Projects
Disclaimer: only up to date as of 12/2025
Disclaimer: only up to date as of 12/2025
We just published some work on an agent for quantum computing simulations - my work was to suggest some problems for the agent to solve (https://arxiv.org/abs/2512.18847)
Some QC Ware work was not published as it was part of customer contracts.
In collaboration with researchers at POSCO Holdings, we applied quantum algorithms to Lithium diffusion in solid state electrolytes (http://dx.doi.org/10.1109/QCE60285.2024.00082) and to carbon capture in metal-organic frameworks (https://arxiv.org/abs/2510.02550).
In collaboration with researchers at Boehringer-Ingelheim, we investigated how orbital optimization could improve the overlap of the initial state with the ground state without consuming quantum resources (https://doi.org/10.1103/PhysRevLett.133.250601).
In collaboration with researchers at Boehringer-Ingelheim, we designed a new method for factorizing Hamiltonians that improved upon the existing state of the art by leveraging symmetry shifting in combination with compressed double factorization techniques (https://doi.org/10.1021/acs.jctc.4c00352). We also investigated the theoretical bounds of symmetry shifting techniques and the maximum speedup achievable for the final QPE algorithm (https://doi.org/10.1103/PhysRevA.110.022420).
Some of the work at Zapata was performed within customer contracts and was not published at the time. Overall, the work of my team focused on near-term algorithms, mostly with applications in chemistry.
In collaboration with my colleagues, I worked on two projects to reduce the number of measurements needed. In the first, we evaluated Robust Amplitude Estimation (designed at Zapata Computing) on chemistry problems and verified that it yields better dependence of the number of measurements on precision at the cost of deeper circuits (https://arxiv.org/abs/2410.00686). In the second, we worked to reduce the cost of measurements by learning a surrogate model for the wavefunction in the form of a neural network, trained on previous measurement outcomes. We did not invent that method but we worked to characterize the measurement efficiency for molecular ground states (https://dx.doi.org/10.1088/2632-2153/acb4df).
My first project was to evaluate the cost of running VQE on simple molecules. With my colleagues, we quickly discovered that the number of measurements required to obtain accurate energies was prohibitive. We evaluated different measurement techniques for small molecules and extrapolated our results to molecules of larger size, showing how using FNOs could result in better scaling when combined with efficient measurement grouping. Overall, the cost of measuring the energy is still excessive (https://doi.org/10.1103/PhysRevResearch.4.033154).
I examined the nature of dispersion interactions by analyzing the CCSD T2 amplitudes in a local basis. Singular Value Decomposition of these amplitudes yields a very compact representation, from which I extract virtual orbitals specifically tailored for dispersion. These orbitals are connected to the multipole expansion of dispersion and represent dipolar-dipolar, dipolar-quadrupolar and quadrupolar-quadrupolar correlations. This work is related to the PNO technique and opens up new ways to efficiently represent dispersion. (10.1063/1.4997186)
The analyses tools were implemented as object-oriented C++ code in the commercial electronic structure software Q-Chem.
In a collaboration with Ivano Tavernelli, technical leader of theoretical quantum computing at IBM Research, Zurich. Optimization of Unitary Coupled-Cluster for Variational Quantum Eigensolver, submitted for publication.
I designed and supervised the research project on which Pauline Ollitrault worked for 6 months before obtaining her Master thesis. Pauline is currenly a PhD candidate at IBM Research laboratories in Zurich, Switzerland.
More to come after publication, collaborative project with Head-Gordon group members and Q-Chem employees to write several thousand lines of C++ code.
IR spectroscopic data from Prof. Bridgette Barry and Udita Brahmachari, supported by simulated IR spectra that I computed, indicate that a protonated water cluster near a chloride ion is involved in an internal proton transfer step in the S1 to S2 transition of the oxygen evolving complex in Photosystem II (10.1021/acs.jpcb.7b08358 and 10.1021/acs.jpcb.9b01523).
I generalized the very efficient closed-shell SAPT code to open-shell cases in the software Psi4 and performed the then largest open-shell SAPT computation to date, examining the change in pi-stacking interactions of a base pair ladder upon oxidation (10.1063/1.4963385). To facilitate these computations, I also improved considerably the efficiency of the UHF stability analysis in Psi4.
I rewrote the disk-based Fock builder in Psi4 to minimize disk access and use batching of integral computation over multiple threads (OpenMP). Each thread now computes integrals in specialized buffers, and one separate thread is performing asynchronous write operations in the appropriate file as results become available. I also added a fully in-code algorithm when enough memory is available. The various algorithms all derive from the same base class for convenience.
To make sure the implementation of open-shell SAPT was correct, since unfortunately most published formulae have typos, I re-derived all terms in open-shell SAPT0 from scratch. The full document is available as a pdf.
In collaboration with Rob Parrish, we published a mathematically simple version of intramolecular SAPT that is based on HF-in-HF embedding. All terms in SAPT0 are available in the intramolecular version since only the zeroth-order wavefunction is different from the intermolecular case (10.1063/1.4927575).
My early work on intramolecular SAPT was devoted to designing a reliable zeroth-order wavefunction that would exclude the non-covalent interactions of interest while still describing all other interactions in a molecule. I implemented in Fortran 90 an approach based on the Chemical Hamiltonian (10.1063/1.4871116) which was later used to derive the necessary SAPT0 energy terms (10.1063/1.4936830).
I participated in the writing of a review summarizing the quantification of a number of chemical concepts. Often, several different methods are available for quantification, for example to compute bond orders and partial charges. We discussed the pros and cons of the most common methods (10.1039/C2CS35037H).
I also designed an index to characterize the propensity of a planar molecule for pi-stacking interactions based on the LOL isosurface for the electrons (10.1039/C2CC33886F).
I collaborated with Dr. Matthew Wodrich and the Cramer experimental group on a Rh(III)-catalyzed dihydroisoquinolone synthesis. Specifically, I wrote and solved the system of differential equations describing the entire kinetic network of the computed catalytic cycle (10.1002/chem.201404515).
I worked in direct collaboration with Dr. Astrid J. Olaya and the Girault group on the metal-free four-electron reduction of dioxygen catalyzed by tetrathiafulvalene (TTF). My computations revealed that a TTF quadruplex likely played the role of the catalyst in this reaction (10.1021/ja203251u).
I modeled the enthalpy of formation of regular, methylated and permethylated linear alkanes by a simple increment system. Branched alkanes are generally known to be more stable than their linear counterparts, however this is not true beyond some degree of branching where steric repulsions start to dominate the interactions (10.1021/ol1010642).
I collaborated with Dr. Matthew Wodrich to investigate the strain in carbomeric cycloalkanes (10.1021/jp1029322) and the thermochemistry of hydrocarbon radicals (10.1021/jp3061653).