Search

Low-Rank Adaptation (LoRA) and other parameter-efficient fine-tuning (PEFT) methods provide low-memory, storage-efficient solutions for personalizing text-to-image models. However, these methods offer little to no improvement in wall-clock training time or the number of steps needed for convergence compared to full model fine-tuning. While PEFT methods assume that shifts in generated distributions (from base to fine-tuned models) can be effectively modeled through weight changes in a low-rank subspace, they fail to leverage knowledge of common use cases, which typically focus on capturing specific styles or identities. Observing that desired outputs often comprise only a small subset of the possible domain covered by LoRA training, we propose reducing the search space by incorporating a prior over regions of interest. We demonstrate that training a hypernetwork model to generate LoRA weights can achieve competitive quality for specific domains while enabling near-instantaneous conditioning on user input, in contrast to traditional training methods that require thousands of steps., Comment: 9 pages, 6 figures

Eye tracking technology offers great potential for improving road safety. It is already being built into vehicles, namely cars and trucks. When this technology is integrated into transit service vehicles, employees, i.e., bus drivers, will be subject to being eye tracked on their job. Although there is much research effort advancing algorithms for eye tracking in transportation, less is known about how end users perceive this technology, especially when interacting with it in an employer-mandated context. In this first study of its kind, we investigated transit bus operators' perceptions of eye tracking technology. From a methodological perspective, we introduce a mixed methods approach where participants experience the technology first-hand and then reflect on their experience while viewing a playback of the recorded data. Thematic analysis of the interview transcripts reveals interesting potential uses of eye tracking in this work context and surfaces transit operators' fears and concerns about this technology.

Zero-shot personalized image generation models aim to produce images that align with both a given text prompt and subject image, requiring the model to effectively incorporate both sources of guidance. However, existing methods often struggle to capture fine-grained subject details and frequently prioritize one form of guidance over the other, resulting in suboptimal subject encoding and an imbalance in the generated images. In this study, we uncover key insights into achieving high-quality balances on subject identity preservation and text-following, notably that 1) the design of the subject image encoder critically influences subject identity preservation, and 2) the text and subject guidance should take effect at different denoising stages. Building on these insights, we introduce a new approach, EZIGen, that employs two main components: a carefully crafted subject image encoder based on the pre-trained UNet of the Stable Diffusion model, following a process that balances the two guidances by separating their dominance stage and revisiting certain time steps to bootstrap subject transfer quality. Through these two components, EZIGen achieves state-of-the-art results on multiple personalized generation benchmarks with a unified model and 100 times less training data. Demo Page: zichengduan.github.io/pages/EZIGen/index.html

We present measurements on Fe$_2$O$_3$ amorphization and melt under laser-driven shock compression up to 209(10) GPa via time-resolved in situ x-ray diffraction. At 122(3) GPa, a diffuse signal is observed indicating the presence of a non-crystalline phase. Structure factors have been extracted up to 182(6) GPa showing the presence of two well-defined peaks. A rapid change in the intensity ratio of the two peaks is identified between 145(10) and 151(10) GPa, indicative of a phase change. Present DFT+$U$ calculations of temperatures along Fe$_2$O$_3$ Hugoniot are in agreement with SESAME 7440 and indicate relatively low temperatures, below 2000 K, up to 150 GPa. The non-crystalline diffuse scattering is thus consistent with the - as yet unreported - shock amorphization of Fe$_2$O$_3$ between 122(3) and 145(10) GPa, followed by an amorphous-to-liquid transition above 151(10) GPa. Upon release, a non-crystalline phase is observed alongside crystalline $\alpha$-Fe$_2$O$_3$. The extracted structure factor and pair distribution function of this release phase resemble those reported for Fe$_2$O$_3$ melt at ambient pressure., Comment: 11 pages, 4 figures, under review

Radiative heat transfer is a fundamental process in high energy density physics and inertial fusion. Accurately predicting the behavior of Marshak waves across a wide range of material properties and drive conditions is crucial for design and analysis of these systems. Conventional numerical solvers and analytical approximations often face challenges in terms of accuracy and computational efficiency. In this work, we propose a novel approach to model Marshak waves using Fourier Neural Operators (FNO). We develop two FNO-based models: (1) a base model that learns the mapping between the drive condition and material properties to a solution approximation based on the widely used analytic model by Hammer & Rosen (2003), and (2) a model that corrects the inaccuracies of the analytic approximation by learning the mapping to a more accurate numerical solution. Our results demonstrate the strong generalization capabilities of the FNOs and show significant improvements in prediction accuracy compared to the base analytic model.

Attention mechanism has been crucial for image diffusion models, however, their quadratic computational complexity limits the sizes of images we can process within reasonable time and memory constraints. This paper investigates the importance of dense attention in generative image models, which often contain redundant features, making them suitable for sparser attention mechanisms. We propose a novel training-free method ToDo that relies on token downsampling of key and value tokens to accelerate Stable Diffusion inference by up to 2x for common sizes and up to 4.5x or more for high resolutions like 2048x2048. We demonstrate that our approach outperforms previous methods in balancing efficient throughput and fidelity.

Resonant inelastic x-ray scattering (RIXS) is a widely used spectroscopic technique, providing access to the electronic structure and dynamics of atoms, molecules, and solids. However, RIXS requires a narrow bandwidth x-ray probe to achieve high spectral resolution. The challenges in delivering an energetic monochromated beam from an x-ray free electron laser (XFEL) thus limit its use in few-shot experiments, including for the study of high energy density systems. Here we demonstrate that by correlating the measurements of the self-amplified spontaneous emission (SASE) spectrum of an XFEL with the RIXS signal, using a dynamic kernel deconvolution with a neural surrogate, we can achieve electronic structure resolutions substantially higher than those normally afforded by the bandwidth of the incoming x-ray beam. We further show how this technique allows us to discriminate between the valence structures of Fe and Fe$_2$O$_3$, and provides access to temperature measurements as well as M-shell binding energies estimates in warm-dense Fe compounds.

We present a new approach to calculating time eigenvalues of the neutron transport operator (also known as $\alpha$ eigenvalues) by extending the dynamic mode decomposition (DMD) to allow for non-uniform time steps. The new method, called variable dynamic mode decomposition (VDMD), is shown to be accurate when computing eigenvalues for systems that were infeasible with DMD due to a large separation in time scales (such as those that occur in delayed supercritical systems). The $\alpha$ eigenvalues of an infinite medium neutron transport problem with delayed neutrons and consequently having multiple, very different relevant time scales are computed. Furthermore, VDMD is shown to be of similar accuracy to the original DMD approach when computing eigenvalues in other systems where the previously studied DMD approach can be used., Comment: 20 pages, 3 tables, 3 figures

In this manifesto, we put forward the idea of data alchemy as a narrative device to discuss storytelling and transdisciplinarity in visualization. If data is the prima materia of modern science, how does one perform the Great Work? We use text-to-image diffusion-based generative art to develop the concept, and structure our argument in ten propositions, as if they were ten issues of a comic novel on data alchemy: Ad Disco Diffusionem. To follow the argument, the reader must immerse themselves in our miro board, and navigate a multimedia semiotic topology that includes comics, videos, code demos, and ergotic literature in a true alchemic sense. By accessing this paradigm one might find new sources of inspiration for scientific inquiry in familiar places, or get lost in the creative exploration of the unknown. Our colorful, sometimes poetic, exposition should not distract the reader from the seriousness of the ideas discussed, but ultimately it is about the journey., Comment: Submission to IEEE alt.vis 2022. Short paper containing a 4-page comic, and an immense amount of content in a linked miro board

While showing great promise, circuit synthesis techniques that combine numerical optimization with search over circuit structures face scalability challenges due to a large number of parameters, exponential search spaces, and complex objective functions. The LEAP algorithm improves scaling across these dimensions using iterative circuit synthesis, incremental re-optimization, dimensionality reduction, and improved numerical optimization. LEAP draws on the design of the optimal synthesis algorithm QSearch by extending it with an incremental approach to determine constant prefix solutions for a circuit. By narrowing the search space, LEAP improves scalability from four to six qubit circuits. LEAP was evaluated with known quantum circuits such as QFT and physical simulation circuits like the VQE, TFIM, and QITE. LEAP can compile four qubit unitaries up to $59\times$ faster than QSearch and five and six qubit unitaries with up to $1.2\times$ fewer CNOTs compared to the QFAST package. LEAP can reduce the CNOT count by up to $36\times$, or $7\times$ on average, compared to the CQC Tket compiler. Despite its heuristics, LEAP has generated optimal circuits for many test cases with a priori known solutions. The techniques introduced by LEAP are applicable to other numerical-optimization-based synthesis approaches., Comment: 21 pages

The success of the current generation of Noisy Intermediate-Scale Quantum (NISQ) hardware shows that quantum hardware may be able to tackle complex problems even without error correction. One outstanding issue is that of coherent errors arising from the increased complexity of these devices. These errors can accumulate through a circuit, making their impact on algorithms hard to predict and mitigate. Iterative algorithms like Quantum Imaginary Time Evolution are susceptible to these errors. This article presents the combination of both noise tailoring using Randomized Compiling and error mitigation with a purification. We also show that Cycle Benchmarking gives an estimate of the reliability of the purification. We apply this method to the Quantum Imaginary Time Evolution of a Transverse Field Ising Model and report an energy estimation and a ground state infidelity both below 1\%. Our methodology is general and can be used for other algorithms and platforms. We show how combining noise tailoring and error mitigation will push forward the performance of NISQ devices., Comment: Data has been updated

Dynamic simulation of materials is a promising application for near-term quantum computers. Current algorithms for Hamiltonian simulation, however, produce circuits that grow in depth with increasing simulation time, limiting feasible simulations to short-time dynamics. Here, we present a method for generating circuits that are constant in depth with increasing simulation time for a subset of one-dimensional materials Hamiltonians, thereby enabling simulations out to arbitrarily long times. Furthermore, by removing the effective limit on the number of feasibly simulatable time-steps, the constant-depth circuits enable Trotter error to be made negligibly small by allowing simulations to be broken into arbitrarily many time-steps. Composed of two-qubit matchgates on nearest-neighbor qubits, these constant-depth circuits are constructed based on a set of multi-matchgate identity relationships. For an $N$-spin system, the constant-depth circuit contains only $\mathcal{O}(N^2)$ CNOT gates. When compared to standard Hamiltonian simulation algorithms, our method generates circuits with order-of-magnitude fewer gates, which allows us to successfully simulate the long-time dynamics of systems with up to 5 spins on available quantum hardware. This paves the way for simulations of long-time dynamics for scientifically and technologically relevant quantum materials, enabling the observation of interesting and important atomic-level physics., Comment: 11 pages, 3 figures, 1 table

The successful implementation of algorithms on quantum processors relies on the accurate control of quantum bits (qubits) to perform logic gate operations. In this era of noisy intermediate-scale quantum (NISQ) computing, systematic miscalibrations, drift, and crosstalk in the control of qubits can lead to a coherent form of error which has no classical analog. Coherent errors severely limit the performance of quantum algorithms in an unpredictable manner, and mitigating their impact is necessary for realizing reliable quantum computations. Moreover, the average error rates measured by randomized benchmarking and related protocols are not sensitive to the full impact of coherent errors, and therefore do not reliably predict the global performance of quantum algorithms, leaving us unprepared to validate the accuracy of future large-scale quantum computations. Randomized compiling is a protocol designed to overcome these performance limitations by converting coherent errors into stochastic noise, dramatically reducing unpredictable errors in quantum algorithms and enabling accurate predictions of algorithmic performance from error rates measured via cycle benchmarking. In this work, we demonstrate significant performance gains under randomized compiling for the four-qubit quantum Fourier transform algorithm and for random circuits of variable depth on a superconducting quantum processor. Additionally, we accurately predict algorithm performance using experimentally-measured error rates. Our results demonstrate that randomized compiling can be utilized to leverage and predict the capabilities of modern-day noisy quantum processors, paving the way forward for scalable quantum computing.

We present an algorithm for compiling arbitrary unitaries into a sequence of gates native to a quantum processor. As accurate CNOT gates are hard for the foreseeable Noisy- Intermediate-Scale Quantum devices era, our A* inspired algorithm attempts to minimize their count, while accounting for connectivity. We discuss the search strategy together with metrics to expand the solution frontier. For a workload of circuits with complexity appropriate for the NISQ era, we produce solutions well within the best upper bounds published in literature and match or exceed hand tuned implementations, as well as other existing synthesis alternatives. In particular, when comparing against state-of-the-art available synthesis packages we show 2.4x average (up to 5.3x) reduction in CNOT count. We also show how to re-target the algorithm for a different chip topology and native gate set, while obtaining similar quality results. We believe that empirical tools like ours can facilitate algorithmic exploration, gate set discovery for quantum processor designers, as well as providing useful optimization blocks within the quantum compilation tool-chain., Comment: Presented at the 3rd International Workshop on Quantum Compilation as part of the International Conference On Computer Aided Design 2019

We present several results related to statistics for elliptic curves over a finite field $\mathbb{F}_p$ as corollaries of a general theorem about averages of Euler products that we demonstrate. In this general framework, we can reprove known results such as the average Lang-Trotter conjecture, the average Koblitz conjecture, and the vertical Sato-Tate conjecture, even for very short intervals, not accessible by previous methods. We also compute statistics for new questions, such as the problem of amicable pairs and aliquot cycles, first introduced by Silverman and Stange. Our technique is rather flexible and should be easily applicable to a wide range of similar problems. The starting point of our results is a theorem of Gekeler which gives a reinterpretation of Deuring's theorem in terms of an Euler product involving random matrices, thus making a direct connection between the (conjectural) horizontal distributions and the vertical distributions. Our main technical result then shows that, under certain conditions, a weighted average of Euler products is asymptotic to the Euler product of the average factors., Comment: 57 pages, to appear in Math. Ann.; corrected the proof of Theorem 3.1; added Remark 4.1; some other minor changes

Letting $p$ vary over all primes and $E$ vary over all elliptic curves over the finite field $\mathbb{F}_p$, we study the frequency to which a given group $G$ arises as a group of points $E(\mathbb{F}_p)$. It is well-known that the only permissible groups are of the form $G_{m,k}:=\mathbb{Z}/m\mathbb{Z}\times \mathbb{Z}/mk\mathbb{Z}$. Given such a candidate group, we let $M(G_{m,k})$ be the frequency to which the group $G_{m,k}$ arises in this way. Previously, the second and fourth named authors determined an asymptotic formula for $M(G_{m,k})$ assuming a conjecture about primes in short arithmetic progressions. In this paper, we prove several unconditional bounds for $M(G_{m,k})$, pointwise and on average. In particular, we show that $M(G_{m,k})$ is bounded above by a constant multiple of the expected quantity when $m\le k^A$ and that the conjectured asymptotic for $M(G_{m,k})$ holds for almost all groups $G_{m,k}$ when $m\le k^{1/4-\epsilon}$. We also apply our methods to study the frequency to which a given integer $N$ arises as the group order $\#E(\mathbb{F}_p)$., Comment: 40 pages, with an appendix by Chantal David, Greg Martin and Ethan Smith. Final version, to appear in the Canad. J. Math. Major reorganization of the paper, with the addition of a new section, where the main results are summarized and explained

Let A be an abelian surface over F_q, the field of q elements. The rational points on A/\F_q form an abelian group A(\F_q) \simeq \Z/n_1\Z \times \Z/n_1 n_2 \Z \times \Z/n_1 n_2 n_3\Z \times\Z/n_1 n_2 n_3 n_4\Z. We are interested in knowing which groups of this shape actually arise as the group of points on some abelian surface over some finite field. For a fixed prime power q, a characterization of the abelian groups that occur was recently found by Rybakov. One can use this characterization to obtain a set of congruences modulo the integers $n_1, n_2, n_3, n_4$ on certain combinations of coefficients of the corresponding Weil polynomials. We use Rybakov's criterion to show that groups \Z/n_1\Z \times \Z/n_1 n_2 \Z \times \Z/n_1 n_2 n_3\Z \times\Z/n_1 n_2 n_3 n_4\Z do not occur if n_1 is very large with respect to n_2, n_2, n_4 (Theorem \ref{splitbound}), and occur with density zero in a wider range of the variables (Theorem \ref{splitbound-average}).