Publications

The restricted active space spin flip (RAS-SF) method is extended to allow ground and excited states of molecular radicals to be described at low cost (for small numbers of spin flips). RAS-SF allows for any number of spin flips and a flexible active space while maintaining pure spin eigenfunctions for all states by maintaining a spin complete set of determinants and using spin-restricted orbitals. The implementation supports both even and odd numbers of electrons, while use of resolution of the identity integrals and a shared memory parallel implementation allow for fast computation. Examples of multiple-bond dissociation, excited states in triradicals, spin conversions in organic multi-radicals, and mixed-valence metal coordination complexes demonstrate the broad usefulness of RAS-SF.

Abstract Immune cells promote the initial metastatic dissemination of carcinoma cells from primary tumors. In contrast to their well-studied functions in the initial stages of metastasis, the specific roles of immunocytes in facilitating progression through the critical later steps of the invasion–metastasis cascade remain poorly understood. Here, we define novel functions of neutrophils in promoting intraluminal survival and extravasation at sites of metastatic dissemination. We show that CD11b+/Ly6G+ neutrophils enhance metastasis formation via two distinct mechanisms. First, neutrophils inhibit natural killer cell function, which leads to a significant increase in the intraluminal survival time of tumor cells. Thereafter, neutrophils operate to facilitate extravasation of tumor cells through the secretion of IL1β and matrix metalloproteinases. These results identify neutrophils as key regulators of intraluminal survival and extravasation through their cross-talk with host cells and disseminating carcinoma cells. Significance: This study provides important insights into the systemic contributions of neutrophils to cancer metastasis by identifying how neutrophils facilitate intermediate steps of the invasion–metastasis cascade. We demonstrate that neutrophils suppress natural killer cell activity and increase extravasation of tumor cells. Cancer Discov; 6(6); 630–49. ©2016 AACR. This article is highlighted in the In This Issue feature, p. 561

STING agonists trigger NK cell–mediated clearance of MHC-deficient tumors resistant to CD8 + T cells and NK-sensitive, MHC-expressing tumors.

Fracture by hydrogen accelerated fatigue crack growth is a severe type of environmental failure. Although fatigue crack growth has been the subject of intense investigation over several decades, a complete mechanistic and predictive model is still lacking. Such a crack growth model is even more rare in the case of hydrogen in view of the lack also of constitutive models for material deformation under cycling loading that account for the hydrogen effect. In this study, we present a model for fatigue crack propagation induced by alternating crack tip plastic blunting and re-sharpening, which in the presence of hydrogen can be accelerated by hydrogen enhanced dislocation motion and generation. The Chaboche constitutive model, which is a nonlinear kinematic hardening model capable of capturing many features of material behavior under cyclic loading, is used for the calculation of the stress and strain fields at the propagating crack tip. The Chaboche model is calibrated using a sequence of experimental data from uniaxial strain-controlled cyclic loading tests and uniaxial stress-controlled ratcheting tests with a low carbon steel, JIS SM490YB, in the absence and presence of hydrogen. The numerical simulation results indicate that the proposed crack propagation model can predict Paris law behavior and can successfully demonstrate acceleration of fatigue crack growth in the presence of hydrogen. Significantly, the profiles of the steady-state opening stress and strain ahead of the fatigue crack tip in a compact tension (C(T)) specimen were found to have sections over which they vary as ln ( 1 / r ) with distance r from the crack tip, consistent with the crack-tip strain field for a non-stationary crack.

Compound AI systems that combine multiple LLM calls, such as self-refine and multi-agent-debate, achieve strong performance on many AI tasks. We address a core question in optimizing compound systems: for each LLM call or module in the system, how should one decide which LLM to use? We show that these LLM choices have a large effect on quality, but the search space is exponential. We propose LLMSelector, an efficient framework for model selection in compound systems, which leverages two key empirical insights: (i) end-to-end performance is often monotonic in how well each module performs, with all other modules held fixed, and (ii) per-module performance can be estimated accurately by an LLM. Building upon these insights, LLMSelector iteratively selects one module and allocates to it the model with the highest module-wise performance, as estimated by an LLM, until no further gain is possible. LLMSelector is applicable to any compound system with a bounded number of modules, and its number of API calls scales linearly with the number of modules, achieving high-quality model allocation both empirically and theoretically. Experiments with popular compound systems such as multi-agent debate and self-refine using LLMs such as GPT-4o, Claude 3.5 Sonnet and Gemini 1.5 show that LLMSelector confers 5%-70% accuracy gains compared to using the same LLM for all modules.

An application of complex dynamics and chaos in neural networks to information processing is studied. Mathematical models based on piecewise-linear maps implementing basic functions of information processing via complex dynamics and chaos are discussed. Realizations of these models by neural networks are presented. In contrast to other methods of using neural networks and associative memory to store information, the information is stored in dynamical attractors such as limit cycles, rather than equilibrium points. Retrieval of information corresponds to getting the state into the basin of attraction of the attractor. We show that noise-corrupted information or partial information are sufficient to drive the state into the basin of attraction of the attractor, thus these systems exhibit the property of associative memory.

The objective of the present study was to develop a heterogeneous catalyst for the hydrogenation of butanal that could function in the presence of CO and propene and, hence, could be used in a tandem reactor to enable the gas-phase conversion of propene and synthesis gas to butanol. To this end, we investigated the activity of silica-supported Shvo's catalyst (Shvo/SiO2) for the gas-phase hydrogenation of butanal. Experiments were performed to determine the kinetics of n- and iso-butanal hydrogenation. The apparent activation energies and the apparent partial pressure dependencies of n- and iso-butanal, H 2, and CO on the rates of n- and iso-butanol formation were determined. A mechanism for butanal hydrogenation was proposed to rationalize the observed kinetics and some of the reaction intermediates were observed by in situ infrared and 31P MAS NMR spectroscopy. It was found that Shvo/SiO2 was inhibited by SX (SX = sulfoxanthphos) and CO, and is inactive for alkene hydrogenation. The tandem catalytic conversion of propene and synthesis gas to butanol was then carried out using a SX-Rh supported ionic liquid phase (SILP) catalyst to promote the hydroformylation of propene to butanal and Shvo/SiO2 to promote the hydrogenation of butanal to butanol. The rate expressions describing the kinetics of each of the catalysts were then used to predict operating conditions required to achieve high conversion of propene to butanol. Under the most favorable conditions examined (H2/CO = 10), an overall yield of 13% to butanol was achieved with 15% propene conversion and 90% aldehyde conversion at a temperature of 413 K. © 2013 Elsevier Inc. All rights reserved.

The design, synthesis, and evaluation of the Copper-Gad (CG) family, a new class of copper-activated magnetic resonance imaging (MRI) contrast agents, are presented. These indicators comprise a Gd(3+)-DO3A core coupled to various thioether-rich receptors for copper-induced relaxivity switching. In the absence of copper ions, inner-sphere water binding to the Gd(3+) chelate is restricted, resulting in low longitudinal relaxivity values (r(1) = 1.2-2.2 mM(-1) s(-1) measured at 60 MHz). Addition of Cu(+) to CG2, CG3, CG4, and CG5 and either Cu(+) or Cu(2+) to CG6 triggers marked enhancements in relaxivity (r(1) = 2.3-6.9 mM(-1) s(-1)). CG2 and CG3 exhibit the greatest turn-on responses, going from r(1) = 1.5 mM(-1) s(-1) in the absence of Cu(+) to r(1) = 6.9 mM(-1) s(-1) upon Cu(+) binding (a 360% increase). The CG sensors are highly selective for Cu(+) and/or Cu(2+) over competing metal ions at cellular concentrations, including Zn(2+) at 10-fold higher concentrations. (17)O NMR dysprosium-induced shift and nuclear magnetic relaxation dispersion measurements support a mechanism in which copper-induced changes in the coordination environment of the Gd(3+) core result in increases in q and r(1). T(1)-weighted phantom images establish that the CG sensors are capable of visualizing changes in copper levels by MRI at clinical field strengths.