Ground state properties of large atoms and quantum dots

We investigate the ground state properties of large atoms and quantum dots described by a d-dimensional N-body Hamiltonian of confinement ZV. In atoms, d = 3 and V is the Coulomb interaction; in dots, d = 2 and V is phenomenologically determined. We express the grand-canonical partition function in a path integral approach, and evaluate its expansion in Z-1. The problem can be seen as that of field theory possessing a saddle point. This saddle point results in a mean-field contribution to the energy, while the fluctuations result in the correlation energy. The mean-field contribution to the energy is self-consistently determined by the Hartree potential and contains an exchange term. Its smooth contribution is evaluated by a semiclassical method, with ε = Z-1/d in the role of ℏ, while its oscillating contribution can be related to the periodic orbits in the corresponding classical Hamiltonian. In the case of atoms, the leading order in ε of the correlation energy contains a term in Z ln Z1/3, which is essential in reproducing the behaviour shown by reference values, and a term in Z. While we have evaluated the contribution to the Z-term provided by the leading fluctuation order, the numerical evaluation of the contributions provided by higher order fluctuations remains an open problem. The self-consistent contribution to the energy corresponds to the statistical atom, composed of Thomas-Fermi and its corrections, comprehensively analysed, including oscillations, by Schwinger and Englert. In the case of dots, the leading order in ε of the correlation energy is a universal contribution of order Z, which we obtain in closed form. We then determine the expansion in ε of the smooth contributions down to this correlation order. We apply the approach to dots of quadratic and quartic confinement, including the oscillating contribution in the case of a chaotic quartic confinement.

Structure-preserving approaches and data-driven closure modeling for model order reduction

Nicolò Ripamonti

In this thesis, we propose model order reduction techniques for high-dimensional PDEs that preserve structures of the original problems and develop a closure modeling framework leveraging the Mori-Zwanzig formalism and recurrent neural networks. Since high-fidelity approximations of PDEs often result in a large number of degrees of freedom, the need for iterative evaluations for numerical optimizations and rapid feedback is computationally challenging.The first part of this thesis is devoted to conserving the high-dimensional equation's invariants, symmetries, and structures during the reduction process. Traditional reduction techniques are not guaranteed to yield stable reduced systems, even if the target problem is stable. In the context of fluid flows, the skew-symmetric structure of the problem entails the preservation of the kinetic energy of the system. By preserving the same structure at the level of the reduced model, we obtain enhanced stability, and accuracy and the reduced model acquires physical significance by preserving a surrogate of the energy of the original problem. Next, we focus on Hamiltonian systems, which, being driven by symmetry, are a source of great interest in the reduction community. It is well known that the breaking of these symmetries in the reduced model is accompanied by a blowup of the system energy and flow volume. In this thesis, geometric reduced models for Hamiltonian systems are further developed and combined with the dynamically orthogonal methods, addressing the poor reducibility in time of advection-dominated problems. The reduced solution is expressed as a linear combination of a finite number of modes and coincides with the symplectic projection of the high-fidelity Hamiltonian problem onto the tangent space of the approximating manifold. An error surrogate is used to monitor the approximation ability of the reduced model and make a change in the rank of the approximating system if necessary. The method is further developed through a combination of DEIM and DMD to reduce non-polynomial nonlinearities while preserving the symplectic structure of the problem and applied to the Vlasov-Poisson system.In the second part of the thesis, we consider several data-driven methods to address the poor accuracy in the under-resolved regime for Galerkin reduced models via a closure term. The closure term is developed systematically from the Mori-Zwanzig formalism by introducing projection operators on the spaces of resolved and unresolved scales, thus resulting in an additional memory integral term. The interaction between different scales turns out to be nonlocal in time and dominated by a high-dimensional orthogonal dynamics equation, which cannot be solved precisely and efficiently. Several classical methods in the field of statistical mechanics are used to approximate the memory term, exploiting the finiteness of the memory kernel support. We conclude this thesis by showing through numerical experiments how long short-term memory networks, i.e., machine learning structures characterized by feedback connections, represent a valid tool for approximating the additional memory term.

EPFL2022

Power-Efficient Sensor Placement and Transmission Structure for Data Gathering under Distortion Constraints

Razvan Cristescu

We consider the joint optimization of sensor placement and transmission structure for data gathering, where a given number of nodes need to be placed in a field such that the sensed data can be reconstructed at a sink within specified distortion bounds while minimizing the energy consumed for communication. We assume that the nodes use either joint entropy coding based on explicit communication between sensor nodes, where coding is done when side information is available, or Slepian-Wolf coding where nodes have knowledge of network correlation statistics. We consider both maximum and average distortion bounds. We prove that this optimization is NP-complete since it involves an interplay between the spaces of possible transmission structures given radio reachability limitations, and feasible placements satisfying distortion bounds.We address this problem by first looking at the simplified problem of optimal placement in the one-dimensional case. An analytical solution is derived for the case when there is a simple aggregation scheme, and numerical results are provided for the cases when joint entropy encoding is used. We use the insight from our 1-D analysis to extend our results to the 2-D case and compare it to typical uniform random placement and shortest-path tree. Our algorithm for two-dimensional placement and transmission structure provides two to three fold reduction in total power consumption and between one to two orders of magnitude reduction in bottleneck power consumption. We perform an exhaustive performance analysis of our scheme under varying correlation models and model parameters and demonstrate that the performance improvement is typical over a range of data correlation models and parameters. We also study the impact of performing computationally-efficient data conditioning over a local scope rather than the entire network. Finally, we extend our explicit placement results to a randomized placement scheme and show that such a scheme can be effective when deployment does not permit exact node placement.

2006

Development and validation of data sets and Monte Carlo methods for electron/positron transport at low and medium energies

Cezarina Elena Negreanu

When exposed to ionising radiation, living tissue can potentially suffer somatic and genetic damage - effects depending mainly on the radiation dose or energy absorbed, the type of radiation, and the type and mass of cells affected. It is well known that large doses of radiation lead to high damage of the cell nucleus and additional cell structures, which results in harmful somatic effects, and even rapid death of the individual exposed, while at low doses, cancer is by far the most important possible consequence. Understanding the mechanisms by which low doses of radiation cause cell damage is thus of great significance, not only from this viewpoint but also from that of practical medical physics applications such as radiotherapy treatment planning. Ionising radiation, such as electrons and positrons, begins to cause damage to the genome of a living cell by direct ionisation of atoms, thus depositing energy in the DNA double helix itself. The energy threshold for inducing strand-breaks by electrons, however, is around 7 eV, well below the energy levels required for direct ionization. The low-energy electrons that are set in motion around the tracks of energetic charged particles, for example, are responsible for a multitude of low-energy events (energy transfer of the order of 10 eV), which play a significant role in inducing molecular damage. Assessing the spatial configuration of energy transfer events and the deposited energy spectra, in regions of cellular and sub-cellular dimensions, can be aimed at via the application of appropriate Monte Carlo simulation tools. Such calculations depend primarily on an accurate knowledge of the production and subsequent slowing down of secondary electrons that form the basic structure of the charged particle track. In the above context, an important requirement is the provision of detailed quantitative information concerning the interaction cross sections of electrons over an energy range extending down to low energies, i.e. including the sub-excitation domain. In addition, developing fast particle-transport simulation algorithms to cover the entire slowing down process efficiently is a key aspect. Thus, the present doctoral research has, as global goal, the development and validation of new Monte Carlo calculational tools for electron and positron transport in biological materials, both at high and low energies. More specifically, it aims at providing (i) a comprehensive and accurate set of appropriate cross section data, and (ii) a fast and reliable algorithm for the simulation of charged particle transport. Thus, the first part of the thesis concerns the assessment, further development and validation of standard theoretical models for generating electron and positron cross sections to cover the main interactions of these particles with matter, in particular with the basic atomic components of biomaterials (water, bio-polymers, etc.). This has been done for bremsstrahlung, and both elastic and inelastic scattering, considering a wide range of atomic numbers and high up to thermal incident particle energies. In particular, the excitation cross sections for medium and low energies (down to 1 eV) have been derived by using a new formalism based on many-body field theory. The accuracy of the presently obtained data sets are assessed against other theoretical models, as also a large experimental database for each type of interaction, so that both a comprehensive coverage and adequate accuracies have been ensured for the cross section data sets generated. In the second part of the thesis, an extension of the Monte Carlo code system PENELOPE is first undertaken such that use can be made of elastic scattering differential cross sections which have been made available in numerical form. Thereby, new computational routines (incorporated into the new code PENELAST) prepare the cross sections, needed for a given energy and scattering angle, by applying a fast and accurate sampling technique to a provided data set. The present development will allow various electron and positron cross section data libraries, appropriately formatted, to be used with PENELOPE for benchmarking purposes. The development of a high calculation-speed (Class I) Monte Carlo tool for charged particle transport in biological materials has then been addressed. Thereby, a numerical algorithm for calculating the multiple-scattering angular distributions of high energy electrons and positrons is developed, based on the multiple-scattering theory of Lewis which accounts for energy losses within the continuous slowing down approximation. Partial-wave elastic scattering differential cross sections made available in numerical form, as indicated above, are used for the calculations, the inelastic scattering differential cross sections being obtained from the Sternheimer-Liljequist generalized oscillator strength model implemented in PENELOPE. The new code LEWIS has been used to calculate multiple-scattering angular distributions for given path lengths and can be readily adopted for Class I Monte Carlo simulations. The simultaneous generation of a large number of Legendre expansion coefficients is rendered possible, both rapidly and accurately. Results from LEWIS have been found to be in satisfactory agreement, both with detailed simulations carried out using PENELAST and with various sets of experimental data for high to medium energy electrons. In brief, the present research represents a significant improvement in the quality of Monte Carlo modelling of charged particle slowing down processes, thus contributing to understanding the. role of low-energy secondary electrons in radiation protection studies. It will also allow the further development of a complete Class I Monte Carlo code, which can then be reliably used in practical applications such as radiation treatment planning.

EPFL2005