Publications

This paper presents a detailed analysis of various oscillatory behaviors observed in relation to the calcium and potassium ions in the third-order Morris–Lecar model of giant barnacle muscle fiber. Since, both the calcium and potassium ions exhibit all of the characteristics of memristor fingerprints, we claim that the time-varying calcium and potassium ions in the third-order Morris–Lecar model are actually time-invariant calcium and potassium memristors in the third-order memristive Morris–Lecar model. We confirmed the existence of a small unstable limit cycle oscillation in both the second-order and the third-order Morris–Lecar model by numerically calculating the basin of attraction of the asymptotically stable equilibrium point associated with two subcritical Hopf bifurcation points. We also describe a comprehensive analysis of the generation of oscillations in third-order memristive Morris–Lecar model via small-signal circuit analysis and a subcritical Hopf bifurcation phenomenon.

An entry from the Cambridge Structural Database, the world’s repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

In higher plants, cellulose is synthesized at the plasma membrane by the cellulose synthase (CESA) complex. The catalytic core of the complex is believed to be composed of three types of CESA subunits. Indirect evidence suggests that the complex associated with primary wall cellulose deposition consists of CESA1, -3, and -6 in Arabidopsis thaliana . However, phenotypes associated with mutations in two of these genes, CESA1 and -6 , suggest unequal contribution by the different CESAs to overall enzymatic activity of the complex. We present evidence that the primary complex requires three unique types of components, CESA1-, CESA3-, and CESA6-related, for activity. Removal of any of these components results in gametophytic lethality due to pollen defects, demonstrating that primary-wall cellulose synthesis is necessary for pollen development. We also show that the CESA6-related CESAs are partially functionally redundant.

This work explores techniques to scale up image-based end-to-end learning for dexterous grasping with an arm + hand system. Unlike state-based RL, vision-based RL is much more memory inefficient, resulting in relatively low batch sizes, which is not amenable for algorithms like PPO. Nevertheless, it is still an attractive method as unlike the more commonly used techniques which distill state-based policies into vision networks, end-to-end RL can allow for emergent active vision behaviors. We identify a key bottleneck in training these policies is the way most existing simulators scale to multiple GPUs using traditional data parallelism techniques. We propose a new method where we disaggregate the simulator and RL (both training and experience buffers) onto separate GPUs. On a node with four GPUs, we have the simulator running on three of them, and PPO running on the fourth. We are able to show that with the same number of GPUs, we can double the number of existing environments compared to the previous baseline of standard data parallelism. This allows us to train vision-based environments, end-to-end with depth, which were previously performing far worse with the baseline. We train and distill both depth and state-based policies into stereo RGB networks and show that depth distillation leads to better results, both in simulation and reality. This improvement is likely due to the observability gap between state and vision policies which does not exist when distilling depth policies into stereo RGB. We further show that the increased batch size brought about by disaggregated simulation also improves real world performance. When deploying in the real world, we improve upon the previous state-of-the-art vision-based results using our end-to-end policies.

Large Language Models (LLMs) are evolving beyond their classical role of providing information within dialogue systems to actively engaging with tools and performing actions on real-world applications and services. Today, humans verify the correctness and appropriateness of the LLM-generated outputs (e.g., code, functions, or actions) before putting them into real-world execution. This poses significant challenges as code comprehension is well known to be notoriously difficult. In this paper, we study how humans can efficiently collaborate with, delegate to, and supervise autonomous LLMs in the future. We argue that in many cases, "post-facto validation" - verifying the correctness of a proposed action after seeing the output - is much easier than the aforementioned "pre-facto validation" setting. The core concept behind enabling a post-facto validation system is the integration of an intuitive undo feature, and establishing a damage confinement for the LLM-generated actions as effective strategies to mitigate the associated risks. Using this, a human can now either revert the effect of an LLM-generated output or be confident that the potential risk is bounded. We believe this is critical to unlock the potential for LLM agents to interact with applications and services with limited (post-facto) human involvement. We describe the design and implementation of our open-source runtime for executing LLM actions, Gorilla Execution Engine (GoEX), and present open research questions towards realizing the goal of LLMs and applications interacting with each other with minimal human supervision. We release GoEX at https://github.com/ShishirPatil/gorilla/.

Abstract not Available.

Catalytic hydrogen/deuterium exchange on a platinum (111) single crystal and its poisoning with carbon monoxide was studied using scanning tunneling microscopy (STM), X-ray photoelectron spectroscopy (XPS), sum frequency generation vibrational spectroscopy (SFG), and mass spectrometry under reaction conditions at pressures in the mTorr to atmospheric range. At 298 K and in the presence of 200 mTorr of hydrogen and 20 mTorr of deuterium the surface is catalytically active, producing HD with activation energy of approximately 5.3 kcal/mol. Addition of 5 mTorr of CO stops the reaction completely. In situ STM images reveal an ordered surface structure of chemisorbed CO. At 353 K the addition of 5 mTorr of CO slows the reaction by 3 orders of magnitude, but HD production continues with an activation energy of 17.4 kcal/mol. Changes in coverage and adsorption site of CO were followed by XPS and SFG up to a temperature of 480 K. From these data, a CO dominated, mobile and catalytically active surface model is proposed.

Sensitivity to social evaluation has been proposed as a potential marker or risk factor for depression, and has also been theorized to increase with pubertal maturation. This study utilized an ecologically valid paradigm to test the hypothesis that adolescents with major depressive disorder (MDD) would show altered reactivity to peer rejection and acceptance relative to healthy controls in a network of ventral brain regions implicated in affective processing of social information. A total of 48 adolescents (ages 11-17), including 21 with a current diagnosis of MDD and 27 age- and gender-matched controls, received rigged acceptance and rejection feedback from fictitious peers during a simulated online peer interaction during functional neuroimaging. MDD youth showed increased activation to rejection relative to controls in the bilateral amygdala, subgenual anterior cingulate, left anterior insula and left nucleus accumbens. MDD and healthy youth did not differ in response to acceptance. Youth more advanced in pubertal maturation also showed increased reactivity to rejection in the bilateral amygdala/parahippocampal gyrus and the caudate/subgenual anterior cingulate, and these effects remained significant when controlling for chronological age. Findings suggest that increased reactivity to peer rejection is a normative developmental process associated with pubertal development, but is particularly enhanced among youth with depression.

Although fatigue is by far the most common mode of failure of structural materials, mechanistic understanding is still lacking. For example, the fundamental Paris law which relates the crack growth rate to stress-intensity factor range is still phenomenological and no reliable mechanistic model has been established for a given material or class of materials despite numerous investigations over a half a century. This work is an attempt to theoretically model fatigue crack propagation induced by alternating crack-tip plastic blunting and re-sharpening in the mid-range of growth rates on the basis of inputs from experiments that measure macroscopic material behavior, e.g., response to uniaxial cycling loading. In particular, we attempt to predict Paris law behavior by accounting for the material constitutive behavior in response to cyclic loading by modeling crack advance solely in terms of the underlying plastic dissipation. We obtain the steady-state condition for crack growth based on plastic dissipation, numerically using finite element analysis, which involves a methodology to address plastic closure upon unloading. For a given stress-intensity range, we calculate the crack propagation rate from the steady-state condition through each cycle of loading and unloading of a cracked compact-tension specimen, without resorting to any specific criterion for crack advance.