The goal of this research is to evaluate the secondary prompt gamma (PG) yield from proton therapy at high characteristic energies from MC model simulations and experimental data. Recent studies indicate that target composition influences PG characteristic energy and yield, and the quantification of PG may be used to offer real-time dose verification for proton therapy. In this study PG analysis was performed for MC simulations to evaluate the characteristic measurements and total yield of secondary PG emitted from a target in the 0-8 MeV range from a proton therapy beam over a range of four different beam energies (70 MeV, 160 MeV, 200 MeV, 220 MeV). This was repeated in several target materials (carbon, calcium oxide, calcium fluoride, PMMA, and HDPE) in order to evaluate the influence of the incident energy and the target material on the PG yield and energy spectra. PG energy spectra determined from the specified target materials indicated that the 3.74 MeV energy peak shows a linear correlation between PG intensity and calcium mass fraction of the target material. 6.13 MeV and 4.44 MeV energy peak are not unique to the respective, oxygen and combined oxygen and carbon, mass fraction of the target material. This relationship is complicated by the addition of calcium within the target. Experimental data was collected in order to validate the computational model based on comparison of relative characteristic energy peek intensities. The relative peak ratio determined from experimental data is in good agreement with model prediction, the combined peak ratio is within 0.2%.

Prompts are implemented often in training procedures, to include conditional discriminations, and this can lead to prompt dependency. The current study extends a prior study that suggested that the effectiveness of supplementary visual stimuli displaying the sample and comparison was dependent on the timing in relation to the selection task, presented as a prompt or feedback, in a match-to-sample procedure. The current research examines if the differences in that effectiveness were due to differences in observing behavior in those two conditions. Measures of observing behavior were determined by making access to the individual visual stimuli contingent on clicking on the visual stimulus and keeping the cursor located on the stimulus. Participants viewed the sample comparison much less than the comparison stimulus in both the prompt and feedback conditions. Latency to select the comparison stimulus was much shorter for the prompt condition suggesting that the participants might have interacted differently with the selection task in the two conditions.

Individuals diagnosed with Autism Spectrum Disorder (ASD) may be taught novel skills using various prompting strategies. Some studies have compared effective prompting procedures for these learners. Results have shown that no single procedure has been effective for all learners. Instead, efficiency of each response prompt type varied across participants. Some research suggests that identifying an individual’s preference may help identify an effective treatment. However, it is unknown if response prompt preference corresponds with effectiveness. Thus, the purpose of the present study is to evaluate whether preference corresponds with the most effective response prompt in children with ASD.

A prompt-gamma, neutron-activation analysis facility earlier developed at the Nuclear Science Center of Texas A&M University could not be used successfully to analyze geologic samples due to high detection background, low neutron fluence rate and poor detection equipment . A systematic investigation into the performance capability of a prompt-gamma, neutron activation analysis facility was undertaken in this research project. The facility was reconstructed and used to obtain prompt-gamma spectra of chlorine and cadmium and from the spectra, the net peak area counts for the most intense prompt-gamma-ray energies were obtained. A theoretical model was developed which can predict the net peak area counts expected on these prompt-gamma-ray energies using the thermal neutron fluence rate at the sample position, the absolute efficiency of the detector, and the mass and partial gamma-ray production cross section data for the samples. The experimental and predicted results were compared to establish the performance capability of the reconstructed facility. Good agreements between experimental and predicted results were obtained for chlorine, but results from cadmium showed larger discrepancies due to self-shielding effects. Corrections for self-shielding effects were applied to results from cadmium and the experimental and predicted results were also in good agreement. The satisfactory results indicate that it is possible to implement the prompt-gamma, neutron-activation analysis technique at Beam Port #1 of the Nuclear Science Center Reactor. To be able to obtain excellent results from other samples, improvements in shielding materials to attain a lower detection background and a highly efficient detection system should be incorporated.

A prompt-gamma, neutron-activation analysis facility earlier developed at the Nuclear Science Center of Texas A&M University could not be used successfully to analyze geologic samples due to high detection background, low neutron fluence rate and poor detection equipment . A systematic investigation into the performance capability of a prompt-gamma, neutron activation analysis facility was undertaken in this research project. The facility was reconstructed and used to obtain prompt-gamma spectra of chlorine and cadmium and from the spectra, the net peak area counts for the most intense prompt-gamma-ray energies were obtained. A theoretical model was developed which can predict the net peak area counts expected on these prompt-gamma-ray energies using the thermal neutron fluence rate at the sample position, the absolute efficiency of the detector, and the mass and partial gamma-ray production cross section data for the samples. The experimental and predicted results were compared to establish the performance capability of the reconstructed facility. Good agreements between experimental and predicted results were obtained for chlorine, but results from cadmium showed larger discrepancies due to self-shielding effects. Corrections for self-shielding effects were applied to results from cadmium and the experimental and predicted results were also in good agreement. The satisfactory results indicate that it is possible to implement the prompt-gamma, neutron-activation analysis technique at Beam Port #1 of the Nuclear Science Center Reactor. To be able to obtain excellent results from other samples, improvements in shielding materials to attain a lower detection background and a highly efficient detection system should be incorporated.

Both response to intervention (RTI) and recognition and response systems recommend the use of evidence-based teaching strategies and individualized data collection to monitor the students' response to such strategies. In addition to the efficacy of interventions, individual stakeholders should have a voice in which intervention is implemented. Constant prompt delay and progressive prompt delay procedures have been routinely implemented and have been proven to be effective at teaching important skills to a variety populations; however, no objective data has been reported on the stakeholders' preference for such procedures. Thus, the purpose of the present study was to evaluate the relative efficacy of as well as the relative preference for prompt delay procedures. Four preschool children experienced three teaching conditions: constant prompt delay, progressive prompt delay, and a control condition. A multi-element design was used to evaluate the relative efficacy of the prompt delay procedures at teaching pre-academic tasks while a modified concurrent chains arrangement was used to evaluate each child's relative preference for each teaching strategy. The results for efficacy of and preference for prompt delay procedures were idiosyncratic for all participants. These results as well implication of the results are discussed.

Many research have shown that reflection is one of the important factors in the learning process. It is then recommended that teacher should adopt various prompt techniques to promote learners¡¦ reflection ability in traditional classrooms. However, there are still very few studies on how to apply this kind of teaching strategy in outdoor learning environments. Therefore, this study designed an adaptive prompts strategy by matching prompt types with learning styles to improve learners¡¦ reflection levels in an outdoor u-learning environment. We conducted an experiment using the designed and developed learning system in a butterfly ecology garden. Two classes of thirty-nine 5th grade primary school students were involved in the experiment. The result shows that the reflection levels of the learners whose learning styles matched with the appropriate prompt types were higher than that the non-matched group of learners. Therefore, it is evidenced that the designed adaptive reflection prompt system is useful in an outdoor u-learning environment.

Emissions of oxides of nitrogen (NOx) from combustion devices is a topic of tremendous current importance. The bulk of the review of NOx emissions has been in the field of turbulent jet flames. However laminar jet flames have provided much insight into the relative importance of NOx reaction pathways in non premixed combustion for various flame conditions. The existing models include detailed chemistry kinetics for various species involved in the flame. These detailed models involve very complex integration of hundreds of chemical reactions of various species and their intermediates. Hence such models are highly time consuming and also normally involve heavy computational costs. This work proposes a numerical model to compute the total production of NOx in a non-premixed isolated circular laminar jet flame. The jet consists of the fuel rich inner region and the O2 rich outer region. The model estimates both thermal NOx and prompt NOx assuming single step kinetics for NOx formation and a thin flame model. Further the amount of air entrainment by jet depends upon the Sc number of fuel. The higher the Sc number, the higher is the air entrained which lowers the flame temperature and hence NOx formation. With increasing Sc number, flame volume increases which leads to an increase in the NOx formation. The effect of the Sc number variation on the net production of NOx and flame structure is also investigated. The effect of equilibrium chemistry for CO2 <-> CO + 1/2 O2 and H2O <-> H2 +1/2 O2 on total NOx emission is studied. Also the effect of both CO2 and H2O equilibrium is considered simultaneously and the net x NO formation for propane is 45 ppm. The split between pre-flame and post-flame regions is also investigated. For Propane, 96% of NO emissions occur in the pre-flame region and about 4% in the post-flame region. The model predictions are compared with experimental values of NOx missions reported elsewhere.

The integration of large language models (LLMs) with third party applications has allowed for LLMs to retrieve information from up-to-date or specialized resources. Although this integration offers numerous advantages, it also introduces the risk of indirect prompt injection attacks. In such scenarios, an attacker embeds malicious instructions within the retrieved third party data, which when processed by the LLM, can generate harmful and untruthful outputs for an unsuspecting user. Although previous works have explored how these attacks manifest, there is no benchmarking framework to evaluate indirect prompt injection attacks and defenses at scale, limiting progress in this area. To address this gap, we introduce InjectBench, a framework that empowers the community to create and evaluate custom indirect prompt injection attack samples. Our study demonstrate that InjectBench has the capabilities to produce high quality attack samples that align with specific attack goals, and that our LLM evaluation method aligns with human judgement. Using InjectBench, we investigate the effects of different components of an attack sample on four LLM backends, and subsequently use this newly created dataset to do preliminary testing on defenses against indirect prompt injections. Experiment results suggest that while more capable models are susceptible to attacks, they are better equipped at utilizing defense strategies. To summarize, our work helps the research community to systematically evaluate features of attack samples and defenses by introducing a dataset creation and evaluation framework. / Master of Science / Large language models (LLMs), such as ChatGPT, are now able to retrieve up-to-date information from online resources like Google Flights or Wikipedia. This ultimately allows the LLM to utilize current information to generate truthful, helpful and accurate responses. Despite the numerous advantages, it also exposes a user to a new vector of attacks known as indirect prompt injections. In this attack, an attacker will write a instruction onto an online resource that an LLM will process when retrieved from the online resource. The primary aim of the attacker is to instruct the LLM to say something it is not supposed to, and thus may manifest as a blatant lie or misinformation given to the user. Prior works have studied and showcased the harmfulness of this attack, however not many works have tried to understand which LLMs are more vulnerable to indirect prompt injection attacks and how we may defend from them. We believe that this is mainly due to the non-availability of a benchmarking dataset which allows us to test LLMs and new defenses. To address this gap, we introduce InjectBench, a methodology that allows the automated creation of these benchmarking datasets, and the evaluation of LLMs and defenses. We show that InjectBench can produce a high quality dataset that we can customize to specific attack goals, and that our evaluation process is accurate and agrees with human judgement. Using the benchmarking dataset created from InjectBench, we evaluate four LLMs and investigate defenses for indirect prompt injection attacks.

Ce travail de thèse a été effectué dans le cadre de la collaboration CLaRyS, qui a pour objectif le développement d'une caméra gamma multi-collimatée et d'une caméra Compton pour les applications médicales notamment pour le contrôle en ligne de l’hadronthérapie. La caméra Compton pourrait également être utilisée en médecine nucléaire. L’objectif principal de ce travail de thèse était la caractérisation complète des détecteurs qui composent les caméras et le test des deux systèmes complets avec des faisceaux d’ions cliniques. En parallèle, des études en simulation ont permis d'estimer les performances de la caméra Compton à la fois pour le contrôle de l’hadronthérapie et la médecine nucléaire / This thesis work has been carried out within the CLaRyS French collaboration, which is involved in the development of a multi-collimated gamma camera and a Compton camera for the application in ion beam therapy monitoring through prompt-gamma detection and in nuclear medicine examinations. The main goal of the thesis was the complete characterization of the camera detector components, and the test of the whole systems on clinical ion beams. In parallel, simulation works have been performed to assess the performance of the Compton camera for the measurement of ion range during proton and carbon therapy, and for single photon emission computed tomography application in the nuclear medicine field