Presentations

Ksenia Bravaya (Boston University)

Electronic Structure Theory for Electronic Resonances

Metastable electronic states, or electronic resonances, lie in the electron ionization or detachment continuum and cannot be described using conventional electronic structure methods developed for bound states. In this talk, I will present recent developments from my group that extend established bound-state methods and software to electronic resonances through the projected complex absorbing potential (CAP) approach [1].

I will discuss the performance of a range of quantum chemistry models for predicting the energies and lifetimes of different classes of resonances and demonstrate that the projected CAP approach yields smooth and self-consistent potential energy surfaces for metastable states [2]. I will also present a scheme for computing analytic nuclear

gradients and nonadiabatic couplings for metastable electronic states using quantities obtained from standard boundstate methods and software [3]. Together, these developments provide a practical framework for exploring the potential energy surfaces of electronic resonances and simulating nonadiabatic dynamics involving metastable electronic states.

J.R. Gayvert and K.B. Bravaya; Projected CAP-EOM-CCSD method for electronic resonances, J. Chem. Phys., 2022, 156, 094108.
S. Mondal and K.B. Bravaya; Complex potential energy surfaces with projected CAP technique: Vibrational excitation of N2, J. Chem. Phys, 2024, 61, 024106.
S. Mondal and K.B. Bravaya; Analytic nuclear gradients for complex potential energy surfaces: a projected CAP approach, J. Chem. Theory Comput., 2025, 21, 5986–5996.

About: Ksenia Bravaya received her Ph.D. in Theoretical Chemistry from Lomonosov Moscow State University and is now Associate Professor of Chemistry at Boston University. Her research focuses on developing computational methods to characterize electron transfer and electronic resonances. Her work has been recognized with an Alfred P. Sloan Research Fellowship (2020).

Kay Carter-Fenk (University of Pittsburgh)

Pushing Single-Reference to Its Limits with Simple, Robust Coupled-Cluster Models

Static correlation is relevant in myriad chemical contexts but incorporating it into quantum chemistry calculations is exponentially expensive. To avoid such high computational scaling, there is an ongoing effort to simplify single-reference coupled-cluster (CC) equations, especially since certain simplifications can improve the performance of CC models in cases where static correlation predominates. Linearized CC theory is a form of simplified CC that offers computational benefits due to its Hermitian structure, but it also suffers from singularities in the strongly-correlated regime. My group has found that the singularities in linearized CC can be avoided by omitting ring diagrams. The resultant linearized-ladder CC theory avoids singularities but remains under-correlated due to a lack of particle-hole correlation. We amend this by including ring terms to second-order in a novel linearized external CC perturbation theory (xlinCC(2)) approach. Our results suggest that xlinCC(2) atop a linearized-ladder CC reference can avoid catastrophic divergences of linearized CC theory, attaining CCD-like accuracy for covalent bond breaking at around one tenth the cost.

Another avenue for improving CC results without increasing the cost is by choosing nonstandard reference densities. We have recently found that using Kohn-Sham references in the gold-standard CCSD(T) method can vastly improve predictions for metal dimers, including the notorious Cr2 potential energy surface. I emphasize that no choice of orbitals is sacred, but the space they span is key to finding different solutions to the CCSD(T) equations. These results will be useful for generating accurate benchmark data for semiempirical and machine-learning models.

About: Kevin (Kay) Carter-Fenk is an Assistant Professor at the University of Pittsburgh. Their work focuses on the development of simple quantum chemistry methods for ground and excited-states of strongly correlated systems. Their group is especially interested in applying these methods to problems in sustainable energy and wildlife conservation.

Ji Chen (Peking University)

Neural Network Wave Function: A nified Framework for Correlated Systems

Neural network wave functions have emerged as a powerful approach for solving correlated electronic structures of real materials and model systems. In this talk, I will present two latest examples of extending neural network wave functions to more complex chemical and physical systems, including large molecules with transition metals (such as iron sulfur clusters) and correlated topological states (such as Wigner crystals and fractional topological insulators). I will highlight the successful applications of neural network wave function as a variational solver that can potentially surpass the current SOTA of ab initio calculations and shed light on exotic physical states beyond the current understanding. Perspective and outlook will also be discussed to highlight future opportunities and challenges in the field.

About: Ji Chen is a Tsientang associate professor at Peking University. He received his BSc from USTC and PhD from Peking University, followed by postdoctoral research at UCL and Max Planck Institute. His research focuses on developing new computational methods for condensed matter physics and chemical physics, and materials science.

Emmanuel Giner (Sorbonne University )

Using DFT in Various Contexts: Curing Finite Basis Set Errors and Implicit Solvation for QMMM Calculations

In this presentation, I will summarise two distinct strategies where a DFT-based approach can be used to effectively account for many degrees of freedom. In the first part, I will present the DFT-based basis set correction that allows to obtain an effective and cheap correction for the finite basis set error of WFT calculations. In the second part, I will show how a QMMM calculation where the MM part consists in many water molecules can be reframed using a classical DFT framework.

About: Emmanuel Giner completed his PhD in Toulouse on QMC and selected CI methods for improving DMC wave functions. After postdoctoral work in Ferrara and Stuttgart on multireference methods, he joined CNRS in Paris as a researcher, developing electronic structure methods spanning DFT, transcorrelated, QMC, and QM/MM approaches.

Diptarka Hait (Columbia University)

Size Consistent Padé Resummed Wavefunction Theories

Padé resummation is widely used in applied mathematics and physical sciences to approximate functions better than truncated Taylor series. Unfortunately, naive Padé resummation of many body perturbation theory is not size consistent, leading to limited use in modern quantum chemistry. We present a matrix based approach to size consistent Padé resummed Møller-Plesset perturbation theory, utilizing only second and third order contributions to correlation energy. This approach provides near chemical accuracy for the standard datasets used to evaluate single-reference quantum chemical methods. In particular, it yields an RMSE of 1.6 kcal/mol for the total atomization energies of the non-multireference subset of the W4-17 thermochemistry dataset (vs 12 kcal/mol for CCSD). Significantly improved performance is also obtained for other properties like barrier heights, noncovalent interactions, dipole moments and transition metal thermochemistry. Our approach thus appears to potentially represent the most accurate O(N^6) scaling noniterative method for single-reference quantum chemistry, and therefore ought to be broadly applicable in many regimes (including the development of double hybrid functionals and/or new composite methods). Time permitting, I will also briefly talk about Padé resummed Coupled Cluster methods for further enhancement of accuracy without any increase in asymptotic cost.

About: Dip Hait is an Assistant Professor at Columbia University and an Associate Research Scientist at the Flatiron Institute. Dip is interested in developing new quantum chemical methods and applying them to practical chemical problems. Areas of current interest include simulating electronic excited states, modeling electron correlation, photocatalysis and computational spectroscopy.

Martin P. Head-Gordon (UC Berkeley)

Postmodern Local Correlation Theory: Simplicity, Efficiency, and Precision

Without a doubt, the two decades of development of local correlation within the dominant paradigm of domain localized pair natural orbital (DL-PNO) theory have yielded great benefit to our field. The result has been practical linear scaling algorithms for standard quantum chemistry methods for electron correlation ranging from MP2 to CCSD(T), and even beyond. However, the DL-PNO approach does have drawbacks: algebraic complications from using non-orthogonal localized virtual orbitals, and practical problems that manifest as slightly inadequate numerical precision, that is difficult to improve in current implementations. In this talk I will describe progress in developing a new alternative local correlation approach that is fully numerical. Usability is promoted by use of just a single master threshold to determine the compute cost versus numerical precision tradeoff. Simplicity is promoted by the use of localized non-redundant virtual orbitals. Several new ideas are employed to achieve high efficiency, with a form of electronic embedding proving to be particularly important. Results will be presented that assess performance versus precision as a function of threshold for a range of standard quantum chemical methods beginning with MP2 and progressing to higher order methods.

About: Martin Head-Gordon is the Kenneth S. Pitzer Distinguished Professor, and Director of the Pitzer Center for Theoretical Chemistry at the University of California Berkeley. His research centers on development and application of quantum chemistry methods that are well-defined theoretical model chemistries, across density functional theory and wave function theory.

Kade Head-Marsden (University of Minnesota)

Electronic Structure for Molecular Spin Decoherence

Understanding spin dynamics is crucial to designing and optimizing molecular systems for use in emerging quantum technologies. At low temperatures, irreversible loss is facilitated by electronic-nuclear spin interactions. We develop combined open quantum systems and electronic structure theories capable of predicting trends in relaxation rates in molecular spins, and apply these theories to several molecular systems pertinent to contemporary quantum technologies. Our theories provide a framework to describe irreversible relaxation effects in molecular spin systems with applications in quantum information science, quantum sensing, molecular spintronics, and other spin systems whose dynamics are dominated by spin-spin relaxation.

About: Kade Head-Marsden is an assistant professor in the Department of Chemistry at the University of Minnesota, where her group's research focuses on open quantum systems, electronic structure, and quantum information. She has a BSc in mathematics and chemistry from McGill University, and a PhD in theoretical chemistry from the University of Chicago.

Joonho Lee (Harvard University)

Periodic Quantum Chemistry for Catalysis at Complex Interfaces

Catalysis at solid interfaces is shaped by surface structure, charge transfer, solvation, non-bonded interactions, and electron correlation. This talk will present recent advances in periodic quantum chemistry for modeling and interpreting these effects across catalytic materials. I will discuss grand-canonical DFT with implicit solvation for constant-potential electrochemical simulations, including applications to corrosion and electrochemical stability. I will also highlight a study of PtGa, where apparent topological electrocatalytic activity is instead traced to Ga corrosion and the formation of Pt-rich active sites under operating conditions.

Complementary developments in solid-state energy decomposition analysis and regularized perturbation theory will be presented as broader tools for chemical interpretation and improved accuracy in catalytic materials. Together, these studies point toward predictive, mechanism-focused simulations of catalysis at complex interfaces.

About: Joonho Lee, Assistant Professor of Chemistry and Chemical Biology at Harvard, is a quantum chemist developing scalable electronic-structure methods for complex materials and molecular systems. His research spans ab initio polarons, quantum algorithms, electron correlation methods, and energy decomposition analysis, with an emphasis on connecting rigorous theory to chemically interpretable mechanisms and high-performance computational applications.

Pierre-François Loos (CNRS and University of Toulouse)

An Algebraic-Diagrammatic Construction for Vertex Corrections to the GW Self-Energy

(Co-authors: Antoine Marie and Johannes Tölle)

In this talk, I will discuss recent efforts toward systematically improving Green's function methods beyond the widely used GW approximation. I will present a general strategy to go beyond GW that builds on the algebraic-diagrammatic construction (ADC) framework. This approach provides a controlled and physically motivated way to improve the electronic self-energy while preserving key analytical properties of the Green's function, such as the correct spectral structure. Rather than relying on straightforward perturbative expansions, the proposed framework can be viewed as a nonperturbative route to incorporate higher-order effects in a consistent and robust manner.

About: Pierre-François Loos is a CNRS researcher at the Laboratoire de Chimie et Physique Quantiques (Toulouse, France). His work focuses on the development of electronic structure methods for excited states, including Green's function approaches, coupled-cluster theory, selected configuration interaction, and density-functional theory.

Miguel Morales (Flatiron Institute)

Ab-Initio Quantum Embedding

I will present CoQui [1], a software package developed at the Flatiron Institute's Center for Computational Quantum Physics. The overarching goal of CoQui is the development and implementation of perturbative, low-scaling approaches to the many-body problem for first-principles electronic structure. CoQui relies on the use of Interpolative Separable Density Fitting to generate compact factorizations of the coulomb interaction, achieving an efficient framework for ab-initio many-body perturbation theory (MBPT). The package implements finite temperature variants of GW at various levels of self-consistency (single shot, quasi-particle and fully self-consistent), as well as several downfolding frameworks based on cRPA. It can be used to generate ab-initio low-energy models of complex materials and molecular systems. The resulting models can be combined with accurate many-body methods to provide new avenues to understand correlated phenomena. As a concrete example, I will discuss an implementation of GW+EDMFT obtained by interfacing CoQui with the TRIQS [2] software package. The new capabilities pave the way for the application of ab-initio quantum embedding on complex systems.

[1] CoQui: https://github.com/AbInitioQHub/coqui
[2] TRIQS: https://github.com/TRIQS

About: Miguel A. Morales Silva is a Research Scientist in the Center for Computational Quantum Physics at the Flatiron Institute. His research is focused on the development of numerical methods for the study of quantum many-body problems. His is mainly interested in predictive, parameter-free, first-principles methods for the study of strongly correlated materials.

James Shee (Rice University)

Subtleties and Surprises in Phaseless Auxiliary-Field Quantum Monte Carlo

Phaseless auxiliary-field quantum Monte Carlo (ph-AFQMC) is a nonperturbative, stochastic technique that has shown promise in capturing many-electron correlation effects in chemical systems. However, especially in the context of strongly correlated states as found, e.g., in active space models of synthetic polynuclear iron-sulfur clusters, there are a number of open questions surrounding the optimal choice and nuanced roles of the trial wavefunction. In this talk I will present a selection of findings from my group's ongoing exploration of various physically-transparent trial wavefunction forms ranging from spin symmetry-broken or symmetry-projected mean-field states to orbital-optimized Configuration State Functions. I will investigate the importance of (spin and orbital) angular momentum symmetries, and expose a peculiar sampling bias that arises when chemically-inadequate trial wavefunctions are employed.

About: James Shee is the Norman Hackerman-Welch Assistant Professor of Chemistry at Rice University. He was an NIH postdoctoral fellow at UC Berkeley with Martin Head-Gordon, and received his Ph.D. from Columbia University in 2019, co-advised by Richard Friesner and David Reichman. James graduated from Princeton University in 2014.

Ravishankar Sundararaman (Rensselaer Polytechnic Institute)

First-Principles Simulations of Spin Dynamics and Transport at Device Scales

The realization of practical spintronic and spin-based quantum information technologies hinges on the design of materials that exhibit long spin lifetimes and coherent spin transport. Computational design of such materials requires parameter-free first-principles techniques to quantitatively predict spin dynamics and transport across a broad range of electronic structures, symmetries and dimensionalities, accounting for both coherent dynamics and incoherent scattering processes. We present a general formalism based on Lindbladian dynamics of first-principles density matrices to accurately calculate the intrinsic spin-phonon relaxation time of bulk materials, providing a unified treatment of the Elliott-Yafet (EY) and D'yakonov-Perel' (DP) mechanisms of spin relaxation. In addition to T1 spin relaxation times, we also demonstrate predictions of magnetic-field-dependent T2 and T2* times, separating decoherence and dephasing ecects from first principles simulations of Hahn echo measurements. Finally, we extend this modeling capability to first principles density-matrix transport at device length scales within the Wigner function formalism. We elucidate the ecects of realistic spin-orbit fields, such as Rashba fields and persistent spin helices, and the strength of scattering on spin transport in several materials. Overall, this general capability to predict complex spin dynamics and transport phenomena in arbitrary materials opens up the possibility for designing materials for spin-based devices.

About: Ravishankar Sundararaman is a Professor of Materials Science and Engineering at Rensselaer Polytechnic Institute, following a PhD in Physics from Cornell in 2013, and a postdoctoral fellowship in Caltech till 2016. He develops quantum-classical techniques and software (JDFTx, QimPy) to extend the scale of first-principles simulations for materials design.

Edward Valeev (Virginia Tech)

Accelerating Exploration of Predictive Many-Body Methods

Emergence of practical reduced-complexity many-body methods, such as various forms of natural-orbital-compressed coupled-cluster variants, allow predictive treatment of electron-correlation for large systems with weakly-correlated ground states. The flip side of their spectacular successes is the formidable technical complexity which hinders their exploration and use for practical applications. Here I will report on our attempts to automate efficient implementation of such methods by a toolchain of symbolic and numerical tensor algebra tools, optionally assisted by agentic workflows.

About: Prof. Valeev received his MS in Chemistry from Higher Chemical College (Moscow, Russia) and PhD in Chemistry from the University of Georgia. He has been a faculty in the Department of Chemistry at Virginia Tech since 2006. His interests include quantum many-body problem, computational science, and applied mathematics. He lives in beautiful Roanoke, Virginia.

Steven R. White (UC Irvine)

Radial and Angular Gausslets

Gausslets basis sets are designed to turn continuum electronic-structure problems into something that looks much like a local lattice Hamiltonian, without giving up the advantages of a basis-set description. They are localized orthonormal functions with a special moment property that makes them integrate like delta functions, allowing a diagonal, two-index approximation for the electron electron interactions. They resemble DVR, sinc functions, or finite difference representations, but with a number of advantages.

In this talk I will first review the original gausslet idea and some of the older Cartesian constructions, including how one- and two-electron Coulomb operators can be represented in a simple real-space form. I will then discuss new radial and angular versions of gausslets. Radial gausslets work directly in the radial coordinate and produce a radial diagonal approximation, while resolve an arbitrary range of length scales associated with large Z. Angular gausslets add to this localized angular degrees of freedom, allowing the diagonal approximation to be applied in full.

About: Steven R. White is a Distinguished Professor at the University of California, Irvine. Inventor of the density matrix renormalization group, he works on both the simulation of strongly correlated condensed matter models and a few topics in quantum chemistry and electronic structure methods.

Yang Yang (University of Wisconsin-Madison)

Constrained Nuclear-Electronic Orbital Time-Dependent Density Functional Theory for Excited-State Simulations with Nuclear Quantum Effects

(Co-authors: Yiwen Wang, Zehua Chen, and Xi Xu)

Nuclear quantum effects (NQEs) can play important roles in electronically excited states, especially in processes involving proton motion, but they are not described in conventional excited-state electronic structure methods that treat nuclei classically. We present constrained nuclear-electronic orbital time-dependent density functional theory (CNEO-TDDFT), a new excited-state method that incorporates quantum nuclear delocalization effects directly into excited-state simulations. Built on the CNEO framework, this approach retains an intuitive geometry-based picture through constrained nuclear expectation positions while introducing NQEs into excited-state potential energy surfaces. Under a frozen nuclear orbital approximation, we derive working equations for excitation energies and analytic gradients, which enable excited-state geometry optimization and dynamics on quantum-corrected effective energy surfaces. We have implemented CNEO-TDDFT in PySCF and NWChem and benchmarked it for a series of gas-phase molecules, where it substantially improves excited-state vibrational frequencies relative to conventional TDDFT, particularly for modes involving hydrogen and deuterium motion, while maintaining similar computational cost. Preliminary applications to excited-state intramolecular proton transfer in HBQ also yield encouraging proton-transfer time scales, highlighting the promise of CNEO-TDDFT for photochemical processes in which NQEs are essential.

About: Yang Yang is an Assistant Professor of Chemistry at the University of Wisconsin–Madison. He develops multicomponent quantum methods for describing nuclear quantum effects in chemistry, particularly through the Constrained Nuclear-Electronic Orbital framework and its applications to spectroscopy, dynamics, and reactivity.

Dominika Zgid (University of Michigan and University of Warsaw)

Homotopy Continuation Method for Solving Dyson Equation Fully Self-Consistently: Theory and Application to NdNiO₂

The solution of the Dyson equation for the small-gap systems can be plagued by large non-converging iterations. In addition to the convergence issues, due to a high non-linearity, the Dyson equation may have multiple solutions. We apply the homotopy continuation approach to control the behavior of iterations. We used the homotopy continuation to locate multiple fully self-consistent GW solutions for the NdNiO₂ solid and to establish the corresponding Hartree–Fock limits. Some of the solutions found are qualitatively new and help to understand the nature of electron correlation in this material. We show that there are multiple low-energy charge-transfer solutions leading to the formation of charge-density waves. Our results qualitatively agree with the experimental conductivity measurements. To rationalize the structure of solutions, we compare the k-point occupations and generalize the concept of natural difference orbitals for correlated periodic solids.

About: Dominika Zgid, Professor of Chemistry and Physics at the University of Michigan and University of Warsaw, focuses on quantum many-body and Green's function methods for molecules and materials. Her group develops systematically improvable theoretical and computational approaches that bridge quantum chemistry, condensed matter physics, and materials science, enabling accurate predictions of complex molecular and solid-state systems' properties.

Hong-Zhou Ye (University of Maryland, College Park)

Correlated Wavefunction Methods for Molecular Solids: Chemically Accurate Lattice Energies and Beyond

In this talk, I will present applications of correlated wavefunction methods, up to the CCSD(T) level, for the accurate prediction of energetic and structural properties of molecular solids under periodic boundary conditions. In the first part, I will focus on the methodological developments needed to converge lattice energies to chemical accuracy (within 1 kcal/mol), with a primary focus on addressing the slowly decaying finite-size errors for approaching the thermodynamic limit. This level of precision enables rigorous benchmarking of a hierarchy of correlated wavefunction methods against previous best theoretical estimates and experimental references across extensive molecular-crystal datasets, including X23, ICE13, and BMCOS, which span a diverse range of systems. In the second part, I will discuss how the accuracy and computational efficiency of our local coupled-cluster framework can be extended beyond lattice energies to predict structural properties of molecular solids, including equilibrium cell volumes and bulk moduli. I will also show that geometry optimization can sometimes have a significant impact on theory–experiment comparisons of lattice energies, and examine this effect at the coupled-cluster level for the first time.

About: Hong-Zhou Ye is an Assistant Professor at the University of Maryland, College Park. His research focuses on developing scalable correlated electronic-structure methods and software for predictive quantum simulations of molecules and materials, with applications in molecular solids, heterogeneous catalysis, and energy-related materials.

Tianyu Zhu (Yale University)

Quantum Embedding for Coupled-Cluster Band Structures and Metal-Surface Adsorption

Accurate many-body predictions for extended systems remain challenging because they require a balanced treatment of local and nonlocal electron correlation in the thermodynamic limit. In this talk, I will present two ab initio quantum embedding frameworks that address these challenges in complementary settings. First, I will introduce a Green's function embedding approach, interacting-bath dynamical embedding theory, for equation-of-motion coupled-cluster (EOM-CC) calculations of band structures in solids. This framework enables substantially denser k-point sampling than previous studies and controlled convergence toward the thermodynamic limit, providing a practical route to benchmark band gaps or bandwidths in semiconductors, insulators, and simple metals at the EOM-CCSD level. Second, I will present a density-matrix-based embedding framework for molecular adsorption on metallic surfaces, designed to make coupled-cluster calculations feasible for molecule–metal interfaces with fractional occupancies. I will show how this approach accelerates convergence toward the full-system limit and opens the door to systematically improvable studies of adsorption energetics at neutral and charged metal surfaces. Together, these developments illustrate how quantum embedding extends high-level electronic structure theory to challenging problems in materials science and catalysis.

About: Tianyu Zhu is an Assistant Professor of Chemistry at Yale University. His research focuses on developing electronic structure theory, quantum chemistry software, and machine learning methods for solid-state materials and molecule-solid interfaces. He received his PhD from MIT and completed postdoctoral research at Caltech.