arXiv subjects

Deniz Gunceler

Publications and source records attributed to Deniz Gunceler.

15 recordsLinked to original sources

arXiv2026

MatterSim-MT: A multi-task foundation model for in silico materials characterization

Accurate property characterization is a major bottleneck in materials design. While first-principles methods and task-specific machine-learning models have driven important progress, they remain fundamentally limited in scalability and generalizability across the vast space of structures and properties relevant to real-world materials design. We present MatterSim-MT, a multi-task foundation model for in silico materials simulation and property characterization. The model is pretrained on over 35 million first-principles-labeled structures covering 89 elements, temperatures up to 5000 K and pressures up to 1000 GPa, and is fine-tuned on various properties including Bader charges, magnetic moments, Born effective charges, and dielectric matrices. Out of the box, MatterSim-MT not only serves as a foundation model for predicting material structure, dynamics and thermodynamics, its multi-task architecture also enables a wide range of complex simulations that cannot be captured by potential energy surfaces alone. For example, we demonstrate pressure-dependent LO-TO phonon splitting in SiC with close agreement with experiment, electric hysteresis in ferroelectric BaTiO3, and the cationic-to-anionic redox transition during delithiation of a Li-rich cathode material. Finally, we show that MatterSim-MT scales well with more data and parameters, can be efficiently fine-tuned to higher levels of theory, and can be efficiently extended to new systems via active learning. Overall, we believe this approach provides a scalable route to accurate in silico materials characterization.

cond-mat.mtrl-sci↗

arXiv2025

Accurate and scalable exchange-correlation with deep learning

Density Functional Theory (DFT) underpins much of modern computational chemistry and materials science. Yet, the reliability of DFT-derived predictions of experimentally measurable properties remains fundamentally limited by the need to approximate the unknown exchange-correlation (XC) functional. The traditional paradigm for improving accuracy has relied on increasingly elaborate hand-crafted functional forms. This approach has led to a longstanding trade-off between computational efficiency and accuracy, which remains insufficient for reliable predictive modelling of laboratory experiments. Here we introduce Skala, a deep learning-based XC functional that surpasses state-of-the-art hybrid functionals in accuracy across the main-group chemistry benchmark set GMTKN55 with an error of 2.8 kcal/mol, while retaining the lower computational cost characteristic of semi-local DFT. This demonstrated departure from the historical trade-off between accuracy and efficiency is enabled by learning non-local representations of electronic structure directly from data, bypassing the need for increasingly costly hand-engineered features. Leveraging an unprecedented volume of high-accuracy reference data from wavefunction-based methods, we establish that modern deep learning enables systematically improvable neural exchange-correlation models as training datasets expand, positioning first-principles simulations to become progressively more predictive.

physics.chem-ph↗

arXiv2020

Bootstrap an end-to-end ASR system by multilingual training, transfer learning, text-to-text mapping and synthetic audio

Bootstrapping speech recognition on limited data resources has been an area of active research for long. The recent transition to all-neural models and end-to-end (E2E) training brought along particular challenges as these models are known to be data hungry, but also came with opportunities around language-agnostic representations derived from multilingual data as well as shared word-piece output representations across languages that share script and roots. We investigate here the effectiveness of different strategies to bootstrap an RNN-Transducer (RNN-T) based automatic speech recognition (ASR) system in the low resource regime, while exploiting the abundant resources available in other languages as well as the synthetic audio from a text-to-speech (TTS) engine. Our experiments demonstrate that transfer learning from a multilingual model, using a post-ASR text-to-text mapping and synthetic audio deliver additive improvements, allowing us to bootstrap a model for a new language with a fraction of the data that would otherwise be needed. The best system achieved a 46% relative word error rate (WER) reduction compared to the monolingual baseline, among which 25% relative WER improvement is attributed to the post-ASR text-to-text mappings and the TTS synthetic data.

eess.AS↗

arXiv2020

Using Synthetic Audio to Improve The Recognition of Out-Of-Vocabulary Words in End-To-End ASR Systems

Today, many state-of-the-art automatic speech recognition (ASR) systems apply all-neural models that map audio to word sequences trained end-to-end along one global optimisation criterion in a fully data driven fashion. These models allow high precision ASR for domains and words represented in the training material but have difficulties recognising words that are rarely or not at all represented during training, i.e. trending words and new named entities. In this paper, we use a text-to-speech (TTS) engine to provide synthetic audio for out-of-vocabulary (OOV) words. We aim to boost the recognition accuracy of a recurrent neural network transducer (RNN-T) on OOV words by using the extra audio-text pairs, while maintaining the performance on the non-OOV words. Different regularisation techniques are explored and the best performance is achieved by fine-tuning the RNN-T on both original training data and extra synthetic data with elastic weight consolidation (EWC) applied on the encoder. This yields a 57% relative word error rate (WER) reduction on utterances containing OOV words without any degradation on the whole test set.

eess.AS↗

arXiv2017

JDFTx: software for joint density-functional theory

Density-functional theory (DFT) has revolutionized computational prediction of atomic-scale properties from first principles in physics, chemistry and materials science. Continuing development of new methods is necessary for accurate predictions of new classes of materials and properties, and for connecting to nano- and mesoscale properties using coarse-grained theories. JDFTx is a fully-featured open-source electronic DFT software designed specifically to facilitate rapid development of new theories, models and algorithms. Using an algebraic formulation as an abstraction layer, compact C++11 code automatically performs well on diverse hardware including GPUs. This code hosts the development of joint density-functional theory (JDFT) that combines electronic DFT with classical DFT and continuum models of liquids for first-principles calculations of solvated and electrochemical systems. In addition, the modular nature of the code makes it easy to extend and interface with, facilitating the development of multi-scale toolkits that connect to ab initio calculations, e.g. photo-excited carrier dynamics combining electron and phonon calculations with electromagnetic simulations.

cond-mat.mtrl-sci↗

arXiv2015

Structure of the photo-catalytically active surface of SrTiO3

A major goal of energy research is to use visible light to cleave water directly, without an applied voltage, into hydrogen and oxygen. Since the initial reports of the ultraviolet (UV) activity of TiO2 and SrTiO3 in the 1970s, researchers have pursued a fundamental understanding of the mechanistic and molecular-level phenomena involved in photo-catalysis. Although it requires UV light, after four decades SrTiO3 is still the gold standard for splitting water. It is chemically stable and catalyzes both the hydrogen and the oxygen reactions without applied bias. While ultrahigh vacuum (UHV) surface science techniques have provided useful insights, we still know relatively little about the structure of electrodes in contact with electrolytes under operating conditions. Here, we report the surface structure evolution of a SrTiO3 electrode during water splitting, before and after training with a positive bias. Operando high-energy X-ray reflectivity measurements demonstrate that training the electrode irreversibly reorders the surface. Scanning electrochemical microscopy (SECM) at open circuit correlates this training with a tripling of the activity toward photo-induced water splitting. A novel first-principles joint density-functional theory (JDFT) simulation constrained to the X-ray data via a generalized penalty function identifies an anatase-like structure for the more active, trained surface.

physics.chem-ph↗

arXiv2015

Anatase TiO$_2$ Nanowires Functionalized by Organic Sensitizers for Solar Cells : A Screened Coulomb Hybrid Density Functional Study

The adsorption of two different organic molecules cyanidin glucoside (C$_{21}$O$_{11}$H$_{20}$) and TA-St-CA on anatase (101) and (001) nanowires have been investigated using the standard and the range separated hybrid density functional theory calculations. The electronic structures and optical spectra of resulting dye--nanowire combined systems show distinct features for these types of photochromophores. The lowest unoccupied molecular orbital of the natural dye cyanidin glucoside is located below the conduction band of the semiconductor while, in the case of TA-St-CA, it resonates with the states inside the conduction band. The wide-bandgap anatase nanowires can be functionalized for solar cells through electron-hole generation and subsequent charge injection by these dye sensitizers. The intermolecular charge transfer character of Donor-$\pi$-Acceptor type dye TA-St-CA is substantially modified by its adsorption on TiO$_2$ surfaces. Cyanidin glucoside exhibits relatively stronger anchoring on the nanowires through its hydroxyl groups. The atomic structures of dye--nanowire systems re-optimized with the inclusion of nonlinear solvation effects showed that the binding strengths of both dyes remain moderate even in ionic solutions.

cond-mat.mes-hall↗

arXiv2015

Stability and surface diffusion at lithium-electrolyte interphases with connections to dendrite suppression

This work presents an ab initio exploration of fundamental mechanisms with direct relevance to dendrite formation at lithium-electrolyte interfaces. Specifically, we explore surface diffusion barriers and solvated surface energies of typical solid-electrolyte interphase layers of lithium metal electrodes. Our results indicate that surface diffusion is an important mechanism for understanding the recently observed dendrite suppression from lithium-halide passivating layers, which were motivated by our previous work. Our results uncover possible mechanisms underlying a new pathway for mitigating dendridic electrodeposition of lithium on metal and thereby contribute to the ongoing efforts to develop stable lithium metal anodes for rechargeable battery systems.

cond-mat.mtrl-sci↗

arXiv2014

Weighted-density functionals for cavity formation and dispersion energies in continuum solvation models

Continuum solvation models enable efficient first principles calculations of chemical reactions in solution, but require extensive parametrization and fitting for each solvent and class of solute systems. Here, we examine the assumptions of continuum solvation models in detail and replace empirical terms with physical models in order to construct a minimally-empirical solvation model. Specifically, we derive solvent radii from the nonlocal dielectric response of the solvent from ab initio calculations, construct a closed-form and parameter-free weighted-density approximation for the free energy of the cavity formation, and employ a pair-potential approximation for the dispersion energy. We show that the resulting model with a single solvent-independent parameter: the electron density threshold ($n_c$), and a single solvent-dependent parameter: the dispersion scale factor ($s_6$), reproduces solvation energies of organic molecules in water, chloroform and carbon tetrachloride with RMS errors of 1.1, 0.6 and 0.5 kcal/mol respectively. We additionally show that fitting the solvent-dependent $s_6$ parameter to the solvation energy of a single non-polar molecule does not substantially increase these errors. Parametrization of this model for other solvents, therefore, requires minimal effort and is possible without extensive databases of experimental solvation free energies.

physics.chem-ph↗

arXiv2014

Range separated hybrid density functional study of organic dye sensitizers on anatase TiO$_2$ nanowires

The adsorption of organic molecules coumarin and the donor-$\pi$-acceptor type tetrahydroquinoline (C2-1) on anatase (101) and (001) nanowires have been investigated using screened Coulomb hybrid density functional theory calculations. While coumarin forms single bond with the nanowire surface, C2-1 additionally exhibits bidentate mode giving rise to much stronger adsorption energies. Nonlinear solvation effects on the binding characteristics of the dye chromophores on the nanowire facets have also been examined. These two dye sensitizers show different electronic charge distributions for the highest occupied and the lowest unoccupied molecular states. We studied the electronic structures in terms of the positions of the band edges and adsorbate related band gap states and their effect on the absorption spectra of the dye-nanowire combined systems. These findings were interpreted and discussed from the view point of better light harvesting and charge separation as well as in relation to more efficient charge carrier injection into the semiconductor nanowire.

cond-mat.mes-hall↗

arXiv2014

Universal iso-density polarizable continuum model for molecular solvents

Implicit electron-density solvation models based on joint density-functional theory offer a computationally efficient solution to the problem of calculating thermodynamic quantities of solvated systems from firstprinciples quantum mechanics. However, despite much recent interest in such models, to date the applicability of such models in the plane-wave context to non-aqueous solvents has been limited because the determination of the model parameters requires fitting to a large database of experimental solvation energies for each new solvent considered. This work presents an alternate approach which allows development of new iso-density models for a large class of protic and aprotic solvents from only simple, single-molecule ab initio calculations and readily available bulk thermodynamic data.

physics.chem-ph↗

arXiv2013

Nanoscale Imaging of Lithium Ion Distribution During In Situ Operation of Battery Electrode and Electrolyte

A major challenge in the development of new battery materials is understanding their fundamental mechanisms of operation and degradation. Their microscopically inhomogeneous nature calls for characterization tools that provide operando and localized information from individual grains and particles. Here we describe an approach that images the nanoscale distribution of ions during electrochemical charging of a battery in a transmission electron microscope liquid flow cell. We use valence energy-loss spectroscopy to track both solvated and intercalated ions, with electronic structure fingerprints of the solvated ions identified using an ab initio non-linear response theory. Equipped with the new electrochemical cell holder, nanoscale spectroscopy and theory, we have been able to determine the lithiation state of a LiFePO4 electrode and surrounding aqueous electrolyte in real time with nanoscale resolution during electrochemical charge and discharge. We follow lithium transfer between electrode and electrolyte and observe charging dynamics in the cathode that differ among individual particles. This technique represents a general approach for the operando nanoscale imaging of electrochemically active ions in a wide range of electrical energy storage systems.

physics.chem-ph↗

arXiv2013

A study of the density functional methods on the photoabsorption of Bodipy dyes

Tunability of the photoabsorption and directional charge injection characteristics of Bodipy-based dye molecules with different carbonyl groups make them promising candidates for photovoltaic applications. In order to study the effect of screening in the Coulomb interaction on the electronic and optical properties of two Bodipy derivatives, we have used linear response time-dependent and exact exchange hybrid density functional approaches. The effect of linear and non-linear solvation models on the electrochemical properties of the dyes has also been discussed.

physics.chem-ph↗

arXiv2013

The importance of nonlinear fluid response in joint density-functional theory studies of battery systems

Delivering the full benefits of first principles calculations to battery materials demands the development of accurate and computationally-efficient electronic structure methods that incorporate the effects of the electrolyte environment and electrode potential. Realistic electrochemical interfaces containing polar surfaces are beyond the regime of validity of existing continuum solvation theories developed for molecules, due to the presence of significantly stronger electric fields. We present an ab initio theory of the nonlinear dielectric and ionic response of solvent environments within the framework of joint density-functional theory, with precisely the same optimizable parameters as conventional polarizable continuum models. We demonstrate that the resulting nonlinear theory agrees with the standard linear models for organic molecules and metallic surfaces under typical operating conditions. However, we find that the saturation effects in the rotational response of polar solvent molecules, inherent to our nonlinear theory, are crucial for a qualitatively correct description of the ionic surfaces typical of the solid electrolyte interface.

cond-mat.mtrl-sci↗

arXiv2010

dc-switchable and single-nanocrystal-addressable coherent population transfer

Achieving coherent population transfer in the solid-state is challenging compared to atomic systems due to closely spaced electronic states and fast decoherence. Here, within an atomistic pseudopotential theory, we offer recipes for the stimulated Raman adiabatic passage in embedded silicon and germanium nanocrystals. The transfer efficiency spectra displays characteristic Fano resonances. By exploiting the Stark effect, we predict that transfer can be switched off with a dc voltage. As the population transfer is highly sensitive to structural variations, with a choice of a sufficiently small two-photon detuning bandwidth, it can be harnessed for addressing individual nanocrystals within an ensemble.

cond-mat.mes-hall↗