We develop a technique for extracting parallelism from ordinary (sequential) programs. The technique combines two fine-grain, instruction level code transformations to achieve effects similar to the wavefront method. Such effects were previously available only as transformations at coarser levels of granularity. By integrating the strengths of the wavefront method with the ability to extract fine-grain, irregular parallelism at the instruction level, our technique exploits previously untapped parallelism.

In California, up to 46% of the state's total water supply is from groundwater and its geogenic contamination may affect a large area, number of wells, and number of people throughout the state. The presence of geogenic contaminants, such as chromium (Cr) and manganese (Mn), in groundwater is due to the interplay between complex processes (i.e., biogeochemical reaction networks) involving mineral dissolution, diffusion, redox transformation, sorption, and precipitation. To effectively evaluate the safety of our drinking water it is critical that we understand the network of biogeochemical reactions that control geogenic metals cycling. The overarching goal of this dissertation work is to explore how fluctuating redox conditions from micron- to field-scale may influence the release of geogenic metals and threaten drinking water sources. To first address this goal on the micro-scale, we investigated how the co-occurrence of iron (Fe) and Cr minerals in reduced zones of soil aggregates may alter the redox potential of Mn oxides within oxidized zones at diffusion-controlled soil interfaces. Altering the oxidation state and structure of Mn oxides can impact the oxidation of Cr(III), a non-toxic mineral, to Cr(VI), a mobile and highly toxic groundwater contaminant. Simultaneously, the oxidation of Cr minerals by Mn oxides will also release soluble Mn, a secondary contaminant, through reductive dissolution. On a state-wide scale, we then investigated the groundwater contamination by redox controlled groundwater contaminants, such as Mn, within public water systems relying on groundwater resources. Larger systems that treated for primary contaminants often saw associated benefits of less Mn, however, this was not observed within small systems or systems without treated groundwater. Further, we investigated differences in Mn contaminant exposure between private, shallow well users and deeper, public well users in California’s Central Valley. We found that, despite predicted differences in redox status groundwater based on well-depth, this led to no differences in Mn concentration in groundwater. However, lack of monitoring or treatment of private wells and the high associated costs may lead to exposure to concentrations of Mn exceeding regulatory standards.

Many program analyses can be reduced to graph reachability problems involving a limited form of context-free language reachability called Dyck-CFL reachability. We show a new reduction from Dyck-CFL reachability to set constraints that can be used in practice to solve these problems. Our reduction is much simpler than the general reduction from context-free language reachability to set constraints. We have implemented our reduction on top of a set constraints toolkit and tested its performance on a substantial polymorphic flow analysis application.

Social protection programs are essential to assisting the poor, but governments and humanitarian agencies are rarely resourced to provide aid to all those in need, so accurate targeting of benefits is critical. In developed economies, targeting decisions typically rely on administrative income data or broad survey-based social registries. In low-income countries, however, poverty information is rarely reliable, comprehensive, or up-to-date. Novel sources of digital data — from mobile phones and satellites, in particular — are well suited to fill this gap: they are predictive of wealth in low-income contexts and ubiquitously collected. The research studies in this dissertation design and evaluate new methods for targeting aid in low-resource contexts using machine learning, satellite imagery, and mobile phone data, and evaluate these methods in large, real-world interventions. Across social protection programs in Togo, Afghanistan, and Bangladesh, the studies in this dissertation show that targeting methods based on machine learning and digital data sources identify poor households more accurately than methods based on categorical eligibility criteria like geography or occupation, but typically less accurately than traditional survey-based poverty measurement approaches. These results highlight the potential for digital data and machine learning to improve the targeting of humanitarian aid, particularly when traditional poverty data are unavailable or out-of-date and in settings where conflict, environmental conditions, or health concerns render primary data collection infeasible. These studies also provide empirical evidence on the limitations and risks of digital and algorithmic targeting approaches, including privacy, transparency, fairness, and digital exclusion.

Two hypotheses on the origin of the ‘Isabella tomographic anomaly,’ which has been interpreted as either a lithospheric delamination (Ducea and Saleeby, 1998) or a remnant of the Farallon plate attached to the Monterey micro-plate (Wang　et al., 2013), were tested. P-wave receiver functions and tomography based on teleseismic events recorded by 37 permanent stations and 4 newly installed stations were used to construct a simple geometry tomographic model of the Isabella anomaly, and to test whether the Monterey micro-plate can be connected with the location of the Isabella anomaly. A rectangular block model was used to fit the arrival-time residual pattern, and the best-fit model has its top surface located at 50km below the Great Valley. The block model dips 65º toward the Sierra Nevada with 100 km thickness. Receiver functions suggest that the remnant Farallon plate is not stalled at the base of the crust beneath the Coast Ranges as has been suggested from refraction studies (Brocher et al., 1999). Instead, the Monterey micro-plate lies in the mantle. Its top and bottom surfaces can be observed as a negative-positive dipole signal in the receiver functions. Negative-positive dipole signals at 35 km observed in the Coast Ranges' receiver functions were interpreted as the P-to-S converted phases from the Monterey micro-plate. These P-to-S converted phases from the top of the oceanic crust could be traced from the Coast Ranges to the Great Valley. The depth of the top surface of the Monterey micro-plate suggested from the receiver functions is consistent with the location of the rectangular block found from tomography, supporting the hypothesis that the Isabella anomaly is a remnant of the Farallon plate that is attached to the Monterey micro-plate. The combined results of the receiver functions and tomography study suggest that the partially subducted Monterey micro-plate extends east beyond the San Andreas fault and that the initially low angle of subduction in the west increases its dip east below the Great Valley.