arXiv subjects

Christopher Gardner

Publications and source records attributed to Christopher Gardner.

7 recordsLinked to original sources

arXiv2026

Generative Artificial Intelligence creates delicious, sustainable, and nutritious burgers

Food choices shape both human and planetary health; yet, designing foods that are delicious, nutritious, and sustainable remains challenging. Here we show that generative artificial intelligence can learn the structure of the human palate directly from large-scale, human-generated recipe data to create novel foods within a structured design space. Using burgers as a model system, the generative AI rediscovers the classic Big Mac without explicit supervision and generates novel burgers optimized for deliciousness, sustainability, or nutrition. Compared to the Big Mac, its delicious burgers score the same or better in overall liking, flavor, and texture in a blinded sensory evaluation conducted in a restaurant setting with 101 participants; its mushroom burger achieves an environmental impact score more than an order of magnitude lower; and its bean burger attains nearly twice the nutritional score. Together, these results establish generative AI as a quantitative framework for learning human taste and navigating complex trade-offs in principled food design.

cs.CE↗

arXiv2025

Phase matching in Vector Beam Driven High Harmonic Generation with 3D-printed Gas Cells

We present experimental results of high harmonic generation(HHG) driven by a 1300 nm beam in three different polarization states: linear, radial, and azimuthal. We found that the optimal pressure for phasematching was roughly twice as high for the vector beam drivers than the linear driver. We attribute this difference in pressure primarily to the Gouy phase, which differs by a factor of two between the linear and vector polarization states. We demonstrate a target for HHG that produces a uniform pressure profile in the interaction region that is nearly identical to the backing pressure, and preserves the mode of the driving beam. We provide characterization and validation of this technique through flow simulations and experimental measurements.

physics.optics↗

arXiv2025

Spatiotemporal shaping of broadband helical light pulses at relativistic intensities

Spatiotemporal control of laser pulses at relativistic intensities is a longstanding goal with broad implications in laser-plasma acceleration, high-brightness radiation sources, and extreme-field science. Laser pulses with helical spatiotemporal intensity profiles, often referred to as light springs, carry multiple spectral and orbital angular momentum (OAM) modes, producing a rotating intensity profile capable of coupling directly to helical plasma waves. Until now, light springs have only been realized on low-power systems, limited by optical damage thresholds and large-aperture beamline constraints. Here, we report the first experimental realization of light springs at relativistic intensities, achieving peak intensities above $1.4\times10^{18}$ W/cm$^{2}$. Our approach spectrally splits a high-power laser pulse, imprints distinct helical phases on each component, and coherently recombines them. Hyperspectral imaging, off-axis holography, and spectral phase reconstruction confirm excellent agreement with theory and reveal the potential to drive superluminal rotational velocities. Introducing spectral chirp demonstrates further control of the temporal evolution of the transverse mode structure. This platform opens new regimes of ultra-intense laser-plasma interaction where laser OAM can directly couple to plasma OAM.

physics.optics↗

arXiv2021

Real-time reconstruction of intense, ultrafast laser pulses using deep learning

Ultrafast lasers ($< 500$ fs) have enabled laser-matter interactions at intensities exceeding $10^{18} \rm{Wcm}^{-2}$ with only millijoules of laser energy. However, as pulse durations become shorter, larger spectral bandwidths are required. Increasing the bandwidth causes the temporal structure to be increasingly sensitive to spectral phase, yet measuring the spectral phase of a laser pulse is nontrivial. While direct measurements of the spectral phase cannot be done using square-integrable detectors, phase information can be reconstructed by measuring the spectral response of a nonlinear optical effect. We introduce a new deep learning approach using the generalized nonlinear Schr\"{o}dinger equation and self-phase modulation, a $\chi_3$ nonlinearity occurring from material propagation. By training a neural network on numerical simulations of pulses propagating in a known material, the features of spectral change can be use to reconstruct the spectral phase. The technique is also sensitive to the local fluence of the pulse, enabling the full temporal intensity profile to be calculated. We demonstrate our method on a simulated large bandwidth pulse undergoing moderate material dispersion, and an experimentally produced broadband spectrum with substantial material dispersion. Error rates are low, even when modest amounts of noise introduced. With a single plate of glass and an optical spectrometer, single shot phase and fluence measurements are possible in real-time on intense ultrafast laser systems.

physics.optics↗

arXiv2019

Activation Analysis of a Byte-Based Deep Neural Network for Malware Classification

Feature engineering is one of the most costly aspects of developing effective machine learning models, and that cost is even greater in specialized problem domains, like malware classification, where expert skills are necessary to identify useful features. Recent work, however, has shown that deep learning models can be used to automatically learn feature representations directly from the raw, unstructured bytes of the binaries themselves. In this paper, we explore what these models are learning about malware. To do so, we examine the learned features at multiple levels of resolution, from individual byte embeddings to end-to-end analysis of the model. At each step, we connect these byte-oriented activations to their original semantics through parsing and disassembly of the binary to arrive at human-understandable features. Through our results, we identify several interesting features learned by the model and their connection to manually-derived features typically used by traditional machine learning models. Additionally, we explore the impact of training data volume and regularization on the quality of the learned features and the efficacy of the classifiers, revealing the somewhat paradoxical insight that better generalization does not necessarily result in better performance for byte-based malware classifiers.

cs.LG↗

arXiv2008

Measuring Value in Healthcare

A statistical description and model of individual healthcare expenditures in the US has been developed for measuring value in healthcare. We find evidence that healthcare expenditures are quantifiable as an infusion-diffusion process, which can be thought of intuitively as a steady change in the intensity of treatment superimposed on a random process reflecting variations in the efficiency and effectiveness of treatment. The arithmetic mean represents the net average annual cost of healthcare; and when multiplied by the arithmetic standard deviation, which represents the effective risk, the result is a measure of healthcare cost control. Policymakers, providers, payors, or patients that decrease these parameters are generating value in healthcare. The model has an average absolute prediction error of approximately 10-12% across the range of expenditures which spans 6 orders of magnitude over a nearly 10-year period. For the top 1% of the population with the largest expenditures, representing 20%-30% of total spending on healthcare, a power-law relationship emerges. This relationship also applies to the most expensive medical conditions in the US. A fundamental connection between healthcare expenditures and mathematical finance is found by showing that the process healthcare expenditures follow is similar to a widely used model for managing financial assets, leading to the conclusion that a combination of these two fields may yield useful results.

physics.soc-ph↗

arXiv1997

The structure of quantum Lie algebras for the classical series B_l, C_l and D_l

The structure constants of quantum Lie algebras depend on a quantum deformation parameter q and they reduce to the classical structure constants of a Lie algebra at $q=1$. We explain the relationship between the structure constants of quantum Lie algebras and quantum Clebsch-Gordan coefficients for adjoint x adjoint ---> adjoint. We present a practical method for the determination of these quantum Clebsch-Gordan coefficients and are thus able to give explicit expressions for the structure constants of the quantum Lie algebras associated to the classical Lie algebras B_l, C_l and D_l. In the quantum case also the structure constants of the Cartan subalgebra are non-zero and we observe that they are determined in terms of the simple quantum roots. We introduce an invariant Killing form on the quantum Lie algebras and find that it takes values which are simple q-deformations of the classical ones.

q-alg↗