arXiv subjects

Satadeep Bhattacharjee

Publications and source records attributed to Satadeep Bhattacharjee.

At least 19 recordsLinked to original sources

arXiv2026

Circuit Depth Compression via Spectral Gap Amplification in Quantum Phase Estimation

We show that quantum phase estimation (QPE) circuits can be significantly compressed in depth by preprocessing the input operator with a sigmoid spectral filter before estimation. For systems with small spectral gaps Delta_lambda, standard QPE requires m = ceil(log2(1/Delta_lambda)) precision qubits and depth Theta(2^m). Applying a soft-step transformation f(lambda; tau,w) amplifies the effective gap to Delta_f > Delta_lambda (for w < 1/4), reducing the required precision to m_f = ceil(log2(1/Delta_f)) and compressing circuit depth by 2^(alpha Delta_m), where alpha = 1 for the LMR density-matrix exponentiation framework and alpha is in [0.11,0.42] for controlled-phase-gate circuits. We prove that this compression is exact, bounded above by log2(1/(4w Delta_lambda)) + 1, and impossible for exactly degenerate spectra. We further show that the threshold parameter tau requires only O(w) accuracy, so classical preprocessing such as covariance diagonalisation or CASSCF avoids circularity. A net resource advantage occurs when 4w^2(2^Delta_m - 1) > Delta_lambda log(1/epsilon). Validation on LiH and BeH2 bond-stretch calculations, classical covariance datasets, and synthetic near-degenerate cases demonstrates depth reductions of up to 27x and CX-gate reductions of up to 21x. For LiH, QPE output fidelity improves from 0.66 to 0.98 at a 1% hardware error rate. The method preserves the principal subspace to machine precision, requires no modification of QPE, and can be combined with readout-stage and state-preparation filtering. Negative-control tests establish the benefit condition: m_raw >= 2 and Delta_lambda > 0.

quant-ph↗

arXiv2026

mCGCNN: A Dual-Stream Crystal Graph Convolutional Neural Network for the Efficient Prediction of Magnetic Properties of Crystalline Materials

Magnetic order in crystals is governed by moment-carrying sublattices and ligand-mediated exchange pathways, yet standard crystal graph neural networks treat all atoms homogeneously and encode bonds primarily through pair distances. We propose mCGCNN, a magnetism-aware crystal graph network that augments the full structural graph with a dedicated magnetic subgraph. The magnetic stream performs angle-aware message passing over magnetic centers using metal-ligand-metal exchange-path descriptors motivated by Goodenough-Kanamori-Anderson physics, while layer-wise cross-coupling transfers structural and ligand-field information from the full crystal graph. A separate magnetic-sublattice pooling operation prevents the magnetic interaction from being diluted by nonmagnetic atoms. Benchmarked on a curated Materials Project spin-polarized DFT data, mCGCNN improves total magnetic moment prediction from a CGCNN test MAE of 2.54~$\mu_B$ to 2.02~$\mu_B$, outperforming a strengthened CGCNN readout baseline and raising the test $R^2$ from 0.644 to 0.776. When pretrained on moment regression, the same magnetic representation improves ferromagnetic/antiferromagnetic classification. The results demonstrate that incorporating exchange geometry directly into graph architectures provides a physically grounded route to predictive models of magnetic materials.

cond-mat.mtrl-sci↗

arXiv2026

Uncertainty-Aware Symbolic Regression through Bayesian Support Selection

The Sure Independence Screening and Sparsifying Operator (SISSO) framework is a powerful symbolic-regression method for extracting compact and interpretable descriptors from large nonlinear feature spaces. However, standard SISSO is deterministic: it returns a single descriptor and point prediction, without quantifying uncertainty in descriptor selection, regression coefficients, or predictions. Here we introduce a probabilistic extension in which the sure independence screening (SIS) stage is kept deterministic to preserve scalability, while the sparsifying operator (SO) stage is reformulated as Bayesian inference over the SIS-screened support space. The resulting deterministic-SIS/Bayesian-SO framework yields posterior probabilities for competing descriptor supports, feature-inclusion probabilities, Bayesian-model-averaged predictions, and predictive credible intervals, while recovering the deterministic SO descriptor of standard SISSO in the maximum-a-posteriori limit. Applied to an $X_2YZ$ Heusler-alloy magnetic-moment dataset, the approach gives modest improvements in five-fold cross-validation RMSE and near-nominal empirical coverage of the 95$\%$ predictive intervals. More importantly, the posterior exposes competing, physically related symbolic descriptor families that would appear artificially unique in a deterministic analysis. These results suggest that deterministic-SIS/Bayesian-SO can be used as an uncertainty-aware diagnostic extension of SISSO: a tool for assessing descriptor confidence, stability, and non-uniqueness in small-data materials regression problems.

cond-mat.mtrl-sci↗

arXiv2026

SR-CGCNN: Shared Recurrent Convolution in Crystal Graph Neural Networks for Materials Property Prediction

Crystal graph neural networks predict materials properties by propagating information through local atomic environments. In conventional crystal graph convolutional neural networks (CGCNNs), this propagation depth is increased by stacking independently parameterized convolutional layers. This coupling between message-passing depth and parameter count raises a simple question: can repeated application of the same learned local update recover most of the benefit of a deeper CGCNN? We address this question by introducing a shared-recurrent CGCNN (SR-CGCNN), in which the main crystal-graph convolutional weights are tied across recurrent message-passing steps. The graph construction, pooling operation, and prediction head are kept unchanged, allowing a controlled comparison with standard CGCNN baselines. On Materials Project-derived formation-energy and band-gap datasets, a three-step SR-CGCNN approaches the accuracy of a standard three-layer CGCNN while using only $34.5\%$ of its trainable convolutional parameters. The formation-energy test mean absolute error changes from $0.0945$ to $0.0986~\mathrm{eV\,atom^{-1}}$, while the band-gap error changes from $0.4346$ to $0.4503~\mathrm{eV}$. These results indicate that repeated shared message passing can provide a parameter-efficient approximation to stacked CGCNN depth, offering a compact recurrent interpretation of crystal graph convolution.

cond-mat.mtrl-sci↗

arXiv2026

Filtered Spectral Projection for Quantum Principal Component Analysis

Quantum principal component analysis (qPCA) is commonly formulated as the extraction of eigenvalues and eigenvectors of a covariance-encoded density operator. Yet in many qPCA settings the practical goal is simpler: projection onto the dominant spectral subspace. Here we introduce a projection-first framework, the Filtered Spectral Projection Algorithm (FSPA), which bypasses explicit eigenvalue estimation while preserving the relevant spectral structure. FSPA amplifies any nonzero warm-start overlap with the leading subspace and remains robust in small-gap and near-degenerate regimes, without artificial symmetry breaking in the absence of bias. We show that FSPA achieves an oracle complexity $\mathcal{O}((\log(1/\epsilon)+\log(1/|a_1|^2))/\log(\lambda_1/\lambda_2))$,which is tight by a matching lower bound, establishing it as an\emph{optimal} projection primitive. We derive a convergence rate for degenerate spectra, give a circuit resource analysis with $n+\mathcal{O}(1)$ qubit overhead independent of system dimension, and extend the method to threshold spectral projection, Threshold-FSPA, which converges in $\mathcal{O}(\log(1/\epsilon))$ calls when the threshold lies between eigenvalues. In the density matrix exponentiation access model, FSPA gives an exponential copy-complexity advantage over classical methods. For classical datasets, we show that for amplitude-encoded centered data the ensemble density matrix $\rho=\sum_i p_i|\psi_i\rangle\langle\psi_i|$ equals the covariance matrix. Numerical tests on chemistry density matrices, noisy circuit outputs, Breast Cancer Wisconsin, handwritten Digits, and 1--4-qubit scalability confirm the theory. A minimal Qiskit implementation validates magnitude invariance, signal amplification, and no spurious symmetry breaking. These results establish FSPA as an optimal and deployable quantum spectral projection primitive.

stat.ML↗

arXiv2025

LLM Meets Diffusion: A Hybrid Framework for Crystal Material Generation

Recent advances in generative modeling have shown significant promise in designing novel periodic crystal structures. Existing approaches typically rely on either large language models (LLMs) or equivariant denoising models, each with complementary strengths: LLMs excel at handling discrete atomic types but often struggle with continuous features such as atomic positions and lattice parameters, while denoising models are effective at modeling continuous variables but encounter difficulties in generating accurate atomic compositions. To bridge this gap, we propose CrysLLMGen, a hybrid framework that integrates an LLM with a diffusion model to leverage their complementary strengths for crystal material generation. During sampling, CrysLLMGen first employs a fine-tuned LLM to produce an intermediate representation of atom types, atomic coordinates, and lattice structure. While retaining the predicted atom types, it passes the atomic coordinates and lattice structure to a pre-trained equivariant diffusion model for refinement. Our framework outperforms state-of-the-art generative models across several benchmark tasks and datasets. Specifically, CrysLLMGen not only achieves a balanced performance in terms of structural and compositional validity but also generates more stable and novel materials compared to LLM-based and denoisingbased models Furthermore, CrysLLMGen exhibits strong conditional generation capabilities, effectively producing materials that satisfy user-defined constraints. Code is available at https://github.com/kdmsit/crysllmgen

cs.LG↗

arXiv2025

Role of Oxygen during Methane Oxidation on Pd$_1$/PdO$_1$@CeO$_2$ Surface: A Combined Density Functional Theory, Microkinetic, and Machine Learning Approach

This work explores the role of oxygen in industrial methane oxidation. Oxygen, a well-known oxidizing agent, drives CH$_4$ conversion to CO$_2$ and H$_2$O. We report how oxygen influences oxidation on single Pd and PdO clusters supported on CeO$_2$(111). Oxygen is introduced by (1) lattice O in PdO and (2) O$_2$ adsorption on an isolated Pd atom, forming PdO$_x$ clusters. Density-functional theory (DFT) mapped multiple reaction pathways on the Pd$_1$/PdO$_1$@CeO$_2$(111) surface; both Pd and PdO clusters were found to thermodynamically favour methane activation. The computed barrier for CH$_4$ activation is 0.63 eV on PdO$_1$@CeO$_2$(111). A single Pd atom markedly accelerates O$_2$ dissociation to PdO$_2$, and the presence of lattice oxygen lowers this barrier by 0.36 eV relative to an oxygen-deficient surface, enhancing catalytic efficiency. Reaction selectivity, coverage-dependent production rates, degree of rate control (DRC), and intrinsic turnover frequency (TOF) were quantified through steady-state microkinetic modelling. The simulations predict full conversion of CH$_4$ to CO$_2$ and H$_2$O above 600 K, whereas partial-oxidation intermediates dominate at lower temperature and high O coverage. Rate constants for all elementary steps were derived via the Sure Independence Screening and Sparsifying Operator (SISSO) symbolic-regression method, yielding a concise predictive expression based on charge, coordination number, and key Pd-O/C-H distances. These combined DFT-microkinetic-SISSO results clarify oxygen's mechanistic participation and provide practical guidelines for designing Pd/CeO$_2$ catalysts with improved activity toward methane oxidation, a reaction of pressing environmental and industrial importance.

cond-mat.mtrl-sci↗

arXiv2025

Mechanistic Insights into Complete Methane Oxidation on Single-Atom Pd Supported by SSZ-13 Zeolite: A First-Principles Study

Complete catalytic oxidation of methane is an effective strategy for greenhouse gas mitigation and clean energy conversion; yet, ensuring both high catalytic activity and stability with palladium-based catalysts remains a challenge. In the present work, we employed a theoretical investigation of methane oxidation over single-atom Pd supported on SSZ-13 zeolite using density functional theory calculations, combined with climbing-image nudged elastic band calculations to determine activation barriers. A systematic assessment of various Al distributions and Pd placements was carried out to identify the most stable configurations for Pd incorporation within the zeolite framework.Further, two mechanistic routes for methane activation were evaluated: (i) direct dehydrogenation under dry conditions, and (ii) O$_2$-assisted oxidative dehydrogenation. Our results demonstrate that the direct (dry) pathway is energetically demanding and overall endothermic, whereas the O$_2$ assisted route facilitates the exothermic energy profile, particularly in the C-H bond cleavage. The formation of stable hydroxyl and CO/CO$_2$ intermediates were also studied. The results emphasize the role of oxygen-rich environments in enabling the complete methane oxidation with improved thermodynamic feasibility. Moreover, we propose an alternate low-energy pathway based on O-assisted and multi-site mechanisms that reduce the overall reaction enthalpy. These insights provide the design principles for highly active and moisture-resistant Pd-zeolite catalysts for sustainable methane utilization.

cond-mat.mtrl-sci↗

arXiv2025

Quantum Simulations of Battery Electrolytes with VQE-qEOM and SQD: Active-Space Design, Dissociation, and Excited States of LiPF$_6$, NaPF$_6$, and FSI Salts

Accurate prediction of excited states in battery electrolytes is central to understanding photostability, oxidative stability, and degradation. We employ hybrid quantum-classical algorithms -- the Variational Quantum Eigensolver (VQE) for ground states combined with the quantum equation of motion (qEOM) for vertical singlet excitations -- to study LiPF$_6$, NaPF$_6$, LiFSI, and NaFSI. Compact active spaces were constructed from frontier orbitals, mapped to qubits, and reduced via symmetry tapering and commuting-group measurements to lower sampling cost. Within $\sim$10-qubit models, VQE-qEOM agrees closely with exact diagonalization of the same Hamiltonians, while sample-based quantum diagonalization (SQD) in larger active spaces recovers near-exact (subspace-FCI) energies. The spectra display clear anion and cation trends: PF$_6$ salts exhibit higher first-excitation energies (e.g., LiPF$_6$ $\approx$13.2 eV) and a compact three-state cluster at 12-13 eV, whereas FSI salts show substantially lower onsets ($\approx$8-9 eV) with a near-degenerate (S$_1$,S$_2$) followed by S$_3$ $\sim$1.3 eV higher. Substituting Li$^+$ with Na$^+$ narrows the gap by $\sim$0.4-0.8 eV within each anion family. Converting S$_1$ to wavelengths places the onsets in the deep-UV (LiPF$_6$ $\sim$94 nm; NaPF$_6$ $\sim$100 nm; LiFSI $\sim$141 nm; NaFSI $\sim$148 nm). All results pertain to isolated species or embedded clusters appropriate to the NISQ regime; solvent shifts can be incorporated a posteriori via classical $\Delta$-solvation or static embedding. These results demonstrate that current quantum algorithms can deliver chemically meaningful excitation and binding trends for realistic electrolyte motifs and provide quantitative baselines to guide electrolyte screening and design.

cond-mat.mtrl-sci↗

arXiv2025

Rashba-induced spin Hall response in a disordered $WTe_2$ four-terminal structure

The paramount acumen for controlling spin transport properties in nonmagnetic materials is the usage of spin-orbit coupling (SOC). We propose a model to calculate the spin hall angle (SHA) for the elemental transition metal dichalcogenide compound, $WTe_2$ entrenched on the intrinsic Rashba SOC. This model, is based on the Landauer-Buttiker formalism for quantum transport, and the $4$-terminal device setup with the presence of disorder from random onsite potential fluctuations. The SHA, including the mean and RMS values, also illustrate the mesoscopic oscillations, and the values obtained are $25\%$ and $30\%$, respectively. The variation pattern of charge and spin current, along with the mean and spin Hall conductance, can be a comparative measure for other TMDs and monolayer Xenes. To validate our outcomes, we compare our results with experimental data and numerically extract real-space simulation results based on the nearest-neighbor tight binding (NNTB) model. Also our results are in line with the scaling theory of localization. This work sets the stage to calculate the spin Hall angle and spin Hall conductivity for other elemental monolayer Xenes and TMDs, considering the intrinsic scattering mechanisms. An extension of this work will be to explore the possible spintronics applications for extrinsic scattering, including side-jump and skew-jump scattering processes.

cond-mat.mes-hall↗

arXiv2025

Testing the spin-bath view of self-attention: A Hamiltonian analysis of GPT-2 Transformer

The recently proposed physics-based framework by Huo and Johnson~\cite{huo2024capturing} models the attention mechanism of Large Language Models (LLMs) as an interacting two-body spin system, offering a first-principles explanation for phenomena like repetition and bias. Building on this hypothesis, we extract the complete Query-Key weight matrices from a production-grade GPT-2 model and derive the corresponding effective Hamiltonian for every attention head. From these Hamiltonians, we obtain analytic phase boundaries and logit gap criteria that predict which token should dominate the next-token distribution for a given context. A systematic evaluation on 144 heads across 20 factual-recall prompts reveals a strong negative correlation between the theoretical logit gaps and the model's empirical token rankings ($r\approx-0.70$, $p<10^{-3}$).Targeted ablations further show that suppressing the heads most aligned with the spin-bath predictions induces the anticipated shifts in output probabilities, confirming a causal link rather than a coincidental association. Taken together, our findings provide the first strong empirical evidence for the spin-bath analogy in a production-grade model. In this work, we utilize the context-field lens, which provides physics-grounded interpretability and motivates the development of novel generative models bridging theoretical condensed matter physics and artificial intelligence.

cond-mat.mtrl-sci↗

arXiv2025

Tunable Thermal Expansion in Functionalized 2D Boron Nitride: A First-Principles Investigation

This study investigates the thermal expansion coefficient of two-dimensional (2D) functionalized boron nitride (f-BN) materials using first-principles density functional theory (DFT). Two-dimensional materials, particularly hexagonal boron nitride (h-BN), have attracted significant attention due to their exceptional mechanical, thermal, and electronic properties. However, the influence of functionalization on the thermal expansion behavior remains largely unexplored. In this work, DFT calculations are employed to analyze how different functionalized forms of h-BN impact the thermal expansion of BN sheets. Density functional perturbation theory (DFPT) and the quasiharmonic approximation (QAH) are utilized to determine the thermal expansion coefficient over a range of temperatures. The results reveal that functionalization induces notable modifications in the in-plane thermal expansion of BN, affecting material stability and suggesting potential applications in nanoelectronics and thermal management. This investigation provides critical insights into the tunability of the thermal properties of 2D BN, underscoring its suitability for next-generation flexible and high-performance devices.

cond-mat.mtrl-sci↗

arXiv2025

Multi-site mixing and entropy stabilization of CsPbI$_{3}$ with potential application in photovoltaics

Metal halide perovskite solar cells have achieved dramatic improvements in their power conversion efficiency in the recent past. Since compositional engineering plays an important role in optimizing material properties, we investigate the effect of alloying at Cs and Pb sites on the energetics and electronic structure of CsPbI$_{3}$ using cluster expansion method in combination with first-principles calculations. For Ge-mixing at Pb-site, the $\alpha$ and $\beta$-phases are considered with emphasis on the electronic structure, transition probability, absorption coefficient, efficiency, and carrier mobility of higher-symmetry configurations. CsPb$_{0.50}$Ge$_{0.50}$I$_{3}$ (Cs$_{2}$PbGeI$_{6}$) which takes up a double perovskite (elpasolite) structure has a direct band gap with no parity-forbidden transitions. Further, we utilize the alloy entropic effect to improve the material stability and optoelectronic properties of CsPbI$_{3}$ by multi-element mixing. For the proposed mixed compositions, the Fr{\"o}hlich electron-phonon coupling constant is determined. Scattering rates and electron mobility are obtained from first-principles inputs. These lower Pb-content inorganic perovskites offer great promise as efficient solar cell materials for photovoltaic applications.

cond-mat.mtrl-sci↗

arXiv2025

Integrating Density Functional Theory with Deep Neural Networks for Accurate Voltage Prediction in Alkali-Metal-Ion Battery Materials

Accurate prediction of the voltage of battery materials plays a pivotal role in the advancement of energy storage technologies and the rational design of high-performance cathode materials. In this work, we present a deep neural network (DNN) model, built using PyTorch, to estimate the average voltage of cathode materials across Li-ion, Na-ion, and other alkali-metal-ion batteries. The model is trained on an extensive dataset from the Materials Project, incorporating a wide range of specific structural, physical, chemical, electronic, thermodynamic, and battery descriptors, ensuring a comprehensive representation of material properties. Our model exhibits strong predictive performance, as corroborated by first-principles density functional theory (DFT) calculations. The close alignment between the DNN predictions and the DFT outcomes highlights the robustness and accuracy of our machine learning framework to effectively select and identify viable battery materials. Using this validated model, we successfully proposed novel Na-ion battery compositions, with their predicted behavior confirmed by rigorous computational assessment. By seamlessly integrating data-driven prediction with first-principles validation, this study presents an effective framework that significantly accelerates the discovery and optimization of advanced battery materials, contributing to the development of more reliable and efficient energy storage technologies.

cond-mat.mtrl-sci↗

arXiv2025

Periodic Materials Generation using Text-Guided Joint Diffusion Model

Equivariant diffusion models have emerged as the prevailing approach for generating novel crystal materials due to their ability to leverage the physical symmetries of periodic material structures. However, current models do not effectively learn the joint distribution of atom types, fractional coordinates, and lattice structure of the crystal material in a cohesive end-to-end diffusion framework. Also, none of these models work under realistic setups, where users specify the desired characteristics that the generated structures must match. In this work, we introduce TGDMat, a novel text-guided diffusion model designed for 3D periodic material generation. Our approach integrates global structural knowledge through textual descriptions at each denoising step while jointly generating atom coordinates, types, and lattice structure using a periodic-E(3)-equivariant graph neural network (GNN). Extensive experiments using popular datasets on benchmark tasks reveal that TGDMat outperforms existing baseline methods by a good margin. Notably, for the structure prediction task, with just one generated sample, TGDMat outperforms all baseline models, highlighting the importance of text-guided diffusion. Further, in the generation task, TGDMat surpasses all baselines and their text-fusion variants, showcasing the effectiveness of the joint diffusion paradigm. Additionally, incorporating textual knowledge reduces overall training and sampling computational overhead while enhancing generative performance when utilizing real-world textual prompts from experts.

cs.LG↗

arXiv2024

Local Symmetry Breaking in Skyrmion-Hosting Centrosymmetric Hexagonal Compounds

Dzyaloshinskii-Moriya interaction (DMI) plays a crucial role in stabilizing the exotic topologically stable skyrmion spin textures in the noncentrosymmetric crystals. The recent discovery of biskyrmions and skyrmions in the globally centrosymmetric crystals has raised debate about the role of the DMI in causing the spin textures, since DMI vanishes in such crystal structures. Theoretical studies, on the other hand, suggest non-vanishing DMI even if there is local inversion symmetry breaking in an otherwise globally centrosymmetric crystal structure. Motivated by such theoretical predictions, we present here the results of a systematic crystal structure study of two skyrmion-hosting Ni2In-type centrosymmetric hexagonal compounds, MnNiGa and MnPtGa, using the atomic pair distribution function (PDF) technique. Our result provides information about structural correlations in the short-range (SR), medium-range (MR) and long-range (LR) regimes simultaneously. The analysis of the experimental PDFs, obtained from high flux, high energy, and high Q synchrotron x-ray powder diffraction patterns, reveals that the local SR structure of both MnNiGa and MnPtGa compounds corresponds to the noncentrosymmetric trigonal space group P3m1, while the structure in the MR+LR regimes remains hexagonal in the centrosymmetric P63/mmc space group. These findings are also supported by theoretical DFT calculations. Our results, in conjunction with the previous theoretical predictions, provide a rationale for the genesis of skyrmions in centrosymmetric materials in terms of non-vanishing DMI due to local inversion symmetry breaking. We believe that our findings would encourage a systematic search of skyrmionic textures and other topological phenomena in a vast family of centrosymmetric materials.

cond-mat.mtrl-sci↗

arXiv2024

Enhancement of spin Hall angle by an order of magnitude via Cu intercalation in MoS$_2$/CoFeB heterostructures

Transition metal dichalcogenides (TMDs) are a novel class of quantum materials with significant potential in spintronics, optoelectronics, valleytronics, and opto-valleytronics. TMDs exhibit strong spin-orbit coupling, enabling efficient spin-charge interconversion, which makes them ideal candidates for spin-orbit torque-driven spintronic devices. In this study, we investigated the spin-to-charge conversion through ferromagnetic resonance in MoS$_2$/Cu/CoFeB heterostructures with varying Cu spacer thicknesses. The conversion efficiency, quantified by the spin Hall angle, was enhanced by an order of magnitude due to Cu intercalation. Magneto-optic Kerr effect microscopy confirmed that Cu did not significantly modify the magnetic domains, indicating its effectiveness in decoupling MoS$_2$ from CoFeB. This decoupling preserves the spin-orbit coupling (SOC) of MoS$_2$ by mitigating the exchange interaction with CoFeB, as proximity to localized magnetization can alter the electronic structure and SOC. First-principles calculations revealed that Cu intercalation notably enhances the spin Berry curvature and spin Hall conductivity, contributing to the increased spin Hall angle. This study demonstrates that interface engineering of ferromagnet/TMD-based heterostructures can achieve higher spin-to-charge conversion efficiencies, paving the way for advancements in spintronic applications.

cond-mat.mtrl-sci↗

arXiv2024

A Hybrid Machine Learning Framework for Predicting Hydrogen Storage Capacities in Metal Hydrides: Unsupervised Feature Learning with Deep Neural Networks

In this study, we present a sophisticated hybrid machine-learning framework that significantly improves the accuracy of predicting hydrogen storage capacities in metal hydrides. This is a critical challenge due to the scarcity of experimental data and the complexity of high-dimensional feature spaces. Our approach employs the power of unsupervised learning through the use of a state-of-the-art autoencoder. This autoencoder is trained on elemental descriptors obtained from Mendeleev software, enabling the extraction of a meaningful and lower dimensional latent space from the input data. This latent representation serves as the basis for our deep multi-layer perceptron (MLP) model, which consists of five layers and shows good precision in predicting hydrogen storage capacities. Furthermore, our results show very good agreement with the results of density functional theory (DFT). In addition to addressing the limitations caused by limited and unevenly distributed data in the field of hydrogen storage materials, we also focus on discovering new materials that show promising opportunities for hydrogen storage. These materials were identified using both feature-based approaches and predictions generated by a large language model. Finally, our investigation into the effectiveness of transferring weights from the autoencoder to the MLP, in addition to the latent features, suggests that while this strategy slightly improves model performance indicated by a slightly higher R$^2$ value and lower RMSE, it emphasizes the intricate challenge of adapting pre-trained weights for specific supervised tasks.

cond-mat.mtrl-sci↗