arXiv subjects

Richard Ho

Publications and source records attributed to Richard Ho.

7 recordsLinked to original sources

arXiv2026

Memory-driven topological ordering during the transition from dormant to migrating epithelia

Transitions from quiescence to collective migration in epithelia underlie wound healing and cancer invasion, yet their physical origin remains poorly understood. Here we show that quiescent epithelial monolayers store spatially contractile stresses that function as a form of mechanical memory. Upon serum-induced reactivation, these pre-stressed regions nucleate extensile asters that emit propagating polarity domain walls. Along these interfaces, topological defects are created, advected and annihilated, leading to defect coarsening with faster kinetics than by elastic interactions. An active elastic model quantitatively reproduces the observed dynamics and identifies stored stress as the origin of rapid topological reorganization. Our results establish a mechanism in which mechanical memory in quiescent epithelia triggers active stress release, driving collective migration via rapid topological ordering, distinct from conventional unjamming and flocking transitions.

cond-mat.soft↗

arXiv2026

AI+HW 2035: Shaping the Next Decade

Artificial intelligence (AI) and hardware (HW) are advancing at unprecedented rates, yet their trajectories have become inseparably intertwined. The global research community lacks a cohesive, long-term vision to strategically coordinate the development of AI and HW. This fragmentation constrains progress toward holistic, sustainable, and adaptive AI systems capable of learning, reasoning, and operating efficiently across cloud, edge, and physical environments. The future of AI depends not only on scaling intelligence, but on scaling efficiency, achieving exponential gains in intelligence per joule, rather than unbounded compute consumption. Addressing this grand challenge requires rethinking the entire computing stack. This vision paper lays out a 10-year roadmap for AI+HW co-design and co-development, spanning algorithms, architectures, systems, and sustainability. We articulate key insights that redefine scaling around energy efficiency, system-level integration, and cross-layer optimization. We identify key challenges and opportunities, candidly assess potential obstacles and pitfalls, and propose integrated solutions grounded in algorithmic innovation, hardware advances, and software abstraction. Looking ahead, we define what success means in 10 years: achieving a 1000x improvement in efficiency for AI training and inference; enabling energy-aware, self-optimizing systems that seamlessly span cloud, edge, and physical AI; democratizing access to advanced AI infrastructure; and embedding human-centric principles into the design of intelligent systems. Finally, we outline concrete action items for academia, industry, government, and the broader community, calling for coordinated national initiatives, shared infrastructure, workforce development, cross-agency collaboration, and sustained public-private partnerships to ensure that AI+HW co-design becomes a unifying long-term mission.

cs.AI↗

arXiv2024

A Survey on Human-AI Collaboration with Large Foundation Models

As the capabilities of artificial intelligence (AI) continue to expand rapidly, Human-AI (HAI) Collaboration, combining human intellect and AI systems, has become pivotal for advancing problem-solving and decision-making processes. The advent of Large Foundation Models (LFMs) has greatly expanded its potential, offering unprecedented capabilities by leveraging vast amounts of data to understand and predict complex patterns. At the same time, realizing this potential responsibly requires addressing persistent challenges related to safety, fairness, and control. This paper reviews the crucial integration of LFMs with HAI, highlighting both opportunities and risks. We structure our analysis around four areas: human-guided model development, collaborative design principles, ethical and governance frameworks, and applications in high-stakes domains. Our review shows that successful HAI systems are not the automatic result of stronger models but the product of careful, human-centered design. By identifying key open challenges, this survey aims to give insight into current and future research that turns the raw power of LFMs into partnerships that are reliable, trustworthy, and beneficial to society.

cs.AI↗

arXiv2020

Effect of spatial dimension on a model of fluid turbulence

A numerical study of the $d$-dimensional Eddy Damped Quasi-Normal Markovian equations is performed to investigate the dependence on spatial dimension of homogeneous isotropic fluid turbulence. Relationships between structure functions and energy and transfer spectra are derived for the $d$-dimensional case. Additionally, an equation for the $d$-dimensional enstrophy analogue is derived and related to the velocity derivative skewness. Comparisons are made to recent four dimensional direct numerical simulation results. Measured energy spectra show a magnified bottleneck effect which grows with dimension whilst transfer spectra show a varying peak in the non-linear energy transfer as the dimension is increased. These results are consistent with an increased forward energy transfer at higher dimensions, further evidenced by measurements of a larger asymptotic dissipation rate with growing dimension. The enstrophy production term, related to the velocity derivative skewness, is seen to reach a maximum at around five dimensions and may reach zero in the limit of infinite dimensions, raising interesting questions about the nature of turbulence in this limit.

physics.flu-dyn↗

arXiv2020

Chip Placement with Deep Reinforcement Learning

In this work, we present a learning-based approach to chip placement, one of the most complex and time-consuming stages of the chip design process. Unlike prior methods, our approach has the ability to learn from past experience and improve over time. In particular, as we train over a greater number of chip blocks, our method becomes better at rapidly generating optimized placements for previously unseen chip blocks. To achieve these results, we pose placement as a Reinforcement Learning (RL) problem and train an agent to place the nodes of a chip netlist onto a chip canvas. To enable our RL policy to generalize to unseen blocks, we ground representation learning in the supervised task of predicting placement quality. By designing a neural architecture that can accurately predict reward across a wide variety of netlists and their placements, we are able to generate rich feature embeddings of the input netlists. We then use this architecture as the encoder of our policy and value networks to enable transfer learning. Our objective is to minimize PPA (power, performance, and area), and we show that, in under 6 hours, our method can generate placements that are superhuman or comparable on modern accelerator netlists, whereas existing baselines require human experts in the loop and take several weeks.

cs.LG↗

arXiv2019

Fluctuations of Lyapunov Exponents in homogeneous and isotropic turbulence

In the context of the analysis of the chaotic properties of homogeneous and isotropic turbulence, direct numerical simulations are used to study the fluctuations of the finite time Lyapunov exponent (FTLE) and its relation to Reynolds number, lattice size and the choice of the steptime used to compute the Lyapunov exponents. The results show that using the FTLE method produces Lyapunov exponents that are remarkably stable under the variation of the steptime and lattice size. Furthermore, it reaches such stability faster than other characteristic quantities such as energy and dissipation rate. These results remain even if the steptime is made arbitrarily small. A discrepancy is also resolved between previous measurements of the dependence on the Reynolds number of the Lyapunov exponent. The signal produced by different variables in the steady state is analyzed and the self decorrelation time is used to determine the run time needed in the simulations to obtain proper statistics for each variable. Finally, a brief analysis on MHD flows is also presented, which shows that the Lyapunov exponent is still a robust measure in the simulations, although the Lyapunov exponent scaling with Reynolds number is significantly different from that of magnetically neutral hydrodynamic fluids.

physics.flu-dyn↗

arXiv2018

Chaotic behavior of Eulerian MHD turbulence

We study the chaotic properties of a turbulent conducting fluid using direct numerical simulation in the Eulerian frame. The maximal Lyapunov exponent is measured for simulations with varying Reynolds number and magnetic Prandtl number. We extend the Ruelle theory of hydrodynamic turbulence to magnetohydrodynamic turbulence as a working hypothesis and find broad agreement with results. In other simulations we introduce magnetic helicity and these simulations show a diminution of chaos, which is expected to be eliminated at maximum helicity. We also find that the difference between two initially close fields grows linearly at late times, which was also recently found in hydrodynamics. This linear growth rate is found to be dependent on the dissipation rate of the relevant field. We discuss the important consequences this linear growth has on predictability. We infer that the chaos in the system is totally dominated by the velocity field and connect this work to real magnetic systems such as solar weather and confined plasmas.

physics.flu-dyn↗