SEARCH · Engineering Papers

Results for “Neural Pathways”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2

All sources NASA NTRS DOE OSTI

NASA NTRS1987

Neural mechanisms by which gravitational stimuli and stress affect the secretion of renin and other hormones

The present goal is to determine by the production of discrete lesions the parts of the hypothalamus and brainstem that are involved in serotonin-mediated increases in renin secretion. A variety of stimuli which act in different ways to increase renin stimuli were developed and standardized. The experiments with p-chloroamphetamine (PCA) demonstrated that there is a serotonergic pathway which projects from the dorsal raphe nuclei to the paraventricular nuclei and the vetromedial nuclei of the hypothalamus; that projection from paraventricular nuclei to the brainstem and spinal cord may be oxytocinergic; and that the pathway from the spinal cord to the renin secreting cells is sympathetic. The demonstration that paraventicular lesions lower circulating renin substrate is important because it raises the possibility that substrate secretion is under neural control, either via the pituitary or by direct neural pathways. The discovery that lesions of the ventromedial nuclei appear to abolish the increase in renin secretion produced by many different stimuli without affecting the concentration of renin substrate in the plasma makes the position of the hypothalamus in the regulation of fluid and electrolyte balance more prominent than previously suspected.

Ganong, William F.↗

NASA NTRS1996

Quantifying the Correlation Between Eye-Movement and Perceptual Responses to Moving Plaids

To determine the extent to which the pursuit eye-movement and perceptual systems share a common neural motion-processing pathway, we simultaneously measure the psychophysical and eye-movement responses to drifting plaids in a direction-discrimination task. Correlation between the noise in the eye movements and percepts would suggest a shared neural substrate. Three observers were asked to track a plaid (Type I 90 deg; TF = 4 Hz, SF = 0.6 c/d; direction = -2, 0, or 2 deg), and to respond whether the motion was to the right or left of vertical. The eye-movement direction is computed from the initial 300 ms of saccade-free tracking (near open-loop). On each trial, we use the eye-movement direction to predict the perceptual response. We then compute the probability of successfully predicting the observer's response. We also use SOC analysis as an alternate measure of the correlation. We compare the data from both these analyses to the predictions of two models in which the eye movements and the percept are driven by either a common noisy signal (correlated model) or by two separate noisy signals (uncorrelated model). We also allow for the fact that our measured eye movements are degraded by eye-tracker noise. This noise causes our measured correlation to be lower than the actual biological correlation. The correlations in our data are higher than those predicted by the uncorrelated model, and are similar to the predictions of the correlated model. The proportion of perceptual responses correctly predicted by the eye movements were 0.66, 0.60, 0.73 for the -2, 0, 2 deg stimuli respectively (correlated model 0.65, 0.60, 0.70; uncorrelated model 0.58, 0.51, 0.60).The SOC proportions were 0.71, 0.64, 0.72 (correlated model 0.66, 0.64, 0.69; uncorrelated model 0.5, 0.5, 0.5). These results show that the oculomotor and perceptual system share a performance-limiting noise source and provide strong evidence that a common neural mechanism (perhaps MT or MST) drives both perception and smooth eye movements.

Beutter, Brent R.↗

DOE OSTI2026

Deciphering the Solvation Structure of Aqueous ZnCl 2 Solutions from X-ray Absorption Spectra Using the Interpretable Graph Neural Network

Machine learning (ML) provides powerful pathways for predicting spectroscopic observables from atomic structures, but its broader impact depends on making model predictions interpretable in terms of physical and chemical principles. Here, we introduce a physics-guided graph neural network (GNN) model that predicts Zn K-edge X-ray spectroscopy (XAS) spectra of aqueous ZnCl 2 solutions. Training data are generated from ab initio XAS calculations on molecular dynamics snapshots obtained using a machine learning interatomic potential. The GNN reproduces experimental spectra across concentrations from dilute (<0.1 m) to highly concentrated (30 m, “water-in-salt”) regimes and scales efficiently to large, disordered liquid systems beyond the reach of conventional ab initio approaches. Gradient-based attribution analysis reveals that the model learns physically meaningful structure-spectrum relationships. Ligand-specific attributions reflect orbital hybridization patterns and the origin of the excitations derived from the density functional theory. Bond-length attributions recover spectral shifts consistent with multiple-scattering theory. Finally, this work bridges data-driven prediction with electronic-structure theory, establishing a general paradigm for interpretable ML that links atomic structure, electronic structure, and spectroscopic observables.

25 ENERGY STORAGE↗

NASA NTRS2020

Dose-Dependent Sensorimotor Impairment in Human Ocular Tracking After Acute Low-Dose Alcohol Administration

Changes in oculomotor behaviours are often used as metrics of sensorimotor disruption due to ethanol (EtOH); however, previous studies have focused on deficits at blood-alcohol concentrations (BACs) above about 0.04%.We investigated the dose dependence of the impairment in oculomotor and ocular behaviours caused by EtOH administration across a range of ultra-low BACs (≤0.035%).We took repeated measures of oculomotor and ocular performance from sixteen participants, both pre- and post-EtOH administration. To assess the neurological impacts across a wide range of brain areas and pathways, our protocol measured 21 largely independent performance metrics extracted from a range of behavioural responses ranging from ocular tracking of radial step-ramp stimuli, to eccentric gaze holding, to pupillary responses evoked by light flashes. Our results show significant impairment of pursuit and visual motion processing at 0.015% BAC, reflecting degraded neural processing within extrastriate cortical pathways. However, catch-up saccades largely compensate for the tracking displacement shortfall caused by low pursuit gain, although there still is significant residual retinal slip and thus degraded dynamic acuity. Furthermore, although saccades are more frequent, their dynamics are more sluggish (i.e. show lower peak velocities) starting at BAC levels as low as 0.035%. Small effects in eccentric gaze holding and no effect in pupillary response dynamics were observed at levels below 0.07%, showing the higher sensitivity of the pursuit response to very low levels of blood alcohol, under the conditions of our study.

visual motion processing↗

DOE OSTI2024

Characterizing climate pathways using feature importance on echo state networks

The 2022 National Defense Strategy of the United States listed climate change as a serious threat to national security. Climate intervention methods, such as stratospheric aerosol injection, have been proposed as mitigation strategies, but the downstream effects of such actions on a complex climate system are not well understood. The development of algorithmic techniques for quantifying relationships between source and impact variables related to a climate event (i.e., a climate pathway) would help inform policy decisions. Data-driven deep learning models have become powerful tools for modeling highly nonlinear relationships and may provide a route to characterize climate variable relationships. In this paper, we explore the use of an echo state network (ESN) for characterizing climate pathways. ESNs are a computationally efficient neural network variation designed for temporal data, and recent work proposes ESNs as a useful tool for forecasting spatiotemporal climate data. However, ESNs are noninterpretable black-box models along with other neural networks. The lack of model transparency poses a hurdle for understanding variable relationships. We address this issue by developing feature importance methods for ESNs in the context of spatiotemporal data to quantify variable relationships captured by the model. We conduct a simulation study to assess and compare the feature importance techniques, and we demonstrate the approach on reanalysis climate data. In the climate application, we consider a time period that includes the 1991 volcanic eruption of Mount Pinatubo. This event was a significant stratospheric aerosol injection, which acts as a proxy for an anthropogenic stratospheric aerosol injection. Furthermore, we are able to use the proposed approach to characterize relationships between pathway variables associated with this event that agree with relationships previously identified by climate scientists.

black-box models↗

DOE OSTI2024

Neural Networks for Prediction of Complex Chemistry in Water Treatment Process Optimization

Water chemistry plays a critical role in the design and operation of water treatment processes. Detailed chemistry modeling tools use a combination of advanced thermodynamic models and extensive databases to predict phase equilibria and reaction phenomena. The complexity and formulation of these models preclude their direct integration in equation-oriented modeling platforms, making it difficult to use their capabilities for rigorous water treatment process optimization. Neural networks (NN) can provide a pathway for integrating the predictive capability of chemistry software into equation-oriented models and enable optimization of complex water treatment processes across a broad range of conditions and process designs. Herein, we assess how NN architecture and training data impact their accuracy and use in equation-oriented water treatment models. We generate training data using PhreeqC software and determine how data generation and sample size impact the accuracy of trained NNs. The effect of NN architecture on optimization is evaluated by optimizing hypothetical black-box desalination processes using a range of feed compositions from USGS brackish water data set, tracking the number of successful optimizations, and testing the impact of initial guess on the final solution. Our results clearly demonstrate that data generation and architecture impact NN accuracy and viability for use in equation-oriented optimization problems.

Dudchenko, Alexander V↗

NASA NTRS2002

The retinoic acid signaling pathway regulates anterior/posterior patterning in the nerve cord and pharynx of amphioxus, a chordate lacking neural crest

Amphioxus, the closest living invertebrate relative of the vertebrates, has a notochord, segmental axial musculature, pharyngeal gill slits and dorsal hollow nerve cord, but lacks neural crest. In amphioxus, as in vertebrates, exogenous retinoic acid (RA) posteriorizes the embryo. The mouth and gill slits never form, AmphiPax1, which is normally downregulated where gill slits form, remains upregulated and AmphiHox1 expression shifts anteriorly in the nerve cord. To dissect the role of RA signaling in patterning chordate embryos, we have cloned the single retinoic acid receptor (AmphiRAR), retinoid X receptor (AmphiRXR) and an orphan receptor (AmphiTR2/4) from amphioxus. AmphiTR2/4 inhibits AmphiRAR-AmphiRXR-mediated transactivation in the presence of RA by competing for DR5 or IR7 retinoic acid response elements (RAREs). The 5' untranslated region of AmphiTR2/4 contains an IR7 element, suggesting possible auto- and RA-regulation. The patterns of AmphiTR2/4 and AmphiRAR expression during embryogenesis are largely complementary: AmphiTR2/4 is strongly expressed in the cerebral vesicle (homologous to the diencephalon plus anterior midbrain), while AmphiRAR expression is high in the equivalent of the hindbrain and spinal cord. Similarly, while AmphiTR2/4 is expressed most strongly in the anterior and posterior thirds of the endoderm, the highest AmphiRAR expression is in the middle third. Expression of AmphiRAR is upregulated by exogenous RA and completely downregulated by the RA antagonist BMS009. Moreover, BMS009 expands the pharynx posteriorly; the first three gill slit primordia are elongated and shifted posteriorly, but do not penetrate, and additional, non-penetrating gill slit primordia are induced. Thus, in an organism without neural crest, initiation and penetration of gill slits appear to be separate events mediated by distinct levels of RA signaling in the pharyngeal endoderm. Although these compounds have little effect on levels of AmphiTR2/4 expression, RA shifts pharyngeal expression of AmphiTR2/4 anteriorly, while BMS009 extends it posteriorly. Collectively, our results suggest a model for anteroposterior patterning of the amphioxus nerve cord and pharynx, which is probably applicable to vertebrates as well, in which a low anterior level of AmphiRAR (caused, at least in part, by competitive inhibition by AmphiTR2/4) is necessary for patterning the forebrain and formation of gill slits, the posterior extent of both being set by a sharp increase in the level of AmphiRAR. Supplemental data available on-line.

NASA Discipline Evolutionary Biology↗

NASA NTRS1994

A Feasibility Study for Perioperative Ventricular Tachycardia Prognosis and Detection and Noise Detection Using a Neural Network and Predictive Linear Operators

To locate the accessory pathway(s) in preexicitation syndromes, epicardial and endocardial ventricular mapping is performed during anterograde ventricular activation via accessory pathway(s) from data originally received in signal form. As the number of channels increases, it is pertinent that more automated detection of coherent/incoherent signals is achieved as well as the prediction and prognosis of ventricular tachywardia (VT). Today's computers and computer program algorithms are not good in simple perceptual tasks such as recognizing a pattern or identifying a sound. This discrepancy, among other things, has been a major motivating factor in developing brain-based, massively parallel computing architectures. Neural net paradigms have proven to be effective at pattern recognition tasks. In signal processing, the picking of coherent/incoherent signals represents a pattern recognition task for computer systems. The picking of signals representing the onset ot VT also represents such a computer task. We attacked this problem by defining four signal attributes for each potential first maximal arrival peak and one signal attribute over the entire signal as input to a back propagation neural network. One attribute was the predicted amplitude value after the maximum amplitude over a data window. Then, by using a set of known (user selected) coherent/incoherent signals, and signals representing the onset of VT, we trained the back propagation network to recognize coherent/incoherent signals, and signals indicating the onset of VT. Since our output scheme involves a true or false decision, and since the output unit computes values between 0 and 1, we used a Fuzzy Arithmetic approach to classify data as coherent/incoherent signals. Furthermore, a Mean-Square Error Analysis was used to determine system stability. The neural net based picking coherent/incoherent signal system achieved high accuracy on picking coherent/incoherent signals on different patients. The system also achieved a high accuracy of picking signals which represent the onset of VT, that is, VT immediately followed these signals. A special binary representation of the input and output data allowed the neural network to train very rapidly as compared to another standard decimal or normalized representations of the data.

Moebes, T. A.↗

NASA NTRSDate not supplied

Oculometric Analysis of Saccadic Compensation for Visual Motion Processing Impairment due to Alcohol and Sleep Disruption

The Visuomotor Control Laboratory at Ames Research Center has developed a 5-minute ocular tracking test that computes 21 largely independent metrics of visuomotor performance, reflecting neural signal processing along a number of distinct pathways through cortex, brainstem, and cerebellum. Human sensorimotor performance is resilient to the challenges and stressors of many operational environments, in part, because overall performance is achieved through multiple parallel systems. Our multidimensional oculometrics allow us to examine impacts on these sub-components separately. To illustrate this, we contrasted the effects of two mild neural stressors, acute sleep-deprivation and low-dose alcohol. We have previously shown that, in both cases, oculometric analysis is a highly sensitive indicator of impairment. Here we quantified not only the observed impact on the performance of one sub-system, smooth pursuit, which uses high-level cortical processing of visual motion to track a moving object, but also the observed (partial) compensation by an evolutionarily older mid-brain and brainstem subsystem, saccades, which generates jumps in eye position to catch up with the target when smooth pursuit is inadequate. Specifically, we examined the dose-response (effect size vs. dose size) of the ground lost (pursuit deficit) and the ground recouped (saccadic compensation) across three separate studies – acute low-dose alcohol administration (16 subjects), acute sleep loss (12 subjects), and acute sleep loss with caffeine intervention (9 subjects). We computed the dose-response slopes using linear regression. The figure below shows that, in the case of acute sleep deprivation, the resulting slopes for ground lost and ground recouped (mean ± SE across subjects) were significantly different (paired t-test, t(11) = 5.17, p < 0.001), indicating poor saccadic compensation. However, when sleep loss was coupled with caffeine ingestion, ground lost was decreased and ground recouped increased such that the slopes were no longer different (t(8) = -0.05, p = 0.965). With alcohol, the two slopes were large albeit not significantly different (t(15) = 0.96, p = 0.351), indicating significant pursuit impairment but effective saccadic compensation. Our findings show that sleep deprivation and alcohol affect oculomotor performance differently. Low-dose alcohol effects appear predominantly cortical, with effective brainstem compensation. Sleep loss and circadian disruption however appears to affect both cortical and brainstem pathways with caffeine providing an effective countermeasure to both effects. Beyond the mere detection of impairment, our oculometric assessment allows us to characterize the nature of the deficit, to provide insight into the neural substrate, and to assess the effectiveness of countermeasures.

pursuit↗

DOE OSTI2025

Wide‐Field Bond Quality Evaluation Using Frequency Domain Thermoreflectance with Deep Neural Network Feature Reconstruction

Heterogeneous integration of microelectronic components provides a pathway to improve circuit/component performance; however, this comes with assembly challenges, in particular due to complex interfaces via subsurface bump bonds. The ability of these bonds to transmit electrical signals and conduct heat to the carrier substrate limits component performance. In this work, hyperspectral frequency‐domain thermoreflectance (FDTR) imaging is demonstrated as a robust technique for evaluating the quality of subsurface indium bump bonds in a surrogate microelectronic sample. By performing microscale FDTR imaging with coarse motion image stitching, thermal phase maps that cover a 4 mm by 4 mm field‐of‐view with subsurface feature sensitivity at depths greater than 50 µm are obtained. The resulting FDTR hyperspectral data contains more than three million pixels and reveal the quality of subsurface microbump arrays. Wide‐field analysis of bonded versus gap regions is enabled by deep neural network feature reconstruction, that after training, rapidly provides an interpretable representation of bond quality. Utility of noisy higher frequency FDTR phase maps, i.e., near the computationally predicted sensing depth limit, results in an average prediction error of 11%. Taken together, FDTR with neural network‐based analysis demonstrates subsurface bond monitoring at length scales relevant for heterogeneously integrated microelectronics.

FDTR↗

DOE OSTI2024

Selective consolidation of learning and memory via recall-gated plasticity

In a variety of species and behavioral contexts, learning and memory formation recruits two neural systems, with initial plasticity in one system being consolidated into the other over time. Moreover, consolidation is known to be selective; that is, some experiences are more likely to be consolidated into long-term memory than others. Here, we propose and analyze a model that captures common computational principles underlying such phenomena. The key component of this model is a mechanism by which a long-term learning and memory system prioritizes the storage of synaptic changes that are consistent with prior updates to the short-term system. This mechanism, which we refer to as recall-gated consolidation, has the effect of shielding long-term memory from spurious synaptic changes, enabling it to focus on reliable signals in the environment. We describe neural circuit implementations of this model for different types of learning problems, including supervised learning, reinforcement learning, and autoassociative memory storage. These implementations involve synaptic plasticity rules modulated by factors such as prediction accuracy, decision confidence, or familiarity. We then develop an analytical theory of the learning and memory performance of the model, in comparison to alternatives relying only on synapse-local consolidation mechanisms. We find that recall-gated consolidation provides significant advantages, substantially amplifying the signal-to-noise ratio with which memories can be stored in noisy environments. We show that recall-gated consolidation gives rise to a number of phenomena that are present in behavioral learning paradigms, including spaced learning effects, task-dependent rates of consolidation, and differing neural representations in short- and long-term pathways.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2024

Selective consolidation of learning and memory via recall-gated plasticity

Lindsey, Jack W. (ORCID:0000000309307327)↗

NASA NTRS1998

Neuronal plasticity: adaptation and readaptation to the environment of space

While there have been few documented permanent neurological changes resulting from space travel, there is a growing literature which suggests that neural plasticity sometimes occurs within peripheral and central vestibular pathways during and following spaceflight. This plasticity probably has adaptive value within the context of the space environment, but it can be maladaptive upon return to the terrestrial environment. Fortunately, the maladaptive responses resulting from neuronal plasticity diminish following return to earth. However, the literature suggests that the longer the space travel, the more difficult the readaptation. With the possibility of extended space voyages and extended stays on board the international space station, it seems worthwhile to review examples of plastic vestibular responses and changes in the underlying neural substrates. Studies and facilities needed for space station investigation of plastic changes in the neural substrates are suggested. Copyright 1998 Elsevier Science B.V.

unmanned↗

NASA NTRS2006

Structure and Function of the Snail Statocyst System after a 16-Day Flight on Foton-M-2

In terrestrial gastropod snail Helix lucorum L. we studied the changes after a 16-day exposure to microgravity in: behavior, neural responses to adequate motion stimulation, intersensory interactions between the photosensory pathways and the statocyst receptors, and in expression of the HPeP gene in the statocyst receptors. In behavioral experiments it was found that the latency of body position change to sudden orientation change (flip from horizontal to downwards position) was significantly reduced in the postflight snails. Extracellularly recorded neural responses of the statocyst nerve to adequate motion stimulation in the postflight snails were independent of the motion direction while in the control animals an orientation selectivity was observed. Significant differences in the HPeP gene mRNA expression pattern in the statocyst receptor neurons were observed in postflight (30h) and control snails. Obtained results confirm the possibility to elucidate the influence of microgravity exposure on mechanisms and function of gravireceptors using this simple model animal.

Balaban, P. M.↗

NASA NTRS2002

Motor scaling by viewing distance of early visual motion signals during smooth pursuit

The geometry of gaze stabilization during head translation requires eye movements to scale proportionally to the inverse of target distance. Such a scaling has indeed been demonstrated to exist for the translational vestibuloocular reflex (TVOR), as well as optic flow-selective translational visuomotor reflexes (e.g., ocular following, OFR). The similarities in this scaling by a neural estimate of target distance for both the TVOR and the OFR have been interpreted to suggest that the two reflexes share common premotor processing. Because the neural substrates of OFR are partly shared by those for the generation of pursuit eye movements, we wanted to know if the site of gain modulation for TVOR and OFR is also part of a major pathway for pursuit. Thus, in the present studies, we investigated in rhesus monkeys whether initial eye velocity and acceleration during the open-loop portion of step ramp pursuit scales with target distance. Specifically, with visual motion identical on the retina during tracking at different distances (12, 24, and 60 cm), we compared the first 80 ms of horizontal pursuit. We report that initial eye velocity and acceleration exhibits either no or a very small dependence on vergence angle that is at least an order of magnitude less than the corresponding dependence of the TVOR and OFR. The results suggest that the neural substrates for motor scaling by target distance remain largely distinct from the main pathway for pursuit.

Non-NASA Center↗

DOE OSTI2026

Detecting and Characterizing Fracture Zones Using a Convolutional Neural Network

This project directly supports the Geothermal Technologies Office (GTO) objectives outlined in the Multi-Year Program Plan (MYPP) by advancing two key research areas: “Exploration and Characterization” and “Data, Modeling, and Analysis.” This project has successfully demonstrated a pre-drilling ability to image and characterize the distribution and connectivity of subsurface faults and fractures, key parameters for identifying permeable pathways that enable geothermal fluids to circulate and produce energy. Specifically, we developed and implemented innovative machine learning methodologies to enhance geothermal exploration. Large-scale faults were detected using a Convolutional Neural Network (CNN), while small-scale fractures were characterized using a novel Double-Beam Neural Network (DBNN). These tools have proven both technically effective and cost-efficient by reducing reliance on expensive exploratory drilling. Through collaboration with our geothermal industry partner, this research has significantly advanced techniques for identifying hidden geothermal systems and extending the productive lifespan of existing geothermal fields. We applied our methods to two geothermal fields—Soda Lake (Nevada) and Lightning Dock (New Mexico)—to identify shallow steam-charged fracture zones and characterize deep faults at depths of 1.5-2 km. The steam zone identified at the Soda Lake geothermal field showed excellent agreement with prior drilling data, validating the effectiveness of our approaches. In addition, the analysis revealed three new prospective drilling targets for further development and verification. The outcomes of this project improve our scientific understanding of geothermal reservoir behavior, enhance exploration efficiency, extend the economic life of existing geothermal plants. Ultimately, these advancements contribute to GTO’s goal of achieving more sustainable, affordable, and data-driven geothermal energy development across the United States.

15 GEOTHERMAL ENERGY↗

DOE OSTI2024

Foundation Models for the Electric Power Grid

Foundation models (FMs) currently dominate news headlines. They employ advanced deep learning architectures to extract structural information autonomously from vast datasets through self-supervision. The resulting rich representations of complex systems and dynamics can be applied to many downstream applications. Therefore, advances in FMs can find uses in electric power grids, challenged by the energy transition and climate change. This paper calls for the development of FMs for electric grids. We highlight their strengths and weaknesses amidst the challenges of a changing grid. It is argued that FMs learning from diverse grid data and topologies, which we call grid foundation models (GridFMs), could unlock transformative capabilities, pioneering a new approach to leveraging AI to redefine how we manage complexity and uncertainty in the electric grid. Finally, we discuss a practical implementation pathway and road map of a GridFM-v0, a first GridFM for power flow applications based on graph neural networks, and explore how various downstream use cases will benefit from this model and future GridFMs.

AI-based power flow simulation↗

DOE OSTI2025

CHEMREASONER: Heuristic Search over a Large Language Model’s Knowledge Space using Quantum-Chemical Feedback

The discovery of new catalysts is essential for the design of new and more efficient chemical processes in order to transition to a sustainable future. We introduce an AI-guided computational screening framework unifying linguistic reasoning with quantum-chemistry based feedback from 3D atomistic representations. Our approach formulates catalyst discovery as an uncertain environment where an agent actively searches for highly effective catalysts via the iterative combination of large language model (LLM)-derived hypotheses and atomistic graph neural network (GNN)-derived feedback. Identified catalysts in intermediate search steps undergo structural evaluation based on spatial orientation, reaction pathways, and stability. Scoring functions based on adsorption energies and barriers steer the exploration in the LLM's knowledge space toward energetically favorable, high-efficiency catalysts. We introduce planning methods that automatically guide the exploration without human input, providing competitive performance against expert-enumerated chemical descriptor-based implementations. By integrating language-guided reasoning with computational chemistry feedback, our work pioneers AI-accelerated, trustworthy catalyst discovery.

artificial intelligence↗