12466 research outputs found
Sort by
Simulation to Practice: Data-driven Characterization of Fairness in Societal Resource Allocation
Get PDFDevising allocation algorithms for limited societal resources, such as housing for homeless individuals or outreach workers to tenants facing evictions, necessitates careful consideration as the implications of algorithmic decisions can significantly affect individuals in vulnerable situations. For instance, deciding whether to allocate a homeless shelter to a household or to remove a household from a shelter can have long-lasting effects on all household members, including children. However, ensuring the fairness of these allocation algorithms is particularly challenging. Direct application of off-the-shelf fairness notions is often notfeasible, as they are standardized for typical allocation and classification settings without accounting for underlying context and policy constraints. Moreover, additional factors such as the dynamic nature of resources and heterogeneity in responses from the individuals receiving the resources, make it more difficult to apply these standard notions of fairness in critical societal domains. In this dissertation, we investigate approaches to defining, characterizing, and auditing fairness in high-stakes settings including eviction prevention, homeless services, and policing. In these different settings, we define appropriate context-aware notions of fairness, validate them through extensive simulated experiments, and analyze their practical implications using novel real-world datasets
COARSE-GRAINED FORCE FIELD DEVELOPMENT FOR OXIDIZED POLYPYRROLES AND MONTE CARLO SIMULATIONS OF THEIR PROPERTIES
No full textThis research puts forward a novel coarse-grained force field (CGFF) for representing the atomistic interactions that occur in the condensed phases of oxidized polypyrrole (PPy) systems. Polypyrrole is a conjugated polymer that conducts electricity when doped with atoms or small molecules that oxidize the polymer matrix. Analytic descriptions for modeling this technically important conducting polymer have been elusive. Hence, the goal of this study is producing a robust classical force field that enables the possibility of predicting properties at the atomic level that either have not yet been measured empirically or are difficult to attain in the laboratory. The newly developed force field allows us to perform an extensive study based on high performance computing Markov chain Monte Carlo simulations, which are well established for yielding realistic outcomes of complex systems such as the one at hand. We adopt the Metropolis Monte Carlo (MMC) characteristics of the Markov chain in the simulations carried out for examining both, thermodynamic properties such as enthalpy, density, specific heat capacity, thermal expansion coefficient, bulk modulus, and structural properties including radial distribution functions, orientation order parameter, vector order parameter, oligomer radius of gyration, and more. Force fields are a collection of mathematical expressions that include parameters. Our CGFF has twenty two parameters. Obtaining these parameters led to a custom created workflow that entails optimizing the CGFF parameters based on a collection of target potential energy points calculated at the quantum level with density functional theory for a library of PPy structures. The process entails a non-linear least square fit as first stage of the parameter optimization. The second stage is a reinforcement learning approach, in which MMC simulations of a small PPy system are carried out using the CGFF with the first set of determined parameters to obtain the system density at ambient conditions, compare it with the experimental result, and make decisions to keep or improve the parameter values. This stage is repeated multiple times until achieving the CGFF most optimal parameters. Atomistic simulations as those in this research require verification and validation of the obtained properties by testing the sensitivity of the PPy properties to the scaling-up of system size. In that pursue, several PPy system sizes are addressed containing 1,024 to 27,618 particles, which permits to demonstrate that the larger system has stable properties that are comparable to experiments within statistical significance. We assert that the obtained solid samples of PPy display strong polymer chain stacking and a preferential ordering of the oligomers along one direction. These samples are not crystalline, in agreement with the amorphous synthesized samples of PPy with chlorine dopants. The CGFF is generalizable to PPy with different dopants by only a re-parametrization of the analytical terms corresponding to polymer-dopant. The CGFF analytic terms are generalizable to polymer systems that give rise to bipolarons in which regions of consecutive monomers within each oligomer sustain a charge of +2e countered by anionic dopants. For large system sizes, the coarse-grained approach of our CGFF implies savings in computational cost by a factor of 3-to-4 when compared to the expenditures required by all-atom force fields. In a nutshell, the CGFF developed along our research is reliable and useful for the calculation of PPy properties including its behavior in composite materials
Using Neural Networks to Solve the Differential Equations of Magnetohydrodynamics
Get PDFThe equations of magnetohydrodynamics (MHD) are a set of coupled nonlinear partial differential equations (PDEs) that describe the behavior of plasmas under certain idealized conditions. The MHD equations combine the Navier-Stokes equations with Maxwell's equations. These equations are typically solved using well-developed numerical methods, such as the finite-element method (FEM) or the finite-difference method (FDM). New artificial neural network (ANN) methods for solving differential equations were developed recently. The ANN methods are versatile, robust, and relatively straightforward to configure and execute. This work investigates two ANN approaches for solving MHD problems: the trial function method and the physics-informed neural network (PINN) method. The trial function method utilizes an approximate functional form for the solution, which uses the network output and incorporates the boundary conditions in a way that is analogous to the method of Lagrange multipliers used in constrained optimization. In the PINN technique, the formulation of the loss function incorporates all knowledge of the physics of the system, including the differential equations describing the relevant physical laws, any constraints and initial or boundary conditions, and any measured data. We developed two novel software suites that implement the trial function and PINN methods and applied them to a collection of MHD problems. Both sets of software have been published on GitHub. While developing this software, we developed a novel algorithm for formulating the boundary condition contribution to the trial function. We applied the trial function software to a set of problems of increasing complexity to assess its suitability for use with MHD problems. The results of these experiments indicated that the trial function method was unsuited for problems with multiple spatial dimensions or complex boundary conditions. We then applied the PINN software to a complementary set of problems, including standard problems from the MHD solver literature. We found that the PINN method was easier to configure and implement than the trial function method. We also found that the PINN method provides results comparable in accuracy to a current state-of-the-art MHD solver (GAMERA) for the problems examined in this work. The PINN method requires more time to train than current numerical solution methods, but the PINN method also requires significantly less storage for the trained solutions
Cross-Cultural Interaction in the Global Economic
Get PDFDoing business internationally is a global thread in the 20th century; the GDP has a 40 percent difference if you connect to a global goods and services network (Manyika, 2024) or not. However, it also means it will have two or more cultures colliding together.
Trompenaars (2021) points out the main dilemma when it comes to cross-cultural business: centralization versus decentralization. Decentralization means each culture goes its own way with the idea of more local better. Centralization means principles need to be the same in different countries. Decentralization can be done in individualistic cultures more like western culture, but not as a family model in eastern countries (Trompenaars, 2021).
Culture differences in the business era often exist in the way people view friendship and contracts, the relationship between business partner and government. Cross-culture management, which provides more culture awareness and flatter organization, is the way to reconcile the culture problem nowadays in international companies like Tesla
Structural Neural Correlates of Social Networks
Get PDFSocial networks play a crucial role in human social interaction, and understanding the neural foundations of their cultivation and maintenance is essential. The social brain hypothesis posits that the neocortex, and subsequently, the social brain, adapted and evolved to meet the demands of maintaining connections. Furthermore, cognitive capacity influences how individuals are maintained in support and sympathy groups, with support groups representing the innermost and emotionally close layer, while sympathy groups expand this circle to include individuals who are less emotionally connected. The difference between these two groups suggests that maintaining sympathy groups involves more mentalizing, whereas support groups rely more on emotional processing. This dissertation investigates the structural neural correlates of social networks by examining both gray and white matter (WM) associations with the Social Network Index (SNI), which measures social network diversity, size, and complexity by assessing 12 different types of relationships, along with mediating factors such as empathy, in a series of two studies. The SNI was chosen as it is inferred to be more suitable for addressing the sympathy group. The first study, highlighted in Chapter Two, investigated the gray matter volume (GMV) associations of social brain regions and the SNI, along with empathy, using voxel-based morphometry, linear regression analyses, and mediation analyses. Our findings revealed an inverse association between the GMV of the dorsomedial prefrontal cortex (dmPFC) and social network size. Cognitive empathy mediated this relationship, which suggested that the smaller the dmPFC’s GMV, the more cognitive empathy was employed, leading to an increased social network size. This association may be explained by cognitive pruning, a mechanism that occurs during adolescence. The second study, highlighted in Chapter Three, explored the association between WM integrity and social network dimensions, focusing on the WM tracts relevant to social cognition (cingulum [CING], superior longitudinal fasciculus [SLF], uncinate fasciculus, inferior fronto-occipital fasciculus). Using diffusion tensor imaging, linear regression models, and mediation analyses, we identified that greater WM integrity within the left SLF was linked to larger social network sizes. Furthermore, mediation analyses showed a distinctive association between the WM integrity of the cingulum and cognitive empathy, which indicated that individuals with greater WM integrity in the CING exhibited less cognitive empathy, which contributed to larger social network sizes. By addressing both structural and behavioral dimensions, this dissertation provided a comprehensive exploration of the neural correlates of social networks, particularly in the context of sympathy groups. These findings shed light on other cognitive processes such as empathy, which may also be involved in the formation and maintenance of social networks. As technology and social media continue to reshape social interactions, the implications of these findings offer more insights into the social cognitive mechanisms involved in mentalizing and empathy regarding social networks
Making Birth Stories Matter(s): Working Toward Reproductive Health Justice - An Examination of Health, Communication, Culture, and Identity Through Blending Critical Autoethnography and Narratives of Women Birthing with Doulas
No full textThis work is embargoed by the author and will not be publicly available until May 2026.Statistics from the CDC show the rate for maternal mortality in the United States continued to rise in 2023 with Black women reported three to four times more likely to die as a result of pregnancy-related complications than non-Hispanic White women (Eissen et al., 2019; Hoyert, 2023). COVID-19 further complicated the maternal health crisis (Hoyert, 2023). The maternal and infant mortality statistics continue to be egregious, yet conversations contextualizing these statics are silenced, rendering people and their stories “disposable”. Uncovering harmful dominant attitudes relating to reproductive health is important for transforming culture and wellbeing for marginalized women. As a community birth doula, advocate, researcher, and doctoral student, I blend multiple intersectional and culture-centered methodologies (Esposito & Evans-Winters, 2022) from critical and qualitative (Corbin & Strauss, 2014) paradigms to conduct a cultural analysis of birthing for women birthing in the United States. Birth doulas are described as trained labor-support professionals who provide continuous emotional, physical, and informational support for families during pregnancy, childbirth, and postpartum periods (BADT, 2021). Birth doulas have increased healthy birth outcomes, including reducing unnecessary medical interventions and decreasing the time spent in the most intense phases of labor (Gruber, Cupito & Dobson, 2013). I present tensions from an analysis of individual semi-structured interviews (Corbin & Strauss, 2014) with women who gave birth with a doula present, along with autoethnographic accounts of my personal experiences related to reproductive justice and birth work. This project uses the tenets of the Culture Centered Approach and Reproductive Justice frameworks to interrogate the intersections of disposability and social justice for birthing women in the United States. These stories are important for transforming narratives around birth and improving health outcomes for birthing women and future generations.2026-05-1
Trust in Higher Education: Faculty Perspectives on the Facets of Trust in Business Writing-Intensive Classrooms
Get PDFTrust between faculty and students in higher education is not as robust as it could be, which compromises both the faculty and student experience. In lieu of recent and ongoing events (i.e., the global COVID-19 pandemic causing a significant shift to remote learning and virtual communication; the racial disparities recognized across the US; and the national debates surrounding the purpose and scope of higher education), the higher education system has been challenged, stretched, and motivated to assess and to adapt, with trust being one area under review. The National Survey of Student Engagement (NSSE) 2020 Annual Report, which was administered at 29 institutions across the US, included a special set of questions related to the notion of trust and found that only 30% of students completely trusted their instructors. While faculty were considered the most trusted among students across the varying dimensions measured, the survey found that 70% of students only trusted faculty somewhat (59%), very little (9%), or even not at all (2%). With 11% of students either barely trusting or not trusting faculty at all, and considering “trust is an important element in human learning because much of what is learned is based on the verbal and written statements of others that the learner is asked to believe [often] without independent evidence,” there is room for improvement (Tschannen-Moran and Hoy, 2000, p. 547). And since research also identifies that the responsibility to build and sustain trusting relationships is the responsibility of the person with greater power (Tschannen-Moran, 2014), this dissertation research aims to illuminate the most critical space for trust to be initiated: faculty perspectives. More specifically, this qualitative case study explores four business instructional faculty members’ notion of trust in relation to their students, writing-intensive classrooms, and pedagogy. Based on participant interviews, teaching statements, and course artifacts, this dissertation examined how the five facets of trust (i.e., benevolence, honesty, openness, reliability, and competency) influenced these faculty members’ perspectives, and examined the relationship trust and writing played in their writing-intensive classrooms. Some of the findings from this study discuss which facets of trust emerged most amongst the faculty participants; how each facet showed up in relation to each faculty member, with varying rationale; as well as the importance of writing and how its understanding varied depending on the faculty members’ experience and background. More broadly, the findings conclude that the notion of teacher-student trust is not prominently known nor considered a pedagogical necessity for these business faculty in higher education teaching these writing-intensive courses, and that intentional disclosure and pedagogical discussions on the five facets of trust and writing to increase thoughtful and measured approaches to the teacher-student trust relationship should be required of all faculty
Networks and Policy Outcomes: The Case of Refugee Integration
Get PDFRefugee immigration poses a global management challenge for international, national, subnational, and local government entities. Facilitating the mass resettlement of vulnerable populations requires the coordination of multiple government and non-government actors. Yet, we know surprisingly little about the conditions that produce successful refugee integration in society. This study draws on our understanding of social networks to identify conditions that encourage successful economic, social, and cultural integration. Specifically, I ask: how do refugee resettlement networks affect refugee integration outcomes? The theory of successful integration is tested on resettled Afghans and sponsor groups who participated in the Sponsor Circle Program, analyzing data from surveys sent to sponsor group members and resettled Afghans. The study’s findings contribute to existing literature by offering insights into the network structures of sponsor groups, revealing them to be small and sparsely connected with moderately levels of centrality and cohesion but with overall strong communication ties. I find evidence indicating that these network structures influence short-term integration outcomes for refugees. Specifically, I observe refugees’ economic integration is greater within groups with loose connections and minimal internal cliques. And cultural integration outcomes are greater for refugees whose sponsor gro’p members engage in less frequent communication with one another. This study serves as a foundational study for future research on how refugee resettlement networks impact integration outcomes
Uranium Isotope Response to Changing Oxygen Minimum Zone Dynamics in the Arabian Sea over the past 60 kyr
Get PDFQuantitative understanding of the δ238U paleo-redox proxy requires constraints on δ238U fractionation in a range of modern environments with different redox conditions. In particular, euxinic environments are known to exert a substantial lever on the global seawater δ238U mass balance through preferential removal of 238U. However, the degree of δ238U fractionation under low-O2, non-euxinic conditions is poorly understood. Here, we present δ238U data for two cores from the Arabian Sea OMZ that span the past ~65 kyr—one core each from the Oman and Pakistan margins. Today, the Arabian Sea is characterized by suboxic, denitrifying, non-euxinic conditions, but the intensity of denitrification has varied over the past ~65 kyr in association with global climate cyclicity. In order to assess the U isotope fractionation, Δ238U values were calculated as the difference between sediment and modern seawater δ238U. We find that the median Δ238U value for both cores over the past ~65 kyr is 0.15‰, indicating only a modest degree of fractionation compared to euxinic settings, like the Black Sea, where the degree of fractionation ranges from 0.60 to 0.80‰. δ238U varies cyclically in both cores, however, while δ238U maxima of the two cores correspond to interstadials, the two cores record δ238U maxima during different interstadials. The maximum Δ238U value of 0.39‰ was recorded during Dansgaard-Oeschger Event 13 on the Oman Margin at 47.3 kyr. Trace metal proxies indicate that higher δ238U values in the dataset may be associated with environments in which sulfide was present in pore waters, which we hypothesize resulted in greater fractionation than observed in other suboxic, non-euxinic environments around the globe. We use a δ238U mass balance model to explore the effect of “Arabian Sea-style” conditions on global seawater δ238U. We find that when a Δ238U value of 0.15‰ is used for the “Arabian Sea-style” sink, even widespread expansion of this type of environment does not exert a substantial lever on the global δ238U mass balance. When the maximum Δ238U value (0.39‰) is used, however, expansion of “Arabian Sea-style” suboxia is capable of producing a seawater δ238U value as low as −0.69‰, which matches the carbonate record of some anoxic events of the geologic past. This suggests that widespread ocean euxinia may not be required to explain some negative δ238U excursions in the ancient carbonate record
Synthesis and Characterization on Noncentrosymmetric Oxides and Intermetallics
Get PDFInorganic solid-state compounds that lack inversion symmetry in the crystal structures are defined as non-centrosymmetric (NCS) materials. These compounds have applications in spintronics, energy conversion, and quantum technology. Understanding the interplay of the NCS crystal structure, electronic structure, and physical properties is essential for discovering novel NCS materials, which are of interest to a wide range of scientific fields, from inorganic chemistry and material science to condensed matter physics. Currently, it is still challenging to discover novel NCS materials due to limited design strategies. Increasing the knowledge of these materials allows for a more comprehensive understanding of the materials, which will allow us to improve our ability to create novel compounds in this area. It is also important to develop novel design strategies for creating novel NCS materials, especially magnetic NCS materials. To achieve this overarching goal, this work primarily investigates NCS oxides as potential magnetic NCS materials and designs novel NCS oxides based on centrosymmetric (CS) parent compounds. Moreover, we study NCS intermetallics as potential polar thermoelectric materials and semimetals. Chapter 2 examines the crystal structure and magnetic properties of CS Li8IrO6 and whether there is a CS to NCS structural transition at high temperatures. Li8IrO6 can be successfully prepared through a simple solid-state reaction. Rietveld refinement of this compound at room temperature using powder neutron diffraction data confirms the CS space group R3 ̅. An in-situ synchrotron X-ray diffraction study of Li8IrO6 indicates that the sample is stable up to 1000 °C and exhibits no structural transitions. The oxidation state of Ir ions in the crystal structure is confirmed to be +4. Magnetic measurements suggest long-range antiferromagnetic ordering with a Néel temperature (TN) of 4 K, which is further corroborated by the heat capacity measurement. The localized effective moment µeff (Ir) = 1.73 µB and the insulating character indicates that Li8IrO6 is a correlated insulator, which is supported by first-principle calculations. Due to the unsuccessful conversion of the CS Li8IrO6 to NCS at high temperatures, Chapter 3 focuses on testing a design strategy that can prepare novel NCS magnetic oxides by introducing disorder in the CS structure containing face-sharing octahedra with more than two transition metal sites. The disorder will be introduced on the metal sites by doping CS Ba4T3O10 (T = Mn, Ru) compounds. Chapter 3 reports the synthesis of a series of Ba4Ru3-xMnxO10 (x = 1, 1.5, 2) polycrystalline NCS materials, which are characterized by various techniques, including electron microscopy, convergent-beam electron diffraction, second harmonic generation, and temperature-dependent neutron diffraction. All results support the formation of the NCS polar crystal structure (Cmc21) for Ba4Ru3-xMnxO10 (x = 1, 1.5, 2) compounds. The magnetic measurement indicates that Ba4Ru3-xMnxO10 (x = 1, 1.5, 2) compounds exhibit antiferromagnetic ordering at low temperatures with strong magnetic frustration. Chapter 4 examines the successful synthesis and thermoelectric properties of Ag2GeS3. Ag2GeS3 adopts an NCS polar orthorhombic crystal structure (Cmc21). UV-vis diffuse reflectance spectra indicate the compound is a semiconductor with an optical indirect band gap. Thermoelectric conductivity measurements show that Ag2GeS3 exhibits glass-like ultralow lattice thermal conductivity. Theoretical calculations of the lattice thermal conductivity based on density functional theory show a good qualitative agreement with experimental results. Investigation of NCS intermetallic compounds was extended to IrGe4, as described in Chapter 5. The convergent-beam electron diffraction and Rietveld refinement using synchrotron powder X-ray diffraction data suggest that IrGe4 adopts a chiral NCS crystal structure (P3121) instead of the polar crystal structure (P31). Temperature-dependent resistivity indicates the metallic behavior, and the Hall resistivity suggests that holes are the majority carrier type. Electronic band structures calculated by the density functional theory reveal a Weyl point above the Fermi level