Greater Boston Area Theoretical Chemistry Lecture Series

2021-2022 Speaker Schedule

Nucleosomes as Liquid-Like Organisers of Chromatin Organization

10/06/21 10:00 am EST

Rosana Collepardo-Guevara

University of Cambridge, Cambridge, UK


Liquid-liquid phase separation (LLPS) is an important mechanism that helps explain the membraneless compartmentalization of the Cell nucleus. Because chromatin phase separation is a collective phenomena (i.e., emerging from dynamic formation of thousands of molecular interactions), linking biophysical features of individual nucleosomes to LLPS modulation remains an open challenge. In this talk, I will discuss our new multiscale chromatin approach—integrating atomistic representations of DNA and proteins, a chemically-specific coarse-grained model of oligonucleosomes, and a minimal model of chromatin—that can resolve individual nucleosomes within sub-Mb chromatin domains and phase-separated systems (Farr et al. Nat Communs 2021). I will also discuss how using this model, we find that nucleosome thermal fluctuations, which become significant at physiological salt concentrations, destabilize the 30-nm fiber. In its place, nucleosome breathing favours stochastic folding of chromatin into a liquid-like ensemble. I will also discuss why nucleosome breathing also promotes the intrinsic LLPS of chromatin by simultaneously boosting the transient nature and heterogeneity of nucleosome-nucleosome contacts and the effective nucleosome valency. Our work highlights how the plasticity of nucleosomes is a key element in the liquid-like behaviour of chromatin, and the regulation of chromatin LLPS.

Systematic Coarse-Graining and Non-Markovian Modeling: From Molecular Vibrations to Chemical Reactions and Protein Folding

1/26/22 2:00 pm EST

Roland Netz

Freie Universität Berlin, Berlin, Germany


Most interesting physical systems are interacting many-body systems. When dealing with the kinetics of such systems, one is typically interested in the dynamics of a low-dimensional reaction coordinate, which in general is influenced by the entire system. The dynamics of the reaction coordinate is governed by the generalized Langevin equation (GLE), an integro-differential stochastic equation, and involves a memory function, which describes how the dynamics depends on previous values of the reaction coordinate. The GLE is thus an intrinsically non-Markovian description of the dynamics of a system in terms of coarse-grained variables. In the first part of my lecture I will review the Zwanzig and Mori projection techniques that are used to derive the GLE from the Hamiltonian mechanics of a general interacting many-body systems. In the second part of my lecture I will treat specific examples: alpha-helix forming in polypeptides, the infrared absorption spectra of molecular liquids, and chemical reactions in solution.

New Quantum Chemistry Methodology: Quantum Embedding and Machine Learning Algorithms for Complex Systems

2/9/22 11:00 am EST

Jason Goodpaster

University of Minnesota, Minneapolis, Mn


Large, condensed phase, and extended systems impose a challenge for theoretical studies due to the compromise between accuracy and computational cost in their calculations. We present two methods that show exciting promise for treating this compromise: machine learning and quantum embedding. We exploit machine learning methods to solve this accuracy and computational cost trade-off by leveraging large data sets to train on highly accurate calculations using small molecules and then apply them to larger systems. We are developing a method to train a neural network potential with high-level wavefunction theory on targeted systems of interest that are able to describe bond breaking. We combine density functional theory calculations and higher level ab initio wavefunction calculations, such as CASPT2, to train our neural network potentials. We first train our neural network at the DFT level of theory. Using an adaptive active learning training scheme, we retrained the neural network potential to a CASPT2 level of accuracy. Quantum embedding methodology exploits the locality of chemical interactions to allow for accurate yet computationally efficient calculations to be performed on complex systems. Quantum embedding allows for the partitioning of the system into two regions. One is treated at a highly accurate level of theory using wave function theory methods, and the other is treated at the more computationally efficient level of DFT. We discuss our recent advancements for quantum embedding, specifically for systems with complicated electronic structure such as homogeneous and heterogeneous catalysts. Together, we believe both methodologies can allow for complex systems to be studied at a significantly reduced computational cost.

Quantum Vacuum in Quantum Chemistry and vice versa

3/2/22 10:00 am EST

Alexandre Tkatchenko

University of Luxembourg, Luxembourg


The quantum nature of the vacuum fields (beyond the "classical" Coulomb potential) leads to observable effects, for example the spontaneous attraction between neutral objects (van der Waals and Casimir interactions), the Lamb shift of atomic energy levels, or the Aharonov-Bohm effect. In this talk, I will discuss notable effects of quantum vacuum on intermolecular interactions [1,2], showing how field theory approaches can be used to extend the applicability of quantum methods to molecular systems with tens of thousands of atoms [3,4]. I will also show how chemists' fascination for computing accurate numbers can be useful for estimating the intrinsic self-interaction energy density of quantum fields [5,6]. I will conclude by discussing how quantum field theory approaches and quantum chemistry can and should evolve in synergy.

[5], Phys. Rev. Lett. (2022).
[6] Tkatchenko and Fedorov, submitted.

Catalyst Discovery for Metal-free, Photoredox CO2 Reduction

3/16/22 3:30-5:00 pm EST

Shaama Sharada

University of Southern California, Los Angeles, CA


Organic photoredox catalysis is an important component of an energy-efficient, sustainable future as these catalysts can access highly reactive states upon excitation and quenching to carry out reactions that are otherwise thermally inaccessible or energy-intensive. Our group aims to identify sustainable photoredox routes for CO2 utilization. Prior experiments show that a simple organic chromophore, p-terphenyl, can reduce and transform CO2 into useful molecules such as amino acids. However, the steps of the photoredox cycle and reasons for low turnover numbers of these catalysts are poorly understood. Our goal is to utilize quantum chemistry methods to delineate mechanisms of key steps in this cycle and leverage these insights to drive discovery of novel chromophores that are both active and yield high turnover numbers. Thus far, we have demonstrated that the electron transfer (ET) step from the p-terphenyl radical anion to CO2 is adiabatic, and ET barriers are lowered when electron-donating groups are substituted to p- terminal positions of the catalyst. To probe degradation pathways from the excited state, we have established a computational protocol for calculation and characterization of excited-state donor-acceptor complexes, or exciplexes. We are also taking our first steps towards driving discovery of new chromophores by implementing a genetic algorithm (GA) whose fitness function factors in both catalyst activity and degradation resistance by means of simple descriptors obtained from routine DFT calculations. The GA yields several candidates that are more viable than experimentally studied terphenyls, highlighting the importance of automated computational tools in accelerating experimental efforts.

Design and Use of Functionals to Describe Materials Properties: Principles, Difficulties, Ways to go

3/30/22 10:00 am EST

Lucia Reining

École Polytechnique, Palaiseau, France


Properties of materials can in principle be calculated as expectation values using many-body wavefunctions. In practice, this is most often impossible, because one cannot calculate or store the wavefunctions. One important alternative is to express observables as functionals of simpler quantities, such as the density or a one-body Green’s function. However, the exact functionals are unknown, approximations often lack precision, and some observables cannot be accessed at all in a satisfactory way. In a first part of this talk, we will discuss the idea and major general concepts of using functionals. We will see the difficulties of finding good approximations, and some strategies for improvement. We will make this introduction tangible with examples from Density Functional Theory and Many-Body Perturbation Theory, where we build functionals of Green’s functions. Based on this introduction, we will then look at cases of our current research.
First, we will introduce a particular way to profit from results of model systems [1,2]. We will show that model results can be used in an in principle exact way, which we term “Connector Theory”, in order to describe materials properties. Within this approach, a quantity of interest is calculated for a model system as a function of a parameter once and forever, and the results are stored and shared. Under certain conditions, the model result for an appropriate choice of parameter (called connector) can then be used to replace the quantity of interest in the real material. We will discuss the principles and general properties of the connector approach, and show that it leads to interesting approximations.
Finally, we will look at recent results obtained in the framework of Green’s functions. We will in particular concentrate on excitons, with results that go beyond standard textbook expectations and illustrate the rich playground of many-body effects in electronic spectra.

[1] M. Vanzini, A. Aouina, M. Panholzer, M. Gatti, and L. Reining, arXiv:1903.07930v4
[2] A. Aouina, M. Gatti, and L. Reining, Faraday Discussions 224, 27 (2020)

The route going from transition path time distributions and protein folding times, to quantum mechanics, tunneling times, nonadiabaitc transition times, uncertainty and lower bounds to atomic energies

5/11/22 3:30 pm ET at PHO 203

Eli Pollak

Weizmann Institute of Science, Israel


Recent experimental measurements of the transition path time distributions of proteins moving from the folded to the unfolded state and vice versa, presented theory with challenges. Analysis of the results suggested barrier heights that are much lower than the free energies of activation of the observed transitions, what are these barrier heights? Secondly, beyond the mere feat of following a protein as it folds or unfolds, is there anything really useful that we can actually learn from such experiments? These questions lead to a few insights. One is the paradigm of a transition path barrier height which should smaller than the activation energy, resolving partially the low barrier height puzzle [1]. A second one is the observation that when analyzed correctly, the measured distributions reveal long time tails which may be identified with a long lived intermediate, between the folded and unfolded states [2].
In this talk I will explain why Temple’s lower bound and its later improvement by Lehmann could not provide lower bounds with “chemical accuracy”. In recent work, we have shown that one may use a very different approach, based on a matrix in which the Ritz eigenstates are coupled via variances to an exact energy eigenstate to obtain lower bounds whose accuracy is competitive with that of the Ritz upper bound. The methodology will be demonstrated for some toy model as well as the He, Li and Be atoms. Upper and lower bounds are obtained with nano-Hartree accuracy. Perhaps the most difficult part of the new algorithm is that like the older Temple based method one needs to compute variances and this is quite challenging when considering Coulomb potentials. A method will be suggested by which one may accurately estimate variances from matrix elements of the Hamiltonian, without having to expressly compute matrix elements of the Hamiltonian squared operator.
Struggling with the concept of transition path times distributions for proteins naturally led to the question of what is the quantum analog and what can we learn from the same in the quantum mechanical context. This presented us with the challenge of understanding quantum mechanical transition times, and more specifically the time scale of quantum tunneling, a question of some practical interest in view of attosecond experiments on ionization of the He and H atoms as well as the so called Larmor tunneling times of Rb atoms. The highlights of these studies are: a. Theoretical resolution of the tunneling flight as the well known Wigner phase time [3]; b. Verification that the tunneling flight time may be superluminal, even when considering Dirac electrons (in one spatial dimension) [4]; Understanding that this superluminal property does not contradict special relativity, in the sense that it does not lead to superluminal signal transition [4]. Our recent study of electronic transition path time distributions will be presented, demonstrating resonance scattering and the inability of the fewest switches surface hopping algorithm to correctly account for such quantum mechanical phenomena [5]. Finally, transition path time distributions will be presented for fermions, bosons and distinguishable particles [6].
The study of time in quantum mechanics led also to the derivation of a weak value time energy uncertainty relation [7], which in turn led to a renewed interest in lower bound theory, a topic which has remained stagnant for many years, due to the fact that the quality of lower bounds was far inferior to the Ritz upper bounds invented by Ritz in 1908 and 1909. Twenty years later, Temple presented a method for obtaining convergent lower bounds. Ninety years later, the Ritz method is a staple of university courses while the Temple lower bound remains relatively hidden, due to its slow convergence. Finding “good” lower bounds remained a challenge. The Temple class of lower bounds is based on a Cauchy-Schwartz inequality. In recent work, we have shown that one may use a very different approach, based on a matrix in which the Ritz eigenstates are coupled via variances to an exact energy eigenstate to obtain lower bounds [8] whose accuracy is competitive with that of the Ritz upper bound [8,9]. The methodology will be demonstrated for some toy models as well as the He, Li and Be atoms. Using correlated Gaussian basis sets, upper and lower bounds may be obtained with nano-Hartree accuracy [9,10]. Perhaps the most difficult part of the new algorithm is that like the older Temple based method one needs to compute variances and this is quite challenging when considering Coulomb potentials. A method will be suggested by which one may accurately estimate variances from matrix elements of the Hamiltonian, without having to expressly compute matrix elements of the Hamiltonian squared operator.

[1] E. Pollak, The transition path time distribution and the transition path free energy barrier, Phys. Chem. Chem. Phys. 18, 28872 – 28882 (2016). DOI: 10.1039/C6CP05052B
[2] R. Dutta and E. Pollak, What can we learn from transition path time distributions for protein folding and unfolding?, Phys. Chem. Chem. Phys. 23, 23787-23795 (2021). DOI: 10.1039/D1CP03296H
[3] T. Rivlin, E. Pollak, and R. S. Dumont, Determination of the tunneling flight time as the reflected phase time, Phys. Rev. A 103, 012225 (2021).
[4] R.S. Dumont, T. Rivlin and E. Pollak, The relativistic tunneling flight time may be superluminal, but it does not imply superluminal signaling, New J. Phys. 22, 093060 (2020).
[5] X. He, B. Wu, J. Liu, T. Rivlin and E. Pollak, Transition path flight times and nonadiabatic electronic transitions, preprint, to be published.
[6] R. Ianconescu and E. Pollak, A comparison of transition path flight times for fermions, bosons and distinguishable particles, Preprint, to be published.
[7] E. Pollak and S. Miret-Artés, Uncertainty relations for time-averaged weak values, Phys. Rev. A 99, 012108 (2019).
[8] E Pollak and R Martinazzo, Lower bounds for Coulombic systems, J. Chem. Th. Comp. 17, 1535-1547 (2021).
[9] R.T. Ireland, P. Jeszenszki, E. Mátyus, R. Martinazzo, M. Ronto, E. Pollak, Lower Bounds for Nonrelativistic Atomic Energies, ACS Phys. Chem. Au 2, 23–37 (2022).
[10] M. Ronto, P. Jeszenszki, E. Mátyus and E. Pollak, Eigenvalue lower bounds as a practical tool for numerical electronic energy calculations, preprint, to be published.



Garnet Chan

California Institute of Technology, Pasadena, CA



Importance of Electrostatics and the Role of Interfaces for Chemical Transformations

6/1/22 3:30 pm ET at PHO 203

Teresa Head-Gordon

University of California, Berkeley, CA


Chemical transformations rarely occur in a single homogeneous aqueous phase, but instead occur in niches, crevices, and impurity sites at confining interfaces between two or more phases of gases, liquids or solids. Fundamentally, interfaces can alter solvent and solution compositions and phases to reformulate the transition states and pathways of chemical reactions and underlying transport mechanisms. Computational modeling of these systems thereby requires an accurate description of molecular interactions, especially electrostatics, of these complex system. I will present results on how electric fields have been used to computationally optimize biocatalytic performance of a synthetic enzyme, and how they could be used as a unifying descriptor for catalytic design across a range of homogeneous and heterogeneous catalysts including recent hypotheses around microdroplet chemistry. I will also discuss some methodological advances from accurate many-body force fields under non-reactive approximations in classical molecular dynamics, to reactive force fields to describe chemical reactions where charge flow is an essential process