Dartmouth Institute for Health Policy and Clinical Practice
Dartmouth Digital Commons (Dartmouth College)Not a member yet
8214 research outputs found
Sort by
ENTANGLEMENT-ASSISTED METROLOGY UNDER SPATIOTEMPORALLY CORRELATED QUANTUM NOISE
Get PDFQuantum sensors operating at the microscale are an emerging branch of quantum technologies where tangible experimental successes have already been reported. State-of-the-art atomic interferometers allow to measure and estimate a variety of physical parameters with unprecedented precision. In principle, exploiting the full power of quantum mechanics would lead to quantitatively better performance bounds over the best possible classical strategies under the same given set of resource constraints. However, the quantum systems\u27 fragility to external disturbances has so far hindered most of these gains to be reached in practice, particularly in the limit of large probe number . Parallel, {\em purely dephasing} noise, is a ubiquitous source of decoherence in the relevant experimental platforms, that proves especially difficult to mitigate. In this Thesis, we analyze the impact of pure dephasing to \emph{asymptotic} () accuracy and precision of quantum frequency estimation protocols.
First, we show how dephasing-induced loss of quantum coherence may render the standard strategy used to estimate the target frequency in a regime of high prior knowledge biased or even ill-defined. We then provide a solution to this problem by modifying the interferometric sequence to construct an estimator which is asymptotically unbiased (accurate), without detriment to precision, at the cost of doubling the resource overhead.
Second, when the noise exhibits \emph{partial} spatial correlations, we prove that superclassical precision scaling can be restored by a clever spatial arrangement of the probes. In the absence of noise characterization, an intermediate limit between the optimal noiseless classical and quantum precision bounds is accomplished by randomizing the position of the sensors. When access to noise spectroscopy is available, further gains may be reached by placing the sensors in a lattice with tunable unit length. Both of these strategies fail, however, in the presence of full spatial correlations (collective noise), as the probes effectively occupy a single position in space.
Third, we establish rigorous, state-independent bounds to precision in the collective dephasing setting. These bounds can be saturated up to a constant factor by using a class of spin-squeezed input states and global measurements that are readily available in current platforms, regardless of the noise correlations. Importantly, while gains over the noiseless classical limit are possible for a dephasing process with a long-tailed spectrum, no quantum advantage may be reached for Markovian dephasing, or when the environment has a rapidly decaying spectrum.
Fourth, we explore if the above limits to performance may be surpassed by including additional resources. When the possibility of encoding the signal non-linearly is available, noiseless precision bounds generalize to accommodate for entanglement buildup throughout the evolution period. We then show that, in the presence of collective dephasing, an improvement on absolute precision with respect to the usual, linear, setting can be achieved. This precision bound may be saturated (up to a constant factor) by protocols using noise-robust states and collective measurements. For white noise and colored noise with a rapidly decaying spectrum, no gains over the non-linear noiseless classical limit is possible, however, hinting at a broader no-go. In both of these cases, we turn back to the linear setting considered before, now augmented by the ability to apply instantaneous, \emph{pulsed} open-loop control throughout the evolution. We further prove that this type of dynamical decoupling sequences are unable to lift the bound on superclassical scaling of precision, provided we additionally assume that the noise source is classical in nature. Extensions of the no-go to a quantum environment and long-tailed spectra have not been yet concluded but seem highly plausible.
Finally, we explore the benefits of \emph{continuous driving} as a resource to enhance interferometric performance in the presence of collective dephasing. We consider a large limit where the dynamics can be approximately described by a time-dependent quadratic bosonic Hamiltonian, and compute the resulting evolution. Formal expressions for the ultimate precision bounds and optimal measurement scheme are derived. While the extent of the resulting advantage for general time-dependent driving fields remains to be fully understood, we explicitly show how the use of parametric squeezing allows us to improve asymptotic scaling of precision with respect to the control less scenario. It is our hope that our results pave the way towards resource-efficient, scalable entanglement-assisted quantum metrology in realistic settings
Incorporating Visual Information into Natural Language Processing
Get PDFNatural language describes entities in the world, some real and some abstract. It is also common practice to complement human learning of natural language with visual cues. This is evident in the heavily graphical nature of children’s literature which underscores the importance of visual cues in language acquisition. Similarly, the notion of “visual learners” is well recognized, reflecting the understanding that visual signals such as illustrations, gestures, and depictions effectively supplement language. In machine learning, two primary paradigms have emerged for training systems involving natural language. The first paradigm encompasses setups where pre-training and downstream tasks are exclusively in natural language. The second paradigm comprises models that require joint reasoning over both language and visual inputs during pre-training and downstream tasks. Given the widely acknowledged role of visual input in human language comprehension, it is pertinent to inquire whether visual information can similarly augment the comprehension of language-only tasks in machine learning. Despite the remarkable advancements in the capabilities of machine learning models across all domains in recent years, the concept of supplementing Natural Language Processing with visual signals remains insufficiently explored. This is in part due to the absence of clear and effective strategies for integrating visual information into language models, given the limited availability of extensive, high-quality image-language paired datasets. In this thesis, we address this challenge and propose two frameworks for incorporating visual information into natural language pre-training leveraging multimodal models as intermediaries between visual information and language models. Empirical evaluations conducted on language pre-training datasets of varying sizes demonstrate the efficacy of the proposed frameworks across diverse downstream language tasks. In addition, we introduce methods for training effective multimodal models through architectural innovations and novel multimodal data augmentation techniques. The representations generated by our multimodal models lead to improved performance in zero-shot image categorization, visual question answering, visual entailment, and cross-modal retrieval tasks in downstream evaluations. Finally, this thesis presents a novel method for constructing effective neural networks by selection from randomly initialized parameters in contrast to the conventional practice of parameter updates via gradient descent
Eulerian Smoke Simulation with Multiple Fields
Get PDFFluid simulation is a cornerstone of computer graphics, enabling the realistic depiction of dynamic phenomena such as smoke, fire, and other gaseous behaviours. This thesis focuses on advancing Eulerian smoke simulation techniques, with a particular emphasis on grid-based simulations that capture intricate vortical structures and fine visual details.
We propose several detail-preserving frameworks that incorporate various scalar and vector fields within the simulation pipeline, including velocity, impulse, and Lamb vectors, along with their decompositions and transformed representations. By mathematically analyzing the properties of impulse, we derive its scalar fields decomposition (ImpSFD), which introduces an alternative numerical interpretation, and Vortex-Particles in Impulse (VPImp) which provides enhanced control over fluid turbulence. Additionally, we integrate error correction post-processing techniques and explore advanced numerical schemes to mitigate dissipation and preserve ultra-fine details.
Furthermore, we initiate discussions on the potential of Lamb vectors, supported by preliminary experiments and results. This vector field, both intuitively and empirically, appears to have a deep connection to the essence of turbulence, warranting further investigation.
This thesis underscores the value of these fields in improving the accuracy, control, and visual quality of fluid simulations, offering directions for future research
Polarimetric Capture and Differentiable Rendering
Get PDFMany scientific fields rely on the capture and modeling of light to extract underlying information about the world. Often, more information can be extracted by capturing more about the nature of the light, such as its spectral shape or polarization state. While polarization is a relatively unexplored topic in computer graphics, when used in tandem with other recent advancements in the field it has enormous potential to improve both forward and inverse models in other scientific disciplines. We demonstrate this potential in two distinct settings in this thesis.
First, we apply the capture of polarized light to an inverse problem in the setting of computational photography. We propose a novel design for a do-it-yourself hyperspectral imaging system driven by taking multiple photographs through tunable, polarization-induced, spectral filters. These filters can generate a continuous family of broadband transmission spectra via simple rotations of stacked polarizers and waveplates. Our prototype demonstrates that our approach can achieve comparable quality to prior work at reduced cost, while the new design space holds ample opportunity for increased quality and flexibility with professional manufacturing.
Next, we investigate the capture of polarized light in the setting of remote sensing. Existing forward models used by the remote sensing community are typically accurate and fast, but sacrifice flexibility by assuming the atmosphere or ocean is composed of plane-parallel layers that are laterally homogeneous. Monte Carlo forward models, such as those favored by the computer graphics community, can handle more complex scenarios such as 3D spatial heterogeneity, but are relatively slow. We demonstrate that Monte Carlo forward models in computer graphics are capable of sufficient accuracy for remote sensing by extending a forward and inverse modeling framework recently developed in the computer graphics community to simulate simple atmosphere-ocean systems. We show that our framework is capable of achieving error on par with codes currently used by the remote sensing community on benchmark results. Lastly, we demonstrate that our framework can be used to retrieve parameters in a variety of simple inverse problems
Excess, Melodrama, and Recessive Action in Madame Bovary (1857 & 1949)
No full textScholars often pose melodrama as a category of critical mediation between genre and form, or as an expressive ‘mode of excess’ uniquely capable of addressing the lifeworld of social institutions and crisis under historical capitalism. Theories of recessive action developed by Lauren Berlant and Anne-Lise François propose an alternate genealogy attending to underperformative or inexpressive styles alongside caricatural expressivities of nineteenth-century bourgeois tragedy’s melodramatic mode. To thread out Berlant and François’ experimental archive of a reticent aesthetics, this presentation considers different instances of excess from Madame Bovary—Flaubert’s 1857 novel and Vincente Minelli’s 1949 film adaptation—in relation to the historical aesthetics of melodrama and minimal realization. Two potential modes of recessive action, mediocrity and the open secret, become melodramatic exemplars for Minelli’s filmic scan of Flaubert’s novel. And in the novel, figural pattern of litotes and hyperbole in scenes of negotiation, as well as musical allegories of articulation, punctation, identification and lyric voice, suspend Bovary’s relations of surface and convention between melodramatic form and content within latent non-identities of affirmative reticence. Gestures of subtractive revelation can show up otherwise: moments of Flaubert’s style indirect libre, positioned in line with reticent or recessive aesthetics of ‘suspended relational clarity,’ develop experimental readings of melodrama as a diffuse formal mode whereby the text calls for re-tuned apprehension of a scene’s viscera
Ritual as Resistance: Orality, Identity, and Soviet Kazakh Childhood in Sayin Muratbekov’s The Smell of Wormwood
No full textThis essay analyzes The Smell of Wormwood (1967), a short novel by Soviet Kazakh author Sayin Muratbekov, set in a Kazakh village during World War II. The narrative centers on Ayan, a nine-year-old orphan with a physical disability, who copes with trauma and marginalization by becoming a storyteller for the village children. Focusing on the role of Kazakh folklore—particularly storytelling and initiation rituals—I explore how these cultural forms shape the novel’s characters, structure, and central conflict. Drawing on Arnold van Gennep and Victor Turner’s theories of ritual, I argue that Ayan’s storytelling creates an informal liminal space through which the children undergo symbolic transformation. In doing so, the novel gestures toward a collective memory of the Kazakh nomadic past—a cultural lineage that Soviet colonialism sought to suppress. Although The Smell of Wormwood passed Soviet censorship and was widely taught in Kazakh schools as ideologically acceptable children’s literature, I contend that it operates as a decolonial text. By embedding oral traditions within a Soviet literary framework, the novel subtly enacts cultural resistance and affirms Kazakh identity
Long-Range Transport of Canadian Wildfire Smoke: Public Health and Earth System Impacts
Get PDFIn the summer of 2023, Canada experienced an unprecedented wildfire season with total burned area reaching 15 million hectares, doubling the previous record. Notably, smoke plumes from Eastern Canada were transported to the Northeast United States, resulting in record-breaking measures of poor air quality in early June 2023. This thesis employs a multi-method analysis to understand the June 2023 Quebec wildfires, focusing on atmospheric science, public health effects in New York City, and albedo changes on the Greenland ice sheet. Through this analysis, I find record-high monthly mean aerosol optical depth values in June 2023 in three regions– the Northeast United States, the North Atlantic, and Western Europe– compared to the historical record that reaches back to 2002. I also find disproportionate increases in asthma-associated emergency department visits in New York City across different demographic covariates. Finally, analysis of aerosol optical depth and ice sheet albedo in Greenland suggests that wildfire smoke did pass over the ice sheet, but the resolution and extent of these datasets makes it difficult to determine whether deposition occurred. This work illustrates the far-reaching nature of climate extremes, which impact people and the planet in unexpected ways. The thesis also highlights climate change as an environmental justice concern and emphasizes the importance of multidisciplinary methods in studying climate extremes