Many computations in biomedical research such as simulations, bootstrapping, database searches (such as BLAST), and many Monte Carlo algorithms are embarrassingly parallel. This means that the computation can be split up into smaller computations; each of those calculations can be performed in parallel threads that do not need to interact with each other. Computations with this feature can be easily distributed,(that is, run on different computer processors), with a gain in speed that is approximately proportional to the number of processors. In this note we introduce some of the concepts behind distributed computing, examples where they have been used, and lay out scenarios where they may be useful for biomedical researchers in the future.

This is a short note on the mating scheme used by the Mutagenesis Project. We present a probabilistic analysis of the distribution of the number of mutants in the G3 generation. It is shown to be a function of the number of G2 mothers and litter sizes. A computer program is provided to make the calculation. We quantify the odds of a G2 mother being a mutation carrier given that none of its progeny are mutants. Finally we analyze some data from the project; we find the data to be consistent with theory.

Physically based rendering is widely used due to its ability to create compelling, photorealistic images. Generating these images requires evaluating a multidimensional integral and is typically done through Monte Carlo (MC) samples of the scene's light transport. However, these samples are computationally costly and many are needed to obtain a converged result, otherwise there is objectionable noise in the final image.

To address these concerns, there has been substantial work aimed at reducing the variance of the MC estimate. For this dissertation, we observe that this work can be categorized based on where in the rendering pipeline these methods are being applied: pre-rendering, intra-rendering, and post-rendering. The first category refers to methods that perform preprocessing to make subsequent rendering more efficient. Meanwhile, intra-rendering methods will provide feedback during rendering to facilitate convergence of the MC estimate. Finally, post-rendering approaches are used after the sampling budget has been exhausted to output images that better match the ground truth rendering.

Previous methods in these categories are far from optimal as a result of their inaccurate approximations, costly initialization, and fallible heuristics. On the other hand, machine learning approaches in countless applications have seen steady progress in recent years thanks to their ability to model complex relationships with significantly more accuracy and robustness relative to their hand-crafted counterparts. Inspired by these approaches, we propose to apply machine learning to all three portions of the pipeline to avoid the aforementioned pitfalls of prior work in MC variance reduction.

For pre-rendering, we introduce the first deep network for prefiltering that can accurately capture the appearance of large environments with complex geometry and materials for level of detail. Our approach foregoes traditional geometric simplification and volumetric modeling and robustly handles the range of cases.

Within the context of intra-rendering, previous approaches for importance sampling are costly since they require many samples and, in some cases, even additional training per scene before they can be effectively used to guide samples towards paths with higher light contributions. We demonstrate the first offline deep network that can guide sampling for almost the entirety of rendering for an arbitrary scene.

Finally, we bring deep learning to post-rendering MC denoising and replace the heuristics used by previous methods to set the parameters of explicit filters. Our network predicts the filter kernel around each pixel, can even directly filter the image, and works on general scenes and varying levels of noise without retraining.

In this thesis, we demonstrate how our proposed, learning-based solutions offer a significant advantage over prior state-of-the-art methods in each of these cases. We also discuss avenues for future work in the research community to bring machine learning to additional applications within all components of the MC rendering pipeline.

Chalcogenide glasses are an important class of materials that consist of sulfides, selenides or tellurides of group IV and/or V elements, namely Ge, As, P and Si with minor concentrations of other elements such as Ga, Sb, In. Because of their infrared transparency that can be tuned by changing chemistry and can be actively altered by exposure to band gap irradiation, chalcogenide glasses find use in passive and active optical devices. Chalcogenide glass infrared lenses have found applications in the auto industry and in night vision security systems for both civil and military use. Infrared transmission in several key wavelength windows has enabled their use as optical fibers, lasers and fiber-amplifiers for telecommunication and as infrared laser power delivery systems for medical applications. Moreover, their large optical nonlinearity and high refractive index make them attractive candidate materials for all-optical switches in optical circuits. Finally, in thin film form, chalcogenide phase-change materials are of extraordinary technological importance in rewritable non-volatile memory storage applications (e.g., CD, DVD and Blu-Ray). A complete knowledge of the atomic structure of any material is of key importance in understanding and formulating accurate predictive models for its properties. Besides structure, the dynamical processes associated with structural relaxation and annealing near the glass transition controls the technological utility of inorganic glasses. In this dissertation Raman and one- and two- dimensional multi-nuclear (31P, 77Se, 125Te) nuclear magnetic resonance (NMR) spectroscopic techniques are utilized to elucidate the compositional evolution of the atomic structure at the short- and intermediate- range and chemical order in a series of binary selenide glass-forming systems, namely: SxSe100-x, Se100-xIx, TexSe100-x, and PxSe100-x. These spectroscopic results yield a rather detailed picture of the compositional evolution of the atomic structure of these glasses, where the constituent structural moieties range from 0-dimensional molecules to 1-dimensional linear chains all the way to a 3-dimensional network. The corresponding physical properties such as the glass transition temperature and the molar volume are shown to be intimately linked to and consistent with the compositional evolution of the atomic structure in each system. On the other hand, the shear-mechanical response of all compositions in the supercooled liquid state are investigated via a combination of small-amplitude oscillatory and steady shear parallel plate rheometry to build an atomic-scale understanding of the frequency dependent viscoelasticity and structural relaxation processes. These results reveal the mechanistic connection between the structure including its connectivity and dimensionality and the shear relaxation processes, such as the rapid segmental chain motion associated with polymeric chains, the slow bond scission/renewal dynamics caused by interconversions between the structural moieties and the strong secondary bonding interactions, as well as the fast rotational motion and cooperative dynamics resulted from molecules. Depending on the structural moieties presented in a liquid, one or more shear relaxation processes may exist and contribute to the viscous flow. In addition, the shear-mechanical response evolves systematically with the structural transformation of these liquids as a function of composition within each system. When taken together, these results enable us to establish fundamental connections between the “microscopic” and the “macroscopic” aspects of viscoelasticity and relaxational phenomena in chalcogenide glass-forming liquids and to develop robust predictive atomistic models of structure-dynamics-relaxation for all studied systems.  

1857 remains a central watershed in the North Indian history, marking an end to the Mughal rule and permanently altering the material and social landscape of North India. This research analyzes the accounts of 1857 and the late Mughal rule by North Indian Urdu authors written in the early 20th century. These texts revisit and reimagine the pre-1857 Mughal rule in explicit contestation with the British hegemonic accounts dominating the narratives around 1857. The authors use highly politically coded pre-modern Indo-Persian literary forms to challenge British narratives and stereotypes associated with the Mughal rule. The metaphor and ornate prose of pre-modern forms emerge as important modes of political critique that employ the subversive charge of emotions, an impulse much censured by both British and Indian reformists. The texts under study perform a vindication of the last Mughal king and retrospectively venerate him as a Sufi-king and patron of intercommunal peace, called the ‘ganga-yamuni tehzeeb’ in the texts. Rich myth-making around his figure takes place whilst employing Karbala allegory, historically used for its revolutionary symbolism- affirming divine support for the late Mughal king and reversing the narrative of the British material victory. The marsiya genre lends itself to the lament and nostalgia for the lost Mughal rule. Spatial memorialization of places associated with the Mughal rule resonates with the Indo-Persian shahr ashob genre, implying the upending of a just rule. This memorialization commemorates the Mughal discourse of sovereignty and its place in the North Indian social imagination, refuting British attempts at its mimicking through colonial re-enactments. This ‘counter-memory’ constitutes a rich yet neglected archive of re-visiting 1857, important in its exploration of intergenerational lore, social memory, and contestations with institutionalized history.