Federico Zahariev
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Federico Zahariev has a total of 58 co-authors. Here are the top co-authors by number of shared publications:
- Mark S. Gordon
Department of Chemistry and Ames National Laboratory
9 shared publications - Mel Levy
Department of Chemistry and Quantum Theory Group, Tulane University, New Orleans, Louisiana 70118, USA
8 shared publications - Marilú Pérez García
Department of Chemistry and Ames Laboratory, Iowa State University , Ames, Iowa 50011, United States
3 shared publications - Theresa L. Windus
Ames National Laboratory
2 shared publications Alistair P. Rendell
Research School of Computer Science, Australian National University, Canberra, ACT 2601, Australia
1 shared publicationAmr Sabry
1 shared publication
Latest Publications
- Open Access
- 4 November 2026
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 28 (42) , 26135-26148
RANGE-X: a gradient-free framework for the discovery of conical intersections and photochemical reaction pathways
RANGE-X is a fully automated, gradient-free and method agnostic framework for locating conical intersections between any states of interest at reduced computational cost. ML post-processing refines the location of transition states.
- 4 November 2026
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 28 (42) , 26135-26148
Open AccessRANGE-X: a gradient-free framework for the discovery of conical intersections and photochemical reaction pathways
RANGE-X is a fully automated, gradient-free and method agnostic framework for locating conical intersections between any states of interest at reduced computational cost. ML post-processing refines the location of transition states.
- Open Access
- 9 September 2026
- Published by Elsevier in SSRN Electronic Journal
Scaling Local Solvers with the Many-Body Expansion
We introduce a many-body expansion (MBE) framework for scaling black-box solvers for discrete optimization, drawing on the organizing logic of fragment-based methods in computational chemistry. The method decomposes a problem into fragments, repeatedly improves each fragment with the rest held fixed, and applies targeted corrections on unions of fragments; the solver is unchanged and receives only restricted subproblems. Mobius inversion gives an exact fragment expansion of the objective, closing at second order for pairwise objectives such as graph coloring, MAX-CUT, and quadratic unconstrained binary optimization (QUBO). The promoted solver is monotone by construction, terminates finitely for non-negative integer-valued objectives, and recovers junction-tree dynamic programming on acyclic structures. While the framework follows the broad decompose-solve-recombine pattern of classical decomposition and neighborhood-search methods, its distinctive contribution is the certified accounting layered onto that pattern: it identifies which fragment unions to revisit, orders those corrections, and supplies guarantees while treating the underlying solver as an arbitrary black box. Comparing each solver in flat and promoted form under identical budgets isolates the wrapper's contribution. On graph coloring and MAX-CUT benchmarks, the promoted versions of TabuSearch and simulated annealing (SA) generally outperform their flat counterparts, often in less computation time, with the gain insensitive to the graph partition. At the best-known chromatic numbers of the DIMACS instances, promotion removes 33-54% of the conflicts left by the flat solve. On facility location the partition must respect the objective's interaction locality, and the promoted solver reaches within 0.3-1.4% of globally optimal values certified by an exact solver.
- 9 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessScaling Local Solvers with the Many-Body Expansion
We introduce a many-body expansion (MBE) framework for scaling black-box solvers for discrete optimization, drawing on the organizing logic of fragment-based methods in computational chemistry. The method decomposes a problem into fragments, repeatedly improves each fragment with the rest held fixed, and applies targeted corrections on unions of fragments; the solver is unchanged and receives only restricted subproblems. Mobius inversion gives an exact fragment expansion of the objective, closing at second order for pairwise objectives such as graph coloring, MAX-CUT, and quadratic unconstrained binary optimization (QUBO). The promoted solver is monotone by construction, terminates finitely for non-negative integer-valued objectives, and recovers junction-tree dynamic programming on acyclic structures. While the framework follows the broad decompose-solve-recombine pattern of classical decomposition and neighborhood-search methods, its distinctive contribution is the certified accounting layered onto that pattern: it identifies which fragment unions to revisit, orders those corrections, and supplies guarantees while treating the underlying solver as an arbitrary black box. Comparing each solver in flat and promoted form under identical budgets isolates the wrapper's contribution. On graph coloring and MAX-CUT benchmarks, the promoted versions of TabuSearch and simulated annealing (SA) generally outperform their flat counterparts, often in less computation time, with the gain insensitive to the graph partition. At the best-known chromatic numbers of the DIMACS instances, promotion removes 33-54% of the conflicts left by the flat solve. On facility location the partition must respect the objective's interaction locality, and the promoted solver reaches within 0.3-1.4% of globally optimal values certified by an exact solver.
- 12 January 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 22 (2) , 848-858
Solvent Effects on Nonadiabatic Coupling: Interfacing Time-Dependent Density Functional Theory with the Effective Fragment Potential Method
To enable the accurate prediction of solvent effects on nonadiabatic processes, time-dependent density functional theory (TDDFT) is combined with the effective fragment potential (EFP) method for the computation of nonadiabatic coupling matrix elements (NACME). The viability of the new TDDFT/EFP NACME method is demonstrated via comparisons to fully TDDFT NACME calculations for both the solute (methylene imine) and solvent (water or methanol). The success of the combined TDDFT/EFP method is promising for the use of the new method for accurate simulations of the nonadiabatic dynamics of solvated molecules.
- 12 January 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 22 (2) , 848-858
Solvent Effects on Nonadiabatic Coupling: Interfacing Time-Dependent Density Functional Theory with the Effective Fragment Potential Method
To enable the accurate prediction of solvent effects on nonadiabatic processes, time-dependent density functional theory (TDDFT) is combined with the effective fragment potential (EFP) method for the computation of nonadiabatic coupling matrix elements (NACME). The viability of the new TDDFT/EFP NACME method is demonstrated via comparisons to fully TDDFT NACME calculations for both the solute (methylene imine) and solvent (water or methanol). The success of the combined TDDFT/EFP method is promising for the use of the new method for accurate simulations of the nonadiabatic dynamics of solvated molecules.
- Open Access
- 31 October 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (21) , 11532-11542
A Comprehensive Machine Learning Model for Metal–Ligand Binding Prediction: Applications in Chemistry and Biology
A machine-learning (ML) model that predicts metal-ligand binding constants was developed using the open-source Chemprop software. The model was trained on over 30,000 experimental log K1 values, which include both protonation and metal-ligand stability constants, comprising over 3500 ligands and 102 metal ions from 73 total elements, thus generalizing beyond existing limited approaches, which focus only on specific metals or ligand families. The best-performing model included a combination of SMILES-based molecular representations along with descriptors for the metal ion and experimental conditions. It had an external test R2 value of 0.942, and MAE value of 0.834. A "SMILES-only" simpler version also produced accurate predictions and preserved the binding trends, serving as a quick and easily accessible alternative for users without computational expertise. The SMILES-only model performed comparably to density functional theory (DFT) calculations but utilized a fraction of the computational resources. The model was successfully applied across diverse domains, including bioinorganic chemistry, heavy metal remediation, and sensor development and demonstrated its effectiveness as a rapid and reliable screening tool for both academic and industrial uses.
- 31 October 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (21) , 11532-11542
Open AccessA Comprehensive Machine Learning Model for Metal–Ligand Binding Prediction: Applications in Chemistry and Biology
A machine-learning (ML) model that predicts metal-ligand binding constants was developed using the open-source Chemprop software. The model was trained on over 30,000 experimental log K1 values, which include both protonation and metal-ligand stability constants, comprising over 3500 ligands and 102 metal ions from 73 total elements, thus generalizing beyond existing limited approaches, which focus only on specific metals or ligand families. The best-performing model included a combination of SMILES-based molecular representations along with descriptors for the metal ion and experimental conditions. It had an external test R2 value of 0.942, and MAE value of 0.834. A "SMILES-only" simpler version also produced accurate predictions and preserved the binding trends, serving as a quick and easily accessible alternative for users without computational expertise. The SMILES-only model performed comparably to density functional theory (DFT) calculations but utilized a fraction of the computational resources. The model was successfully applied across diverse domains, including bioinorganic chemistry, heavy metal remediation, and sensor development and demonstrated its effectiveness as a rapid and reliable screening tool for both academic and industrial uses.
- Open Access
Quantum Simulation Programming via Typing
Quantum simulations are designed to model quantum systems, and many compilation frameworks have been developed for executing such simulations on quantum computers. Most compilers leverage the capabilities of digital and analog quantum computers by representing quantum particle systems with Pauli strings or digital quantum circuits, making it challenging for users in physics, chemistry, and biology to program simulations effectively. QBLUE is proposed as the first programming language for describing the behaviors of quantum systems in terms of second quantization Hamiltonians. Within QBLUE, a novel type system is proposed to clearly define states across different quantum systems and treat quantum computers as quantum particle systems of specific types. The type system is compatible with the compilation of quantum simulations expressed in QBLUE for digital and analog quantum computers. With QBLUE, users can specify the desired quantum particle system and model the system on quantum computers.
Open AccessQuantum Simulation Programming via Typing
Quantum simulations are designed to model quantum systems, and many compilation frameworks have been developed for executing such simulations on quantum computers. Most compilers leverage the capabilities of digital and analog quantum computers by representing quantum particle systems with Pauli strings or digital quantum circuits, making it challenging for users in physics, chemistry, and biology to program simulations effectively. QBLUE is proposed as the first programming language for describing the behaviors of quantum systems in terms of second quantization Hamiltonians. Within QBLUE, a novel type system is proposed to clearly define states across different quantum systems and treat quantum computers as quantum particle systems of specific types. The type system is compatible with the compilation of quantum simulations expressed in QBLUE for digital and analog quantum computers. With QBLUE, users can specify the desired quantum particle system and model the system on quantum computers.
- Open Access
- 28 January 2024
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 160 (4)
Prediction of stability constants of metal–ligand complexes by machine learning for the design of ligands with optimal metal ion selectivity
The new LOGKPREDICT program integrates HostDesigner molecular design software with the machine learning (ML) program Chemprop. By supplying HostDesigner with predicted log K values, LOGKPREDICT enhances the computer-aided molecular design process by ranking ligands directly by metal–ligand binding strength. Harnessing reliable experimental data from a historic National Institute of Standards and Technology (NIST) database and data from the International Union of Pure and Applied Chemistry (IUPAC), we train message passing neural net algorithms. The multi-metal NIST-based ML model has a root mean square error (RMSE) of 0.629 ± 0.044 (R2 of 0.960 ± 0.006), while two versions of lanthanide-only IUPAC-based ML models have, respectively, RMSE of 0.764 ± 0.073 (R2 of 0.976 ± 0.005) and 0.757 ± 0.071 (R2 of 0.959 ± 0.007). For relative log K predictions on an out-of-sample set of six ligands, demonstrating metal ion selectivity, the RMSE value reaches a commendably low 0.25. We showcase the use of LOGKPREDICT in identifying ligands with high selectivity for lanthanides in aqueous solutions, a finding supported by recent experimental evidence. We also predict new ligands yet to be verified experimentally. Therefore, our ML models implemented through LOGKPREDICT and interfaced with the ligand design software HostDesigner pave the way for designing new ligands with predetermined selectivity for competing metal ions in an aqueous solution.
- 28 January 2024
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 160 (4)
Open AccessPrediction of stability constants of metal–ligand complexes by machine learning for the design of ligands with optimal metal ion selectivity
The new LOGKPREDICT program integrates HostDesigner molecular design software with the machine learning (ML) program Chemprop. By supplying HostDesigner with predicted log K values, LOGKPREDICT enhances the computer-aided molecular design process by ranking ligands directly by metal–ligand binding strength. Harnessing reliable experimental data from a historic National Institute of Standards and Technology (NIST) database and data from the International Union of Pure and Applied Chemistry (IUPAC), we train message passing neural net algorithms. The multi-metal NIST-based ML model has a root mean square error (RMSE) of 0.629 ± 0.044 (R2 of 0.960 ± 0.006), while two versions of lanthanide-only IUPAC-based ML models have, respectively, RMSE of 0.764 ± 0.073 (R2 of 0.976 ± 0.005) and 0.757 ± 0.071 (R2 of 0.959 ± 0.007). For relative log K predictions on an out-of-sample set of six ligands, demonstrating metal ion selectivity, the RMSE value reaches a commendably low 0.25. We showcase the use of LOGKPREDICT in identifying ligands with high selectivity for lanthanides in aqueous solutions, a finding supported by recent experimental evidence. We also predict new ligands yet to be verified experimentally. Therefore, our ML models implemented through LOGKPREDICT and interfaced with the ligand design software HostDesigner pave the way for designing new ligands with predetermined selectivity for competing metal ions in an aqueous solution.
- 16 August 2021
- Published by American Physical Society (APS) in Physical Review A
- Vol. 104 (2) , 022815
Energy components in spin-density functional theory
It has recently been shown how to replace spin-wave functions with purely spatial wave functions in the formulation of spin-density functional theory (SDFT), without any change to the resultant density functional expressions [F. Zahariev and M. Levy, Phys. Rev. A 100, 062507 (2019)]. The purely spatial wave functions that are obtained are much more convenient to use and thus allow for a relatively easy decomposition and manipulation of terms in the constrained-search formulation of SDFT. All the essential ingredients of SDFT, including the kinetic, exchange, and correlation contributions, are explicitly defined in terms of the universal functional expressed in this spin-free manner. Constrained-search derivations of the Oliver-Perdew relations for the kinetic and exchange terms are presented. The up-down spin component of the correlation term is found to contain the up-down spin component of the Hartree term. A spin-dependent generalization of the adiabatic connection is put forth and a connection with spin-dependent coordinate scaling is established for the correlation energy.
- 16 August 2021
- Published by American Physical Society (APS) in Physical Review A
- Vol. 104 (2) , 022815
Energy components in spin-density functional theory
It has recently been shown how to replace spin-wave functions with purely spatial wave functions in the formulation of spin-density functional theory (SDFT), without any change to the resultant density functional expressions [F. Zahariev and M. Levy, Phys. Rev. A 100, 062507 (2019)]. The purely spatial wave functions that are obtained are much more convenient to use and thus allow for a relatively easy decomposition and manipulation of terms in the constrained-search formulation of SDFT. All the essential ingredients of SDFT, including the kinetic, exchange, and correlation contributions, are explicitly defined in terms of the universal functional expressed in this spin-free manner. Constrained-search derivations of the Oliver-Perdew relations for the kinetic and exchange terms are presented. The up-down spin component of the correlation term is found to contain the up-down spin component of the Hartree term. A spin-dependent generalization of the adiabatic connection is put forth and a connection with spin-dependent coordinate scaling is established for the correlation energy.
- 15 June 2021
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 23 (26) , 14308-14314
Combined quantum Monte Carlo – effective fragment molecular orbital method: fragmentation across covalent bonds
The previously developed combined Quantum Monte Carlo-Effective Fragment Molecular Orbital (QMC-EFMO) method is extended to systems in which the fragmentation process cuts across covalent molecular bonds. The extended QMC-EFMO capability is demonstrated on a few model systems: the glycine tetramer, the diglycine reaction to form a dipeptide, silica-based rings, and polyalanine chains of increasing length. The agreement between full QMC and QMC-EFMO for the correlation energy is within 2 kcal mol−1 and for the correlation energy differences is within 1 kcal mol−1.
- 15 June 2021
- Published by Royal Society of Chemistry (RSC) in Physical Chemistry Chemical Physics
- Vol. 23 (26) , 14308-14314
Combined quantum Monte Carlo – effective fragment molecular orbital method: fragmentation across covalent bonds
The previously developed combined Quantum Monte Carlo-Effective Fragment Molecular Orbital (QMC-EFMO) method is extended to systems in which the fragmentation process cuts across covalent molecular bonds. The extended QMC-EFMO capability is demonstrated on a few model systems: the glycine tetramer, the diglycine reaction to form a dipeptide, silica-based rings, and polyalanine chains of increasing length. The agreement between full QMC and QMC-EFMO for the correlation energy is within 2 kcal mol−1 and for the correlation energy differences is within 1 kcal mol−1.
- Top Cited PapersOpen Access
- 16 April 2020
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 152 (15) , 154102
Recent developments in the general atomic and molecular electronic structure system
A discussion of many of the recently implemented features of GAMESS (General Atomic and Molecular Electronic Structure System) and LibCChem (the C++ CPU/GPU library associated with GAMESS) is presented. These features include fragmentation methods such as the fragment molecular orbital, effective fragment potential and effective fragment molecular orbital methods, hybrid MPI/OpenMP approaches to Hartree–Fock, and resolution of the identity second order perturbation theory. Many new coupled cluster theory methods have been implemented in GAMESS, as have multiple levels of density functional/tight binding theory. The role of accelerators, especially graphical processing units, is discussed in the context of the new features of LibCChem, as it is the associated problem of power consumption as the power of computers increases dramatically. The process by which a complex program suite such as GAMESS is maintained and developed is considered. Future developments are briefly summarized.
- 16 April 2020
- Published by AIP Publishing in The Journal of Chemical Physics
- Vol. 152 (15) , 154102
Top Cited PapersOpen AccessRecent developments in the general atomic and molecular electronic structure system
A discussion of many of the recently implemented features of GAMESS (General Atomic and Molecular Electronic Structure System) and LibCChem (the C++ CPU/GPU library associated with GAMESS) is presented. These features include fragmentation methods such as the fragment molecular orbital, effective fragment potential and effective fragment molecular orbital methods, hybrid MPI/OpenMP approaches to Hartree–Fock, and resolution of the identity second order perturbation theory. Many new coupled cluster theory methods have been implemented in GAMESS, as have multiple levels of density functional/tight binding theory. The role of accelerators, especially graphical processing units, is discussed in the context of the new features of LibCChem, as it is the associated problem of power consumption as the power of computers increases dramatically. The process by which a complex program suite such as GAMESS is maintained and developed is considered. Future developments are briefly summarized.
- 13 December 2019
- Published by American Physical Society (APS) in Physical Review A
- Vol. 100 (6) , 062507
Density- and spin-density-functional theories through spin-free wave functions
It is proven that entirely spin-free wave functions can be utilized in the basic definitions of the universal functionals in density-functional theory and spin-density-functional theory. Then, for the purpose of approximating these functionals, it is shown that the knowledge of certain properties of the functionals, such as those involving the coordinate scaling of just one of the spin densities, is made feasible precisely because the wave functions in the functional definitions are free of explicit spins. Among the examples, a conjectured spin-density scaling relation for the correlation energy is studied.
- 13 December 2019
- Published by American Physical Society (APS) in Physical Review A
- Vol. 100 (6) , 062507
Density- and spin-density-functional theories through spin-free wave functions
It is proven that entirely spin-free wave functions can be utilized in the basic definitions of the universal functionals in density-functional theory and spin-density-functional theory. Then, for the purpose of approximating these functionals, it is shown that the knowledge of certain properties of the functionals, such as those involving the coordinate scaling of just one of the spin densities, is made feasible precisely because the wave functions in the functional definitions are free of explicit spins. Among the examples, a conjectured spin-density scaling relation for the correlation energy is studied.