational morphology, it is hard to find word formation patterns that may be applied to every base satisfying its context of application. This means that the productivity of derivational patterns is often limited. Among the causes of this limitation, we find that potential words are blocked by existing lexical items in many cases; in other cases, they are preempted by expressions formed by rival derivational patterns. This work devotes itself to exploring these instances of competition. We start by exploring the concept of productivity and by investigating how to understand quantitative differences between rival patterns in this respect. We then proceeed to a more detailed discussion of morphological competition and the fundamental notion of blocking, comparing a pragmatic approach to this phenomenon with lexicalist grammatical theories. Finally, we present the model of Albright e Hayes (1999) and later works, which explores the idea that the reliability of morphological patterns in different phonological contexts is a key determinant of the productivity of these patterns and the competition between them. We test this model on data formed by the suffixes -ção and -mento, which have been in competition for a long time in Portuguese. These data come from Dicionário Houaiss 3.0 and from a corpus created from newspapers and blogs with the help of software developed for this research. Our results suggest that the continued productivity of -mento throughout history, even after -ção had become the dominant nominalization pattern in the language, was supported by the existence of phonological contexts in which -mento reaches a high degree of reliability. Given the productivity of the first conjugation, contexts of application of -mento in words of this verbal class have shown to be especially important. We show that a statistical model equipped with these generalizations is able to predict the choice between these affixes in most cases.