SEARCH · Search NASA

Results for “Deep potential”

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 343 records · Page 19

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Million-scale data integrated deep neural network for phonon properties of heuslers spanning the periodic table

Existing machine learning potentials for predicting phonon properties of crystals are typically limited on a material-to-material basis, primarily due to the exponential scaling of model complexity with the number of atomic species. We address this bottleneck with the developed Elemental Spatial Density Neural Network Force Field, namely Elemental-SDNNFF. The effectiveness and precision of our Elemental-SDNNFF approach are demonstrated on 11,866 full, half, and quaternary Heusler structures spanning 55 elements in the periodic table by prediction of complete phonon properties. Self-improvement schemes including active learning and data augmentation techniques provide an abundant 9.4 million atomic data for training. Deep insight into predicted ultralow lattice thermal conductivity (<1 Wm –1 K –1 ) of 774 Heusler structures is gained by p–d orbital hybridization analysis. Additionally, a class of two-band charge-2 Weyl points, referred to as “double Weyl points”, are found in 68% and 87% of 1662 half and 1550 quaternary Heuslers, respectively.

36 MATERIALS SCIENCE↗

DOE OSTI2023

The potential of a TeV-scale muon-ion collider

We propose the development of a novel muon-proton and muon-nucleus collider facility at the TeV scale that is capable of performing precision deep inelastic scattering measurements in new regimes and providing a rich program in nuclear and particle physics. Such a facility could seed, or leverage, the development of a muon-antimuon collider and make use of the existing hadron accelerator infrastructure when sited at a facility such as Brookhaven National Laboratory, Fermilab, or CERN. Here, we discuss the possible energy and luminosity design parameters for several collider configurations, and illustrate the science potential with several studies on deep inelastic scattering kinematics, Higgs and vector boson production, top quark production, and beyond Standard Model leptoquark production. Detector design considerations and a possible road map toward development are also given.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2025

Toward Complete Merger Identification at Cosmic Noon with Deep Learning

As we enter the era of large imaging surveys such as Roman, Rubin, and Euclid, a deeper understanding of potential biases and selection effects in optical astronomical catalogs created with the use of ML-based methods is paramount. This work focuses on a deeper understanding of the performance and limitations of deep learning-based classifiers as tools for galaxy merger identification. We train a ConvNeXT-Pico model on mock HST CANDELS images from the IllustrisTNG50 simulation. Our focus is on a more challenging classification of galaxy mergers and non-mergers at higher redshifts 1 < z < 1.5, including minor mergers and lower mass galaxies down to the stellar mass of 108M⊙. We demonstrate, for the first time, that a deep learning model, such as the one developed in this work, can successfully identify even minor and low mass mergers even at these redshifts. Our model achieves overall accuracy, purity, and completeness of over 73%. We show that some galaxy mergers can only be identified from certain observation angles, leading to a potential upper limit in overall accuracy. Using Grad-CAMs and UMAPs, we more deeply examine the performance and observe a visible gradient in the latent space with stellar mass and specific star formation rate, but no visible gradient with merger mass ratio or merger stage.

Schechter, Aimee L. [U. Colorado, Boulder]↗

DOE OSTI2025

Experimental investigation of flow distribution in enhanced geothermal systems with deep eutectic solvent

Geothermal energy has been recognized as a valuable alternative to fossil fuels and nuclear power, as it is renewable and reliable. Enhanced Geothermal Systems (EGSs) have the potential to expand geothermal energy production by enabling access to previously untapped geothermal resources. Geothermal short-circuiting poses a significant challenge to EGS development, leading to reduced heat extraction. Deep Eutectic Solvent (DES) exhibits favorable thermal and rheological properties, making it a candidate for geothermal applications. Here, this paper examines Choline Chloride-Based Deep Eutectic Solvent (DES) as a working fluid in geothermal applications and its potential to mitigate geothermal short-circuiting. Hydraulic experiments using a dual fracture flow loop were conducted at high temperatures. The results showed that DES exhibited higher differential pressure behavior compared to water. Flow distribution results revealed that DES enhances flow allocation within the small fracture, particularly when a temperature difference exists between fractures. Specifically, DES increased flow distribution by an average of 11% when the temperature difference was 85°C, and by 13% when the difference was 45°C, relative to water. These findings suggest that DES responds to thermal fracture differences, making it a potential remedy to address geothermal short-circuiting.

15 GEOTHERMAL ENERGY↗

DOE OSTI2022

Making Invisible Visible: Data-Driven Seismic Inversion With Spatio-Temporally Constrained Data Augmentation

Deep learning and data-driven approaches have shown great potential in scientific domains. The promise of data-driven techniques relies on the availability of a large volume of high-quality training datasets. Due to the high cost of obtaining data through expensive physical experiments, instruments, and simulations, data augmentation techniques for scientific applications have emerged as a new direction for obtaining scientific data recently. However, existing data augmentation techniques originating from computer vision yield physically unacceptable data samples that are not helpful for the domain problems that we are interested in. In this article, we develop new data augmentation techniques based on convolutional neural networks. Specifically, our generative models leverage different physics knowledge (such as governing equations, observable perception, and physics phenomena) to improve the quality of the synthetic data. To validate the effectiveness of our data augmentation techniques, we apply them to solve a subsurface seismic full-waveform inversion using simulated CO 2 leakage data. Our interest is to invert for subsurface velocity models associated with very small CO 2 leakage. We validate the performance of our methods using comprehensive numerical tests. Here via comparison and analysis, we show that data-driven seismic imaging can be significantly enhanced by using our data augmentation techniques. Particularly, the imaging quality has been improved by 15% in test scenarios of general-sized leakage and 17% in small-sized leakage when using an augmented training set obtained with our techniques.

58 GEOSCIENCES↗

DOE OSTI2023

Microbial life in 25-m-deep boreholes in ancient permafrost illuminated by metagenomics

This study describes the composition and potential metabolic adaptation of microbial communities in northeastern Siberia, a repository of the oldest permafrost in the Northern Hemisphere. Samples of contrasting depth (1.75 to 25.1 m below surface), age (from ~ 10 kyr to 1.1 Myr) and salinity (from low 0.1–0.2 ppt and brackish 0.3–1.3 ppt to saline 6.1 ppt) were collected from freshwater permafrost (FP) of borehole AL1_15 on the Alazeya River, and coastal brackish permafrost (BP) overlying marine permafrost (MP) of borehole CH1_17 on the East Siberian Sea coast. To avoid the limited view provided with culturing work, we used 16S rRNA gene sequencing to show that the biodiversity decreased dramatically with permafrost age. Nonmetric multidimensional scaling (NMDS) analysis placed the samples into three groups: FP and BP together (10–100 kyr old), MP (105–120 kyr old), and FP (> 900 kyr old). Younger FP/BP deposits were distinguished by the presence of Acidobacteriota, Bacteroidota, Chloroflexota_A, and Gemmatimonadota, older FP deposits had a higher proportion of Gammaproteobacteria, and older MP deposits had much more uncultured groups within Asgardarchaeota, Crenarchaeota, Chloroflexota, Patescibacteria, and unassigned archaea. The 60 recovered metagenome-assembled genomes and un-binned metagenomic assemblies suggested that despite the large taxonomic differences between samples, they all had a wide range of taxa capable of fermentation coupled to nitrate utilization, with the exception of sulfur reduction present only in old MP deposits.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2024

On the Feasibility of Deep Geothermal Wells Using Numerical Reservoir Simulation

This study examines the geothermal energy extraction potential from the basement rock within the Denver–Julesburg Basin, focusing on the flow performance and heat extraction efficiency of different geothermal well configurations. It specifically compares U-shaped, V-shaped, inclined V-shaped, and pipe-in-pipe configurations against enhanced geothermal system setups. Through numerical modeling, we evaluated the thermal behavior of these systems under various operational scenarios and fracture conditions. The results suggest that while closed-loop systems offer moderate temperature increases, Enhanced geothermal system configurations show substantial potential for high-temperature extraction. This underscores the importance of evaluating well configurations in complex geological settings. The insights from this study aid in strategic geothermal energy planning and development, marking significant advancements in geothermal technology and setting a foundation for future explorations and optimizations.

15 GEOTHERMAL ENERGY↗

DOE OSTI2025

Deep learning model for fast, science-based forecasting of fluid migration along faults in geologic carbon storage scenarios

Effective long-term geologic storage depends on robust site selection and credible, science-based forecasting of subsurface behavior to ensure storage integrity. For this work, we develop a deep learning–based reduced-order model (ROM) to quantify potential carbon dioxide (CO₂) and brine migration through geological faults. The ROM combines a Transformer model for binary classification and a Stacked Ensemble for regression, trained on a comprehensive dataset generated from 1400 physics-based reservoir simulations. Key geologic and operational parameters—including fault geometry, reservoir structure, and injection conditions—were systematically varied to capture a wide range of fluid migration scenarios. The ROM accurately predicts the onset of migration, cumulative migration volumes of both CO₂ and brine, and associated migration rates, as compared to an independent set of validation simulations, while significantly reducing computational cost compared to traditional simulation methods. Model performance was evaluated across diverse fault configurations, revealing that shallow reservoir geometry and fault angle are among the most influential factors governing migration behavior. Sensitivity analysis using SHapley Additive exPlanations (SHAP) provided interpretability, revealing distinct patterns in how geological and operational features drive transient versus cumulative migration outcomes. The ROM’s ability to rapidly simulate fault migration scenarios enables efficient sensitivity analyses, scenario evaluations, and decision support for site selection and monitoring design. This approach enhances the safety, scalability, and long-term operational performance of geologic carbon storage (GCS) systems by providing a robust, interpretable tool for predicting subsurface fluid migration and assessing fault-related migration potential.

42 ENGINEERING↗

DOE OSTI2021

Co-expression networks in Chlamydomonas reveal significant rhythmicity in batch cultures and empower gene function discovery

The unicellular green alga Chlamydomonas reinhardtii is a choice reference system for the study of photosynthesis and chloroplast metabolism, cilium assembly and function, lipid and starch metabolism, and metal homeostasis. Despite decades of research, the functions of thousands of genes remain largely unknown, and new approaches are needed to categorically assign genes to cellular pathways. Growing collections of transcriptome and proteome data now allow a systematic approach based on integrative co-expression analysis. We used a dataset comprising 518 deep transcriptome samples derived from 58 independent experiments to identify potential co-expression relationships between genes. We visualized co-expression potential with the R package corrplot, to easily assess co-expression and anti-correlation between genes. We extracted several hundred high-confidence genes at the intersection of multiple curated lists involved in cilia, cell division, and photosynthesis, illustrating the power of our method. Surprisingly, Chlamydomonas experiments retained a significant rhythmic component across the transcriptome, suggesting an underappreciated variable during sample collection, even in samples collected in constant light. Our results therefore document substantial residual synchronization in batch cultures, contrary to assumptions of asynchrony. We provide step-by-step protocols for the analysis of co-expression across transcriptome data sets from Chlamydomonas and other species to help foster gene function discovery.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2025

Measurement of magnetic dipole strength in 40 Ar between 9.3 and 10.3 MeV excitation energy and impact on neutral current neutrino interaction

The magnetic dipole strength 𝐵⁡(𝑀⁢1 ↑) of 40 Ar was measured in the excitation energy range between 9.3 and 10.3 MeV. The nuclear resonance fluorescence technique was used with monoenergetic, linearly polarized incident photon beams. The new data increase the known 𝐵⁡(𝑀⁢1 ↑) strength in this important excitation energy range by a factor of 2, resulting in an equivalent increase of the calculated neutral current neutrino cross section of 40 Ar . This observation is of special interest for liquid-argon (LAr)-based neutrino detectors. 𝐵⁡(𝑀⁢1 ↑) strength has not been found between 10.3 and 11.7 MeV excitation energy. In conclusion, with the new data, a more reliable detection capability of the large Deep Underground Neutrino Experiment LAr detectors can be derived for potential core-collapse supernova neutrinos via the neutral current interaction.

Energy levels↗

DOE OSTI2020

Carbon-Carbon Aeroshell Chemical Reaction Model in an Open RPS in Martian Environment

A new RPS is being designed to be open to the vacuum of space to improve its thermal efficiency during deep space missions. This design, however, could lead to some potential complications if it is ever used in the low-pressure gas environment of mission destinations like Mars. This report documents the development of a chemical kinetics reaction model of the FWPF carbon-carbon aeroshell in the Martian gaseous environment. This model uses an “infinite sink” assumption where the RPS is open to the Martian environment with open exchange of flowing gases, so the gas chemistry over the aeroshell does not change with the reaction and represents the worst case scenario of the chemical reaction of the FWPF in the Martian environment. The model combines the kinetics equation of both oxidation reactions that would occur on Mars: the Langmuir-Hinshelwood kinetics of the Boudouard reaction of CO 2 with the chemical kinetics of O 2 oxidation of the carbon-carbon. The FWPF chemical reaction model was benchmarked using the Opila’s experimental chemical reaction test data (1) in a simulated Martian environment to increase the validity of the model’s predicted mass loss rate. The output of the benchmarked chemical reaction model showed a trade-off between the maximum temperature of the aeroshell to have less than 10% reaction at the mission life. For example, an RPS with a five-year intended mission life on Mars has a maximum aeroshell temperature of 835 K / 562 °C, increasing the mission life to ten years reduces the maximum operating temperature to 810 K / 527 °C. If the aeroshell operating temperature in the new RPS on the surface of Mars is 960 Kelvin, the model calculates that the life is limited to 0.22 years / 81 days. Increasing the aeroshell temperature to 1343 Kelvin, further reduces the life to 0.375 days / 9 hrs. This illustrates the thermally activated nature of the chemical kinetics model in limiting the life of the aeroshell.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2022

Albedo susceptibility of northeastern Pacific stratocumulus: the role of covarying meteorological conditions

Abstract. Quantification of the radiative adjustment of marine low clouds to aerosol perturbations, regionally and globally, remains the largest source of uncertainty in assessing current and future climate. One of the important steps towards quantifying the role of aerosol in modifying cloud radiative properties is to quantify the susceptibility of cloud albedo and liquid water path (LWP) to perturbations in cloud droplet number concentration (Nd). We use 10 years of spaceborne observations from the polar-orbiting Aqua satellite to quantify the albedo susceptibility of marine low clouds to Nd perturbations over the northeast (NE) Pacific stratocumulus (Sc) region. Mutual information analysis reveals a dominating control of cloud state (e.g., LWP and Nd) on low-cloud albedo susceptibility, relative to the meteorological states that drive these cloud states. Through a LWP–Nd space decomposition of albedo susceptibilities, we show clear separation among susceptibility regimes (brightening or darkening), consistent with previously established mechanisms through which aerosol modulates cloud properties. These regimes include (i) thin non-precipitating clouds (LWP < 55 g m−2) that exhibit brightening (occurring 37 % of the time), corresponding to the Twomey effect; (ii) thicker non-precipitating clouds, corresponding to entrainment-driven negative LWP adjustments that manifest as a darkening regime (36 % of the time); and (iii) another brightening regime (22 % of the time) consisting of mostly precipitating clouds, corresponding to precipitation-suppression LWP positive adjustments. Overall, we find an annual-mean regional low-cloud brightening potential of 20.8±2.68 W m−2 ln(Nd)−1, despite an overall negative LWP adjustment for non-precipitating marine stratocumulus, owing to the high occurrence of the Twomey–brightening regime. Over the NE Pacific, clear seasonal covariabilities among meteorological factors related to the large-scale circulation are found to play an important role in grouping conditions favorable for each susceptibility regime. When considering the covarying meteorological conditions, our results indicate that for the northeastern Pacific stratocumulus, clouds that exhibit the strongest brightening potential occur most frequently within shallow marine boundary layers over a cool ocean surface with a stable atmosphere and a dry free troposphere above. Clouds that exhibit a darkening potential associated with negative LWP adjustments occur most frequently within deep marine boundary layers in which the atmospheric instability and the ocean surface are not strong and warm enough to produce frequent precipitation. Cloud brightening associated with warm-rain suppression is found to preferably occur either under unstable atmospheric conditions or humid free-tropospheric conditions that co-occur with a warm ocean surface.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

Carbon-Carbon Aeroshell Chemical Reaction Model in an Open RPS in Martian Environment

A new RPS is being designed to be open to the vacuum of space to improve its thermal efficiency during deep space missions. This design, however, could lead to some potential complications if it is ever used in the low-pressure gas environment of mission destinations like Mars. This report documents the development of a chemical kinetics reaction model of the FWPF carbon-carbon aeroshell in the Martian gaseous environment. This model uses an “infinite sink” assumption where the RPS is open to the Martian environment with open exchange of flowing gases, so the gas chemistry over the aeroshell does not change with the reaction and represents the worst case scenario of the chemical reaction of the FWPF in the Martian environment. The model combines the kinetics equation of both oxidation reactions that would occur on Mars: the Langmuir-Hinshelwood kinetics of the Boudouard reaction of CO2 with the chemical kinetics of O2 oxidation of the carbon-carbon. The FWPF chemical reaction model was benchmarked using the Opila’s experimental chemical reaction test data (1) in a simulated Martian environment to increase the validity of the model’s predicted mass loss rate. The output of the benchmarked chemical reaction model showed a trade-off between the maximum temperature of the aeroshell to have less than 10% reaction at the mission life. For example, an RPS with a five-year intended mission life on Mars has a maximum aeroshell temperature of 835 K / 562 °C, increasing the mission life to ten years reduces the maximum operating temperature to 810 K / 527 °C. If the aeroshell operating temperature in the new RPS on the surface of Mars is 960 Kelvin, the model calculates that the life is limited to 0.22 years / 81 days. Increasing the aeroshell temperature to 1343 Kelvin, further reduces the life to 0.375 days / 9 hrs. This illustrates the thermally activated nature of the chemical kinetics model in limiting the life of the aeroshell.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Radionuclide Transport Simulations Supporting Proposed Borehole Waste Disposal in Israel

A scientific collaboration between the U.S. and Israel is underway to assess the suitability of a potential site for subsurface radioactive waste disposal in the Negev Desert, Israel. The Negev Desert has several favorable attributes for geologic disposal, including an arid climate, a deep vadose zone, interlayered low-permeability lithologies, and carbonate rocks with high uranium-sorption potential. These features may provide a robust natural barrier to radionuclide migration. Geologic and laboratory characterization data from the Negev Desert are incorporated into multiphase flow and transport models, solved using PFLOTRAN, to aid in site characterization and risk analysis that will support decision-making for waste disposal in an intermediate-depth borehole design. The lithology with the greatest uranium sorption potential at the site is phosphorite. We use modeling to evaluate the ability of this layer to impact uranium transport around a proposed disposal borehole. The current objective of the simulations is focused on characterizing hypothetical leakage from waste canisters and subsequent uranium migration under three infiltration scenarios. Here, we describe a hydrogeologic model based on data from a local exploratory borehole and present results for uranium flow and transport simulations under varying infiltration scenarios. We find that under the current climate conditions, it is likely that uranium will remain in the near-field of the borehole for thousands of years. However, under a hypothesized extreme climate scenario representing an increase in infiltration by a factor of 300x above present-day values, uranium may break through the phosphorite layer and exit the base of the model domain (~200 m above the water table) within 1000 years. Simulation results have direct implications for the planning of nuclear waste disposal in the Negev Desert, and specifically in intermediate-depth boreholes.

58 GEOSCIENCES↗

DOE OSTI2020

Active sulfur cycling in the terrestrial deep subsurface

The deep terrestrial subsurface remains an environment where there is limited understanding of the extant microbial metabolisms. At Olkiluoto, Finland, a deep geological repository is under construction for the final storage of spent nuclear fuel. It is therefore critical to evaluate the potential impact microbial metabolism, including sulfide generation, could have upon the safety of the repository. As such, we investigated a deep groundwater where sulfate is present, but groundwater geochemistry suggests limited microbial sulfate-reducing activity. Examination of the microbial community at the genome-level revealed microorganisms with the metabolic capacity for both oxidative and reductive sulfur transformations. Deltaproteobacteria are shown to have the genetic capacity for sulfate reduction and possibly sulfur disproportionation, while Rhizobiaceae, Rhodocyclaceae, Sideroxydans, and Sulfurimonas oxidize reduced sulfur compounds. Further examination of the proteome confirmed an active sulfur cycle, serving for microbial energy generation and growth. Our results reveal that this sulfide-poor groundwater harbors an active microbial community of sulfate-reducing and sulfide-oxidizing bacteria, together mediating a sulfur cycle that remained undetected by geochemical monitoring alone. The ability of sulfide-oxidizing bacteria to limit the accumulation of sulfide was further demonstrated in groundwater incubations and highlights a potential sink for sulfide that could be beneficial for geological repository safety.

58 GEOSCIENCES↗

DOE OSTI2025

Genetically engineered poplar wood effectively enhances the efficiency of deep eutectic solvent-mediated one-pot processing

Although lignocellulosic biomass is a renewable resource with the potential to replace fossil-derived fuels and chemicals, its recalcitrance, largely due to lignin, limits its utilization. Recent advancements in genetic engineering have produced transgenic trees with reduced lignin content and/or modified lignin structure without compromising growth traits. Here, three engineered poplar varieties are evaluated as feedstocks using a biocompatible one-pot deep eutectic solvent-mediated process that integrates biomass fractionation and enzymatic saccharification within a single reactor, eliminating water washing and reconditioning. All transgenic poplars exhibit higher fermentable sugar yields than wild-type (WT) trees. Notably, QsuB poplar, incorporating 3,4-dihydroxybenzoate in lignin, achieves the highest glucose conversion yield of 91.3% (vs. 73.0% from WT). AT5 and MdCHS3 poplars, incorporating ferulate esters and naringenin, also demonstrate improved glucose yields (86.7 and 84.7%, respectively), confirming reduced biomass recalcitrance. Additionally, residual lignins are valorized via hydrogenolysis into phenolic compounds, with comparable alkylphenol production across all lines. These findings demonstrate that the transgenic poplar lines not only serve as superior feedstocks for sugar conversion but also provide a rich resource for phenolic compound production, enhancing the operational and economic viability of integrated biorefinery processes.

biomasses↗

DOE OSTI2023

Automated classification of big X-ray diffraction data using deep learning models

Abstract In current in situ X-ray diffraction (XRD) techniques, data generation surpasses human analytical capabilities, potentially leading to the loss of insights. Automated techniques require human intervention, and lack the performance and adaptability required for material exploration. Given the critical need for high-throughput automated XRD pattern analysis, we present a generalized deep learning model to classify a diverse set of materials’ crystal systems and space groups. In our approach, we generate training data with a holistic representation of patterns that emerge from varying experimental conditions and crystal properties. We also employ an expedited learning technique to refine our model’s expertise to experimental conditions. In addition, we optimize model architecture to elicit classification based on Bragg’s Law and use evaluation data to interpret our model’s decision-making. We evaluate our models using experimental data, materials unseen in training, and altered cubic crystals, where we observe state-of-the-art performance and even greater advances in space group classification.

Chemistry↗

DOE OSTI2024

A Deep Multimodal Representation Learning Framework for Accurate Molecular Properties Prediction

Drug discovery is a complex and challenging process, requiring the optimization of candidate compounds to identify those with the potential to become safe and effective drugs. Predicting molecular properties is an indispensable step in the drug discovery pipeline. Traditionally, this process is costly and time-intensive, involving multiple rounds of experiments and clinical trials, rendering it impractical for every candidate compound. Deep learning techniques have emerged as a promising approach to drug discovery to reduce the cost and time required to identify novel drugs. However, prevalent research in deep learning models focused on predicting molecular properties has primarily fixated on single-modal models, which utilize a single modality of data, neglecting the potential benefits of combining different data modalities. To overcome this limitation, we introduce MRL-Mol: a deep \textbf{M}ultimodal \textbf{R}epresentation \textbf{L}earning framework for accurate \textbf{Mol}ecular properties prediction. MRL-Mol harnesses three data modalities: sequence, graph, and image, augmenting the depth of comprehension. Leveraging a large-scale unlabeled dataset~($\sim$1M unique molecules), we pretrain MRL-Mol to extract inter- and intra-modal information. Our study demonstrates the superior performance of MRL-Mol in predicting molecular properties across six benchmark datasets, including both classification and regression tasks. Notably, MRL-Mol outperforms other state-of-the-art molecular properties prediction models. These findings suggest that by combining information from multiple data modalities, MRL-Mol can comprehend molecules better than single-modal deep learning models and identify molecular properties with better accuracy.

Yang, Yuxin↗