Projects

To each early-stage researcher (ESR) in the COSINE network, there is an associated scientific project. Below is a brief description of these projects.

ESR1: This project of Maximilien Ambroise at UHEI (University of Heidelberg) is supervised by Prof. Andreas Dreuw, and will focus on the development of a highly-parallel, novel implementation of the second-order Algebraic Diagrammatic Construction (ADC) method. ADC can be regarded as “MP2 for excited states”, and hence linear scaling schemes for MP2 are expected to be directly transferable to ADC. In a first step, the ADC(2) equations will be reformulated in the atomic orbital (AO) basis, which then allows to introduce strict screening procedures of the transformed integrals and amplitudes, similar to AOMP2 or AOCCSD. The latter approach differs from local correlation methods which heavily rely on (arbitrary) screening as a function of the distance between local molecular orbitals (LMO). In the limit of large molecules, most quantities in the AO basis scale linearly with system sizes, and in order to exploit the inherent sparsity, tensor libraries will be used which employ sparse tensor algebra.

The resulting AOADC method, similarly to AOMP2, is expected to have a large overhead, meaning that the low scaling regime is only reached for large system sizes. One solution consists in using a distributed memory tensor library to brute force the problem with enough processing power in HPC environments. Furthermore, other methods will be investigated such as domain-based approaches, Cholesky decomposition (CD), spin-opposite scaling (SOS) and resolution of the identity (RI). AOADC will be developed in close cooperation with the node at LMU, led by Prof. Christian Ochsenfeld.

ESR2: Mikael Scott works at UHEI (University of Heidelberg) under the supervision of Prof. Andreas Dreuw on the implementation of higher order response functions within the Algebraic Diagrammatic Construction (ADC) Scheme. Initially implementing expressions for Electronic Circular Dichroism within the Intermediate State Representation (ISR) in the Q-Chem package of quantum chemistry programs. This marks the first implementation of ADC-ECD within the ISR formalism. Further development of quadratic and cubic response functions within ADC in Q-Chem will follow as simulation of non-linear spectroscopies represents a promising field of research and vital tools within in several different disciplines (e.g. medicine, biology, material science etc). Especially sum-frequency spectroscopy (SFS) and second-harmonic generation (SHG) provide a vital link between theory and experiment.

An extensive secondment is planned at KTH (Kungliga Tekniska Högskolan) where Mikael Scott will work in Prof. Patrick Norman's group on the development of VeloxChem package, and focus on the implementation of higher order response functions at the Hartree-Fock (HF) and Density Functional Theory (DFT) level. Initially implementing Magnetic Circular Dichroism time-dependent DFT. The overall aim of this project is to achieve the most efficient quadratic response solver for high performance computing.

ESR3: This ESR project at KTH Royal Institute of Technology is supervised by Prof. Patrick Norman and aims at the development of complex polarization propagator (CPP) methods in first-principles electronic structure theory in order to address dispersive and absorptive molecular properties in spectroscopy. The focus to begin with will be to re-formulate the algorithmic procedures for solving linear and nonlinear response functions at the level of density functional theory in order to become highly efficient. This will include the development of an iterative complex linear response solver for multi-frequencies and multi-perturbation operators as well as a highly parallel contraction scheme of generalized Hessian matrices with response vectors. The framework for this project is a newly developed program platform named VeloxChem that, with the ease of Python library modules offers a front end to quantum chemical calculations on contemporary high-performance computing (HPC) systems and aims at harnessing the future compute power within the EuroHPC initiative. At the heart of this software lies a module for the evaluation of electron-repulsion integrals (ERIs) using the Obara-Saika recurrence scheme, where a high degree of efficiency is achieved by employing architecture-independent vectorization via OpenMP SIMD pragmas in the auto-generated C++ source code. The software is topology aware and with a Python-controlled work and task flow, the idle time is minimized using an MPI/OpenMP partitioning of resources.

With efficient use of computer memory, we will enable the simultaneous reference to, and solving of, in the order of 1,000 response equations for sizeable biochemical systems without spatial symmetry, and we can thereby determine electronic response spectra in arbitrary wavelength regions, including UV/vis and X-Ray, without resolving the sometimes embedded excited states in the spectrum. E.g. the electronic CD spectrum (involving the Cartesian sets of electric and magnetic perturbations) over a range of some 10 eV is obtained at a computational cost comparable to that of determining the transition energy of the lowest excited state, or optimizing the electronic structure of the reference state.

In close collaboration with the node at the University of Heidelberg under the lead of Prof. Andreas Dreuw, this work will be integrated with the development of CPP response theory in the algebraic diagrammatic construction (ADC) scheme.

ESR4: This ESR project at KTH Royal Institute of Technology is supervised by Prof. Patrick Norman and aims at the development of multiscale simulation strategies tailored for an accurate description of small and medium-sized conjugated oligothiophene derivatives in complex protein environments. The binding modes and the photophysical properties of aggregate specific ligands bound to distinct protein aggregate morphotypes will be revealed by means of fully atomistic molecular dynamics (MD) and quantum mechanics/molecular mechanics (QM/MM) simulations. These combined tools represent an established approach, used previously by us in collaboration with leading experimentalists in this field. Within the present proposal, in addition, we will introduce state-of-the-art machine learning techniques to derive the force fields and simulate the ultra-fast dynamics in the electronically excited states of the biomarker probes. It is this excited-state MD that governs the photophysical responses observed in the emission spectroscopies that are used in the experimental characterization of the supramolecular protein–probe complexes. A correlation between the aggregate specific ligands binding configuration and the aggregate structure will be afforded and aid in deciphering the biophysical properties underlying aggregate polymorphism. Based on MD simulations of protein– chromophore complexes, the binding mechanisms and sites for the anionic probes will be studied at the atomistic level. New concrete suggestions will be made for molecules to be used as early-stage detectors of neurodegenerative diseases such as Alzheimer’s and Parkinson’s disease.

This project is carried out in collaboration with the node at the Southern University of Denmark (SDU) under the lead of Prof. Jacob Kongsted.

ESR5: This project at the Ludwig-Maximilians-University of Munich (LMU) under the supervision of Prof. Christian Ochsenfeld is focused on the development of efficient non-adiabatic molecular dynamics methods (NAMD) on graphics processing units (GPU).

Such methods allow for investigating light-matter interactions in energy materials, e.g., charge separation, electron transport, and recombination. In order to simulate medium to large molecular systems for a significant time span, it is crucial to develop highly efficient, low-prefactor ab initio methods.

As a first effort, an accelerated fewest-switches surface hopping algorithm (FSSH) has been developed which can be executed on GPUs both using CUDA and OpenCL. Additionally, the use of newly developed exchange-correlation functionals, that show the potential to strongly improve on the well-known short-comings of conventional DFT approximations for systems with charge-transfer or strong-correlation character, will be investigated in the context of NAMD. Here, the development of GPU-based Real-Time TDDFT and Ehrenfest molecular dynamics methods constitute a further target of this project.

ESR6: This project at the Ludwig-Maximilians-University of Munich (LMU) under the supervision of Prof. Christian Ochsenfeld is focused on the development of low-/linear-scaling Laplace-based AO-CC2 and AO-ADC2 methods on graphics processing units (GPU).

Since the ADC scheme provides a straightforward path to evaluate excited state properties, a further target is the development of an efficient, low-scaling method to evaluate excited state forces and non-adiabatic coupling vectors on the ADC2 level of theory. The resulting methods can therefore be employed in non-adiabatic molecular dynamics simulations developed by ESR5.

This project is a close collaboration with the group of Prof. Andreas Dreuw at the University of Heidelberg.

ESR7: This project is developed at SNS - Scuola Normale Superiore, Pisa Italy- and is supervised by Prof. Chiara Cappelli. The aim of the project is to develop and test QM/MM computational protocols for the study of Resonance Raman and Resonance Raman Optical Activity spectra. The target systems to which the protocols are applied consist of medium-sized molecular systems, embedded in complex extrnal environments (solvents or proteins). Special attention will be devoted to the description of spectral differences for drugs in solvent or intercalated in DNA.

ESR8: This project is developed at SNS - Scuola Normale Superiore, Pisa Italy- and is supervised by Prof. Chiara Cappelli. The aim of the project is to develop and test computational protocols, based on a hierarchy of QM and QM/MM techniques, for the calculation of absorption spectra of single solvated dyes and donor-acceptor couples of potential interest for photovoltaic devices.

ESR9: This ESR project at SDU – University of Southern Denmark – is supervised by Prof. Jacob Kongsted. The overall aim of the PhD project is to develop and implement novel methods for the description of molecular (structured) environments coupled to quantum chemistry methods thereby targeting calculation of spectroscopic properties of molecules embedded in complex environments such as a solvent, surface or protein. By developing such multi-scale methods, we will be able not only to describe the spectroscopic probe accurately, but also to include environmental and dynamical effects into the quantum chemistry calculations. The developed methods will build upon polarizable embedding approaches, which recently have been designed to address excited states as well as their properties in the framework of response theories. Of special interest within this project will be the development of rational design strategies for formulation of novel optical molecular probes that can be used to sense in biological environments such as for example a membrane.

This project is carried out in collaboration with the node at the KTH Royal Institute of Technology and the node at Scuola Normale Superiore Pisa.

ESR10:

ESR11: This project at NTNU - Norwegian University of Science and Technology - is supervised by Prof. Henrik Koch. It aims at the development and benchmark of multilevel methods for the study of large molecular systems. While an accurate study of properties of interest can require high-level quantum mechanical (QM) calculations, their cost scales steeply with respect to system size. However, many properties are localized on a specific region of the system, albeit affected by its environment. When this is the case, different levels of theory can be applied to the different regions. Based on this rationale, various multilevel methods have been developed.

The goal of this project is to implement and optimize a multilevel model based on Hartree-Fock theory, denoted as multilevel Hartree-Fock (MLHF), in the program eT (www.etprogram.org). MLHF can provide the reference wave function for various standard and multilevel coupled cluster methods. The coupling of these quantum embedding models to an external molecular mechanics (MM) layer will be explored in collaboration with the node at Scuola Normale Superiore (SNS). The resulting QM/QM/MM protocols will be tested and applied to the study of local properties in solution, such as solvatochromic shifts.

ESR12: This project at NTNU - Norwegian University of Science and Technology - under the supervision of Prof. Henrik Koch is focused on the development and implementation of efficient algorithms to compute photochemical properties like excitation energies within the Coupled Cluster hierachy of methods. Coupled cluster theory, especially in its variants including triple substitutions, is well suited to model excited states and their properties. However, due to its steep scaling, highly optimized parallel implementations are required to be able to compute excitation energies of medium-sized molecules. Additional savings can be achieved by splitting the orbital space and treating only parts of a system at a high level of theory, while introducing only a small error compared to treating the full system with the high-level method. This latter approach is called multilevel coupled cluster theory (MLCC).

The goal of this project is the implementation of transition properties for the coupled cluster with singles doubles and perturbative triples (CC3) method in its standard and multilevel variants within the electronic structure program eT (www.etprogram.org). These new methods will then be applied to various molecular systems in close collaboration with the node at Scuola Normale Superiore (SNS) and the Technical University of Denmark (DTU).

ESR13: This ESR at the Technical University of Denmark (DTU) is supervised by Prof. Sonia Coriani. Triggered by third‐generation synchrotron radiation sources and free‐electron lasers, high‐quality X‐ray spectroscopies like Near‐Edge Xray Absorption Fine Structure (NEXAFS) and Auger spectroscopy for ground and excited electronic states, X‐ray Circular Dichroism (XCD), Resonant Inelastic X‐ray Scattering (RIXS), multiphoton X‐ray absorption and other non‐linear X‐ray spectroscopies are becoming increasingly popular. For their meaningful interpretation, accurate computational approaches are critically important. In this PhD project, one of the aims is to develop the necessary theoretical methodologies within standard/damped response theory for both ground‐state and electronic excited states. The latter are essential to be able to simulate and interpret pump/probe experiments. Photoionisation/photodetachment processes and photoelectron spectroscopy will also be targeted. The development will take place primarily at the Coupled Cluster (CC) level of electronic structure theory. With the developed methods, non‐radiative decay channels in molecular systems of interest in natural sciences and engineering will be studied. Light‐induced processes will be investigated and the corresponding spectroscopic signals computed along the proposed reaction channels for direct comparison with time‐resolved pump/probe spectroscopy. In particular X‐ray (linear and non‐linear) absorption spectroscopy, as well as photoelectron spectroscopy, will be simulated.The description of photoionization/detachment processes at a highly correlated level is a very challenging task, as it requires the description of the electronic continuum. Methods including a correlated treatment of the continuum are still lagging behind those for pure bound states. We will pursue the extension of methods based on Stieltjes Imaging and Analytic Continuationto X‐ray photoionization, excited‐state VUV photoionisation, auto‐ionisation and intermolecular decay phenomena and dichroism effects in the continuum, including a B‐spline basis to improve the description of the continuum at CC level. Photoelectron spectroscopy (PES) and its dynamical descriptors shall also be computed using Dyson‐orbitals, that is, the overlap between an N‐electron wave function of the molecule and the N–1/N+1 electron wavefunction of the ion. This gives access to information like angular distributions of photoelectrons, Compton profiles or electron momentum spectra. Dyson orbitals will be obtained from the elements of the CC transition density matrices for all members of the CC hierarchy and coupled them to an efficient multicenter Bspline DFT code for the description of the outgoing photoelectron. These new schemes will be extremely useful for the investigation of photoelectron processes in which electron correlation plays a critical role, as for instance, the inner‐valence shell satellite bands.

Close collaboration with NTNU and external partner UNITS is involved.

ESR14: This ESR at the Technical University of Denmark (DTU) is supervised by Prof. Sonia Coriani. In this specific PhD project, one of the focus points will be on three different, but conceptually related, optical effects connected to the differential absorption of circularly polarized light in presence of additional magnetic fields, namely Magnetic Circular Dichroism (MCD), Magneto‐Chiral Dichroism (MChD) and Nuclear Spin‐Circular Dichroism (NSCD). MCD, MChD, and NSCD can be theoretically ascribed to tensors connected to quadratic response functions. MCD and MChD contain the magnetic dipole operator, that calls for specific measures to ensure gauge invariance, e.g. use of gauge‐including atomic orbitals. Both wave‐function (CC, ADC) and density‐functional based (TD‐DFT) protocols will be pursued, based on both standard and damped response theory.

MCD is a powerful technique for probing electronic states indistinguishable in the unpolarized absorption spectrum. Building on previous development of complex‐polarization‐propagator (CPP) TD‐DFT‐based computational protocols for MCD, analogous complex polarization propagator methods based on the Coupled Cluster (CC) and the Algebraic Diagrammatic Construction (ADC) formalisms will be pursued, in collaboration with the UHEI node (Prof. Dreuw).

The experimental detection of MChD in biologically relevant molecules remains elusive, so computational simulations will be of fundamental help. Our previously defined protocols for MChD will be implemented and used to guide the development of MChD experiments, in particular on molecular systems important in life science. We want to identify molecular compounds possessing MChD signals above the detectability limit of current experimental setups and, thereby, help verify the hypothesis that MChD is a key element in explaining homochirality in nature. Ab initio, as well as DFT methodologies, will be used, and solvent effects included. We will moreover pursue a reinvestigation of the existing discrepancies between magnetochiral birefringence effects observed using different experimental setups and between experimental and computational results. NSCD is an example of nuclear magneto‐optical spectroscopy where optical measurements of dichromism are coupled with themagnetization of atomic nuclei to induce changes in the polarization state of light, potentially providing nucleus‐specific spectroscopic signals. While the first experimental NSCD apparatus is being constructed, it is crucial to provide a basic understanding of the NSCD observable to guide these first experiments.

Close collaboration with UHEI and KTH is involved.