Prediction Of Physical Properties Research Articles

A New Transfer Free Energy Based Implicit Solvation Model for the Description of Disordered and Folded Proteins.

Most biological events occur on time scales that are difficult to access using conventional all-atom molecular dynamics simulations in explicit solvent. Implicit solvent techniques offer a promising solution to this problem, alleviating the computational cost associated with the simulation of large systems and accelerating the sampling compared to explicit solvent models. The substitution of water molecules by a mean field, however, introduces simplifications that may penalize accuracy and impede the prediction of certain physical properties. We demonstrate that existing implicit solvent models developed using a transfer free energy approach, while satisfactory at reproducing the folding behavior of globular proteins, fare less well in characterizing the conformational properties of intrinsically disordered proteins. We develop a new implicit solvent model that maximizes the degree of accuracy for both disordered and folded proteins. We show, by comparing the simulation outputs to experimental data, that in combination with the a99SB-disp force field, the implicit solvent model can describe both disordered (aβ40, PaaA2, and drkN SH3) and folded ((AAQAA)3, CLN025, Trp-cage, and GTT) peptides. Our implicit solvent model permits a computationally efficient investigation of proteins containing both ordered and disordered regions, as well as the study of the transition between ordered and disordered protein states.

Dielectric constant prediction of pure organic liquids and their mixtures with water based on interpretable machine learning

The thermodynamic properties of mixed-solvent electrolytes are functions of pressure, temperature, and composition (PTC), and are generally considered to be characterized by their dielectric constant. In this work, an interpretable dielectric constant model is proposed based on a machine learning algorithm. The model combines machine learning algorithms, Abraham Solvation Parameters (ASP) and SHapley Additive exPlanations (SHAP) methods to accurately predict the dielectric constants of pure organic liquids and their mixtures with water, the significance of each feature and its impact on the results are explained. The predictions demonstrate extremely low mean square errors, and the effect of each feature on the dielectric constant is clearly characterized. This model provides the ability to accurately predict the dielectric constant for any pure organic liquids and their mixtures with water, and can analyze the sensitivity and influence of each feature to the dielectric constant. Furthermore, this work extends the application of the Abraham solvation parameter to prediction of solution properties. The interpretability of the model will make this a great resource to direct the prediction of physical and chemical properties of materials.

Structural Phase Transitions in Anthracene Crystals.

Anthracene (C14 H10 ) and its derivatives, π-conjugated molecules in acenes, have been widely researched in terms of their reactions, physical properties, and self-assembly (or crystal engineering). These molecules can be functionalized to tune reactivities, optoelectronic properties, and self-assembling abilities. Structural changes in the molecular assemblies, solid states, and crystals have recently been discovered. Therefore, a systematic discussion of anthracene's molecular structure, packing, and optical properties based on its intermolecular structure and phase transitions is important for future chemical and structural design. In the present review, we discuss anthracene's molecular design, dimer packing, and crystal structure, focusing on the structural phase transitions of its crystals. We also provide examples of the phase transitions of anthracene crystals. Changes to edge-to-face of CH-π interaction and face-to-face packing of π-π interaction affect the thermodynamic stabilities of various crystal structures. These structures can inform the prediction of structural and physical properties.

Cover: International Journal of Intelligent Systems, Volume 37 Issue 6 June 2022

Cover Caption: The cover image is based on the Research Article Development of physical property prediction models for polypropylene composites with optimizing random forest hyperparameters by Chonghyo Joo et al., https://doi.org/10.1002/int.22700.

Green lubricants production from Nile tilapia waste and prediction of physical properties through molecular dynamics simulations

AbstractFish farming is a worldwide growing activity and a large amount of residues is produced in this process. The present work describes a cleaner and sustainable way to produce new lubricants from fish waste oil. Oil extracted from the Nile tilapia (Oreochromis niloticus) viscera was utilized as raw material to produce basic oil for lubricants. The products were synthesized by esterification with polyols, trimethylolpropane (TMP) and pentaerythritol (PE), using p‐toluenesulfonic acid (p‐TSA) as catalyst. The synthesized esters were characterized by infrared (IR) and nuclear magnetic resonance (NMR). Computational methods were used to predict the physical characteristics of the material. In addition, the main physicochemical properties were evaluated, as well as the thermal behavior and toxicity of the products against Artemia salina. The synthesized esters showed high viscosity indexes (VI > 150) and viscosities that fit the degree of application ISO‐46 and 150. Molecular dynamics simulations indicated that at room temperature the lubricants Tilapia fatty acid ‐ trimethylolpropane ester (T‐TMPE) and Tilapia fatty acid ‐ pentaerythritol ester (T‐PEE) are in liquid and gel states, respectively, confirming the experimental data. The products did not present toxicity against A. salina. In this research, we reinforce the potential of using tilapia oil from waste to produce green lubricants as a strategy to reduce damage to the environment, as well as the use of computational methods that collaborate to predict physical properties of lubricants.

Retrieving electrical and structural carbonate formation properties from measurements on crushed rock using a multidata inversion approach

SUMMARY Quantification understanding, and prediction of physical rock properties rely so far predominantly on laboratory analyses of cores and plugs. Based on such data, petrophysical models are found that relate both microstructural properties and environmental conditions to geophysically accessible quantities. When considering reactive rock–fluid–gas systems, for example in geothermal energy, enhanced oil recovery or carbon dioxide sequestration, especially with carbonatic rock matrix, this approach is costly and time-consuming at best, or impossible to implement at worst. This is based on the two following reasons: First, porosity, permeability and accessible internal surface area in solid rock plugs are often so low that experimental time duration of many months or even years would be required to achieve chemical equilibrium. Secondly, plugs are single specimens of their — generally heterogeneous — original rock formation, which strongly questions the representativeness of single-plug data. To overcome these shortcomings, we present a new methodology based on the combination of systematic crushing, multimethod laboratory measurements and model-based computational evaluation with solving an inverse problem. As a first step, a large amount of undisturbed rock is intentionally crushed and divided in several particle size classes. Then, petrophysical laboratory measurements are carried out on all particle size classes. The resulting data set is finally inverted for the intended properties of the undisturbed rock. This inverse problem entails a three-level forward model, which parametrizes the undisturbed rock properties, particle characteristics and particle packings, but can also be freely adapted to other tasks by any suitable model representation. The three-level model is designed to enforce the petrophysical correlation of all properties at all levels while using a minimal set of model parameters, thus keeping the inverse problem overdetermined. For the inversion, we have developed a publicly available software tool (AnyPetro) based on a Gauss–Newton inversion scheme to minimize a damped least-squares objective function. To demonstrate and validate the proposed methodology, we present a study using five rock types — four carbonates and one sandstone as a reference. Laboratory measurements of complex electrical conductivity (from spectral induced polarization), specific surface (from nitrogen adsorption) and intraparticle porosity (from mercury intrusion) have been carried out on eight particle size classes and on plugs of each rock for comparison. Supportive and complementary analyses include, for example particle geometry, nuclear magnetic resonance, scanning electron microscopy, computer tomography, uniaxial compression strength and mineralogical composition. We show that our new methodology is highly capable of robustly recovering the complex electrical conductivity, specific surface and porosity of the undisturbed rocks from the measured data. The resulting sets of model parameters are petrophysically reasonable and verifiable. The presented methodology can further be applied to the use of drill cuttings as sample material, which is often the only available rock material from deep wells. Our findings also represent a methodological advance for laboratory experiments on reactive systems and both the interpretation and prediction of petrophysical rock properties in such systems.

Accurate Physical Property Predictions via Deep Learning.

Neural networks and deep learning have been successfully applied to tackle problems in drug discovery with increasing accuracy over time. There are still many challenges and opportunities to improve molecular property predictions with satisfactory accuracy even further. Here, we proposed a deep-learning architecture model, namely Bidirectional long short-term memory with Channel and Spatial Attention network (BCSA), of which the training process is fully data-driven and end to end. It is based on data augmentation and SMILES tokenization technology without relying on auxiliary knowledge, such as complex spatial structure. In addition, our model takes the advantages of the long- and short-term memory network (LSTM) in sequence processing. The embedded channel and spatial attention modules in turn specifically identify the prime factors in the SMILES sequence for predicting properties. The model was further improved by Bayesian optimization. In this work, we demonstrate that the trained BSCA model is capable of predicting aqueous solubility. Furthermore, our proposed method shows noticeable superiorities and competitiveness in predicting oil–water partition coefficient, when compared with state-of-the-art graphs models, including graph convoluted network (GCN), message-passing neural network (MPNN), and AttentiveFP.

Engineering geological and petrological characterization of paleoweathered rock in the K1/J2 contact zone in the Ordos Basin, China

Various geological processes (mineral composition, structure, tectonics, weathering, etc.) affect the physical–mechanical properties of rock. Petrological and engineering geological characteristics of paleoweathered rock (PWR) from the K1/J2 contact zone are described in detail via field investigation and experimental testing. This PWR exhibits mainly sandy grains and mud structures, layered and massive strata, and calcareous and argillaceous cements; fissures are developed and often filled with argillaceous and detrital materials; nine minerals and seven oxides are present, and quartz is present in each sample. Long-term weathering results in a consistent bulk density and high total porosity due to the transformation of primary minerals into secondary clay minerals, forming PWR that undergoes argillization in water. The axial point load strength (PLS) is the largest among the tested PLSs, followed by the diametral PLS, and the irregular PLS. The uniaxial compressive strength (UCS) varies widely, but the results are reliable. The mineralogical, physical and mechanical properties of the PWR are compared to predict one parameter from another and study their mutual influence. The PLS and UCS of the PWR are negatively correlated with the elastic mineral group (EMG) content, weathering alteration indexes, water content, and total porosity and positively correlated with the quartz content, brittle mineral group (BMG) content, bulk density, real density, and longitudinal wave velocity. The UCS and the PLS, axial (diametral) PLS and irregular PLS are positively correlated. These results provide a theoretical basis for physical and mechanical property prediction of PWR masses and rapid estimation of UCS in engineering.

Design of mosquito repellent molecules via the integration of hyperbox machine learning and computer aided molecular design

The use of mosquito repellents is an efficient way to prevent mosquito-borne diseases. Despite the accumulation of information about repellents, there remains the challenge of the lack of understanding of their mechanism of action. There is also a need for systematic methods for discovering new alternatives that mitigate the drawbacks of repellents currently in use. To address these research gaps, a computer-aided molecular design (CAMD) framework is developed for the optimal molecular design of mosquito repellents. In this framework, the mosquito repelling attribute of molecules are predicted using a data-driven hyperbox-based machine learning approach in the absence of a mechanistic prediction model. The best set of rules is selected from plausible alternative models developed. For the prediction of important physical properties, a group contribution-based method using reliable models is implemented. Subsequently, the CAMD formulation is developed as a mixed-integer linear programming model to obtain structures with minimum viscosity. Results show that of the structures generated, the hyperbox classifier correctly predicted the repelling ability of all molecules found to be known repellents in literature. The molecules not found in the databases provide key insights on where experimental research to develop new repellents should be targeted. Thus, this newly developed framework can be applied as a systematic technique to screen and narrow down the search space for candidate mosquito repellent molecules before final experimental verification.

Simulation of the soybean oil hydrotreating process for green diesel production

Renewable diesel, or green diesel, is the biofuel with the fastest growing use in the world. Its use as a replacement for fossil diesel is important to reduce the impact of carbon emissions. Green diesel is formed by a mixture of hydrocarbons with a chemical composition similar to that of fossil fuel, making it a drop-in biofuel. Commercial processing employs hydrotreatment of vegetable oil, which, in addition to producing green diesel, can also be used to produce biofuel for aviation, bionaphtha and biopropane. The present paper proposes a model for the hydrotreatment of soybean oil for the production of green diesel. The process was simulated in Aspen Plus® v10; the decarboxylation, decarbonylation and hydrodeoxygenation reactions were carried out in a stoichiometric reactor in the presence of an NiMo/Al2O3 catalyst. In addition to the cracking reactions and fractionation, the fuels produced green diesel, aviation biofuel, and light gases. Brazilian soybean oil was used as the raw material to produce the green diesel. Sensitivity analyses were performed to adjust the best fractionation conditions, considering the composition of the product and the distillation column energy. The best results were chosen when the distillation column was configured with 22 stages, feed at stage 12 and reflux ratio equal to 0.8. The hydrogen/soybean oil ratio was 0.031. The energy used in the process was 1181 kW/h. More than 80% of the soybean oil was converted to renewable fuels, with approximately 65% to green diesel. In addition, a prediction of the simulated physical chemical properties was made and compared with the current Brazilian legislation for green diesel obtaining values within the established specifications. The ASTM D86 distillation curve was compared with simulated results from green diesel obtained from experimental data and fossil diesel. The experimental and simulated distillation curves agreed, showing only a small difference between curves up to 25% of recovered volume. This can be explained by the lighter components present in the experimental green diesel. The proposed simulation supported the potential of Brazilian soybean oil for the production of quality renewable fuels.

Conceptual DFT-Based Computational Peptidology, Pharmacokinetics Study and ADMET Report of the Veraguamides A-G Family of Marine Natural Drugs.

As a continuation of our research on the chemical reactivity, pharmacokinetics and ADMET properties of cyclopeptides of marine origin with potential therapeutic abilities, in this work our already presented integrated molecular modeling protocol has been used for the study of the chemical reactivity and bioactivity properties of the Veraguamides A–G family of marine natural drugs. This protocol results from the estimation of the conceptual density functional theory (CDFT) chemical reactivity descriptors together with several chemoinformatics tools commonly considered within the process of development of new therapeutic drugs. CP-CDFT is a branch of computational chemistry and molecular modeling dedicated to the study of peptides, and it is a protocol that allows the estimation with great accuracy of the CDFT-based reactivity descriptors and the associated physical and chemical properties, which can aid in determining the ability of the studied peptides to behave as potential useful drugs. Moreover, the superiority of the MN12SX density functional over other long-range corrected density functionals for the prediction of chemical and physical properties in the presence of water as the solvent is clearly demonstrated. The research was supplemented with an investigation of the bioactivity of the molecular systems and their ADMET (absorption, distribution, metabolism, excretion, and toxicity) parameters, as is customary in medicinal chemistry. Some instances of the CDFT-based chemical reactivity descriptors’ capacity to predict the pKas of peptides as well as their potential as AGE inhibitors are also shown.

Prediction of Thermal Properties of Zeolites through Machine Learning

The use of machine learning for the prediction of physical and chemical properties of crystals based on their structure alone is currently an area of intense research in computational materials science. In this work, we studied the possibility of using machine-learning-trained algorithms in order to calculate the thermal properties of siliceous zeolite frameworks. We used as training data the thermal properties of 120 zeolites, calculated at the DFT level, in the quasi-harmonic approximation. We compared the statistical accuracy of trained models (based on the gradient-boosting regression technique) using different types of descriptors, including ad hoc geometrical features, topology, pore space, and general geometric descriptors. While geometric descriptors were found to perform best, we also identified limitations on the accuracy of the predictions, especially for a small group of materials with very highly negative thermal expansion coefficients. We then studied the generalizability of the technique, demonstrating that the predictions were not sensitive to the refinement of framework structures at a high level of theory. Therefore, the models are suitable for the exploration and screening of large-scale databases of hypothetical frameworks, which we illustrate on the PCOD2 database of zeolites containing around 600 000 hypothetical structures.

Liquid ethylene glycol: prediction of physical properties, conformer population and interfacial enrichment with a refined non-polarizable force field.

Periodic density functional theory based molecular dynamics simulations confirm the fraction of molecules in neat liquid ethylene glycol with their central OCCO dihedral in the trans conformation to be 21% at ambient conditions, while the rest are gauche conformers. Using this result as a benchmark, two non-polarizable force fields are developed herein to reproduce the conformer populations in the liquid, an important aspect inadequately addressed in several generic force fields. The mean dipole moment of a molecule in the liquid is estimated to be about 40% enhanced over its value in the gas phase, a feature discerned via AIMD simulations and fairly reproduced by our force fields. They are also shown to quantitatively predict all the physical properties of the liquid. Molecules present at the liquid-vapor interface of ethylene glycol are oriented with their methylene groups pointing towards the vapor phase, a requirement that enriches the interface with gauche conformers, in line with polarized sum frequency generation spectroscopy results.

Predicting the Materials Properties Using a 3D Graph Neural Network With Invariant Representation

Accurate prediction of physical properties is critical for discovering and designing novel materials. Machine learning technologies have attracted significant attention in the materials science community for their potential for large-scale screening. Graph Convolution Neural Network (GCNN) is one of the most successful machine learning methods because of its flexibility and effectiveness in describing 3D structural data. Most existing GCNN models focus on the topological structure but overly simplify the three-dimensional geometric structure. However, in materials science, the 3D-spatial distribution of atoms is crucial for determining the atomic states and interatomic forces. This paper proposes an adaptive GCNN with a novel convolution mechanism that simultaneously models atomic interactions among all neighboring atoms in three-dimensional space. We apply the proposed model to two distinctly challenging materials properties prediction problems. The first is Henry’s constant for gas adsorption in Metal-Organic Frameworks (MOFs), which is notoriously difficult because of its high sensitivity to atomic configurations. The second is the ion conductivity in solid-state crystal materials, which is difficult because of the few labeled data available for training. The new model outperforms existing graph-based models on both data sets, suggesting that the critical three-dimensional geometric information is indeed captured.

Ab initio predictions of structures and physical properties of the KCuX (X = Se and Te) phases under pressure

Phase transition, elastic, vibrational and thermal properties of KCuSe and KCuTe are investigated via performing ab-initio calculations combined with quasi-harmonic Debye model. Our results show that these two compounds undergo two first order phase transitions. The calculated equilibrium structural parameters with the modified generalized gradient approximation (PBEsol) agree closely with available experimental values. The mechanical stability is investigated using the calculated elastic constants for all phases. The tetragonal structure of KCuSe is not mechanically stable beyond 8.04 GPa. For KCuTe, C66−P is negative below 7.76 GPa. The mechanical parameters for polycrystalline materials are determined within the framework of the Voigt–Reuss–Hill approximation. From the Poisson’s ratio and the B/G relationship both compounds are brittle in the P63/mmc, while they are ductile in the other phases. The variation of the static and electronic elements of the dielectric tensor with pressure in the different phases is investigated. They show strong anisotropy and discontinuity at the transition pressure. The phonon spectra are also computed along high symmetry lines for both compounds in their different possible phases. The effect of pressure and temperature on the thermal expansion coefficient, bulk modulus, heat capacity and the Debye temperature is analyzed in detail.

Petroleum Physical Properties Prediction Application in Enhanced Oil Recovery Process

The Enhanced Oil Recovery (EOR) process is one of the ways in the petroleum exploitation process so that thick oil can be lifted to the surface and produced. The EOR process referred to in this study is the EOR process carried out in previous studies at the EOR laboratory of UPN Veteran Yogyakarta Indonesia by adding biosurfactants and adjusting the temperature. In laboratory experiments, each time an amount of biosurfactant concentration is added and the temperature is adjusted, the calculation must be done repeatedly to determine the amount of viscosity, interfacial tension (IFT), and density. This experiments takes a long time, requires high cost and variety limitation of the condition. The previous research has succeeded in building a model with multivariate polynomial regression equations to predict the value of the physical properties of crude oil from existing data then classify it into three categories using Naive Bayes, i.e., light oil, medium oil, and heavy oil. The physical properties of petroleum measured in the research are viscosity, interfacial tension, and density. The model uses laboratory data which are taken from the test results of Pertamina's KW-55 well as validation. The validation result shows that Multivariate Polynomial Regression has succeeded in predicting the value of viscosity, interfacial tension, and density with error values ranging from 0% to 1% from the sample data. With a low error value, the application can make forecasting with more variable conditions. The model still cannot be used independently without the Python environment, so to be used easily by more users, the model must be built into an independent application that can be installed on the user's device. In this research, the prediction application of petroleum physical properties has been built. The application is made using the Multivariate Polynomial Regression method according to the model in the previous study to predict the physical properties of petroleum, then uses Naïve Bayes to classify the oil. The application completed the several adjustment to shift from model to application, including user interface, system, and database adjustments.

Meta-learning prediction of physical and chemical properties of magnetized water and fertilizer based on LSTM

BackgroundDue to the high cost of data collection for magnetization detection of media, the sample size is limited, it is not suitable to use deep learning method to predict its change trend. The prediction of physical and chemical properties of magnetized water and fertilizer (PCPMWF) by meta-learning can help to explore the effects of magnetized water and fertilizer irrigation on crops.MethodIn this article, we propose a meta-learning optimization model based on the meta-learner LSTM in the field of regression prediction of PCPMWF. In meta-learning, LSTM is used to replace MAML’s gradient descent optimizer for regression tasks, enables the meta-learner to learn the update rules of the LSTM, and apply it to update the parameters of the model. The proposed method is compared with the experimental results of MAML and LSTM to verify the feasibility and correctness.ResultsThe average absolute percentage error of the meta-learning optimization model of meta-learner LSTM is reduced by 0.37% compared with the MAML model, and by 4.16% compared with the LSTM model. The loss value of the meta-learning optimization model in the iterative process drops the fastest and steadily compared to the MAML model and the LSTM model. In cross-domain experiments, the average accuracy of the meta-learning optimized model can still reach 0.833.ConclusionsIn the case of few sample, the proposed model is superior to the traditional LSTM model and the basic MAML model. And in the training of cross-domain datasets, this model performs best.

Quantitative structure–property relationship of Ev‐degree and Ve‐degree based topological indices : physico‐chemical properties of benzene derivatives

AbstractA quantitative structure–property relationship study was performed for the prediction of physical properties of 22 benzene derivatives using ev‐degree and ve‐degree topological indices. These topological indices correlate certain physico‐chemical property with the chemical structure (like molecular weight, enthalpy, boiling point and π‐electron energy of β‐unit of benzene derivatives). At first, a simple linear regression model was developed using ve‐degree and ev‐degree topological indices and the certain physical properties of the considered benzene derivatives. It is found that the atom bond connectivity index () possesses the best correlating ability among several topological indices to predict the boiling point, the ve‐degree based Randić index is the best predictor of enthalpy, the ev degree based Randić index is the best predictor of π‐electron energy, the sum connectivity index is the best predictor of molecular weight. Second, multiple linear regressions were used for prediction the physical properties (molecular weight, boiling point, enthalpy and π‐electron energy of β‐unit) on the basis of six predictor such as , , , , , and indices. This multi‐linear regression model shows the 100% variation in π‐electron energy of β‐unit. A Maple software based algorithm is used for the calculations of topological indices and the statistical analysis is performed with the SPSS software. The calculations of the ev and ve degree based topological indices of chemical structure of styrene‐butadiene rubber are presented at the end to illustrate the given Maple algorithm.

Exploring DFT+U parameter space with a Bayesian calibration assisted by Markov chain Monte Carlo sampling

The density-functional theory is widely used to predict the physical properties of materials. However, it usually fails for strongly correlated materials. A popular solution is to use the Hubbard correction to treat strongly correlated electronic states. Unfortunately, the values of the Hubbard U and J parameters are initially unknown, and they can vary from one material to another. In this semi-empirical study, we explore the U and J parameter space of a group of iron-based compounds to simultaneously improve the prediction of physical properties (volume, magnetic moment, and bandgap). We used a Bayesian calibration assisted by Markov chain Monte Carlo sampling for three different exchange-correlation functionals (LDA, PBE, and PBEsol). We found that LDA requires the largest U correction. PBE has the smallest standard deviation and its U and J parameters are the most transferable to other iron-based compounds. Lastly, PBE predicts lattice parameters reasonably well without the Hubbard correction.

Quantification of the Properties of Organic Molecules Using Core‐Loss Spectra as Neural Network Descriptors

Artificial neural networks are applied to quantify the properties of organic molecules by introducing a new descriptor, a core‐loss spectrum, which is typically observed experimentally using electron or X‐ray spectroscopy. Using the calculated C K‐edge core‐loss spectra of organic molecules as the descriptor, the neural network models quantitatively predict both intensive and extensive properties, such as the gap between highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) (HOMO–LUMO gap) and internal energy. The prediction accuracy estimated by the mean absolute errors for the HOMO–LUMO gap and internal energy is 0.205 and 97.3 eV, respectively, which are comparable with those of previously reported chemical descriptors. This study indicates that the neural network approach using the core‐loss spectra as the descriptor has the potential to deconvolute the abundant information available in core‐loss spectra for both prediction and experimental characterization of many physical properties. The study shows the practical potential of machine‐learning‐based material property measurements taking advantage of experimental core‐loss spectra, which can be measured with high sensitivity, high spatial resolution, and high temporal resolution.