paper / markusreiher / 10d ago
Computational chemistry heavily relies on potential energy hypersurfaces, traditionally computed via DFT, which is computationally expensive. A new class of machine learning interatomic potentials offers quantum accuracy at force-field speeds. Recent advancements, specifically "foundation machine learning interatomic potentials," have overcome the prior limitation of requiring large, system-specific training datasets, positioning them to replace DFT as the primary method for computational chemistry within a decade.
computational-chemistrymachine-learning-interatomic-potentialsdensity-functional-theoryai-accelerated-discoverymaterials-science
“Computational chemistry is greatly dependent on potential energy surfaces.”
paper / markusreiher / 23d ago
To achieve practical advantage, quantum chemistry algorithms must move beyond solving isolated, strongly correlated molecular cases and instead integrate into high-throughput computational pipelines. The window for quantum advantage is narrowing due to concurrent advancements in classical wavefunction-theory methods, necessitating hardware-aware algorithm development.
quantum-computingcomputational-chemistrymaterials-sciencequantum-algorithmshigh-throughput-screening
“Classical wavefunction-theory methods are reducing the potential window for quantum advantage in chemistry.”
paper / markusreiher / Feb 21
This paper investigates the impact of orbital transformations on Trotter error in quantum phase estimation (QPE) for molecular calculations. The authors explore different strategies to mitigate Trotter error through orbital basis selection and dynamic basis changes. A key finding challenges prior assumptions by demonstrating that localized orbitals do not inherently lead to large Trotter errors, contradicting existing literature.
quantum-computationtrotter-errororbital-transformationsquantum-phase-estimationmolecular-calculationsquantum-algorithms
“Trotter error is a continuous function of the Givens-rotation parameter, implying its continuity upon orbital transformation.”
paper / markusreiher / Feb 13
Neural Quantum States (NQS) offer a flexible parameterization for wave functions in strongly correlated quantum chemistry. This study compares two NQS-based approaches, Variational Monte Carlo (NQS-VMC) and Selected Configuration (NQS-SC), for electronic ground state optimizations. NQS-SC demonstrates superior accuracy and robustness, particularly for statically correlated molecules, but both methods struggle with dynamical correlation.
neural-quantum-statesquantum-chemistryvariational-monte-carloelectron-correlationcomputational-physicsnqs-sc
“Neural Quantum States (NQS) provide a flexible and highly expressive parameterization of wave functions for strongly correlated problems in quantum chemistry.”
paper / markusreiher / Nov 6
The paper introduces a novel approach for simulating photochemical processes by quantizing n-mode anharmonic vibronic Hamiltonian terms. This framework integrates a second-quantized perspective with the density matrix renormalization group algorithm, enabling accurate and reliable quantum dynamics calculations for vibronic systems. The method demonstrates improved accuracy in modeling anharmonic effects and off-diagonal nonadiabatic coupling, crucial for understanding spectral features and excited state dynamics.
quantum-dynamicsvibronic-hamiltoniansmatrix-product-statedensity-matrix-renormalization-groupphotochemical-processesmolecular-modeling
“Accurate quantum dynamics calculations for vibronic systems necessitate precise modeling of anharmonic effects and off-diagonal nonadiabatic coupling terms.”