Besprechungsraum Astrophysik (BZA, F.04.122)
Cosmology Seminar Göttingen
Pedro R. Capelo
University of Michigan
Dynamics of black holes, AGN activity, and star formation in galaxy mergers
I present the latest results from a suite of high-resolution hydrodynamical simulations of mergers between two late-type galaxies, in which I vary the mass ratio, gas fraction, and orbital configuration, and where I include realistic descriptions of cooling, stellar and black hole (BH) feedback, and star formation. I study both the evolution of the central BHs and the fate of the galaxies themselves, from the moment the two galaxies can still be considered in isolation to coalescence. In the first part of my talk, I focus on the dynamics of the central stellar nuclei during these galactic encounters and on their importance in determining the time-scale of BH pairing. I then shift the focus to the accretion and growth of the central BHs, with emphasis on the role of the gas angular momentum: I examine which mergers preferentially trigger AGN activity, distinguish between merger-related and secular-related accretion, and trace the resultant evolution of the BH mass ratios in the pairing process. I further discuss the link between BH accretion and central star formation, in an attempt to explain the observed relation between BH mass and bulge mass.
Ludwig-Prandtl-Hörsaal, Am Faßberg 11, 37077 Göttingen
Prof. Alexander Smits
Mechanical and Aerospace Engineering Princeton University Princeton, NJ, USA
Logarithmic scaling in wall-bounded turbulence
Logarithmic scaling is one of the corner stones of our understanding of wall-bounded turbulent flows. In 1938, Clark B. Millikan advanced an overlap argument that framed the logarithmic variation of the mean velocity in simple dimensional terms. Seventy-five years later, however, basic aspects of this logarithmic region, such as its slope (described by von Karman’s constant), and its spatial extent, are still being debated. In addition, Townsend in 1976 proposed a logarithmic scaling for the streamwise and spanwise components of turbulence based on the attached eddy hypothesis, but to date the experimental verification has been elusive. Here, we use pipe and boundary layer flow measurements over a very large Reynolds number range to examine these expectations of logarithmic scaling, and to show that at sufficiently high Reynolds number these flows reveal both expected and unexpected implications for our understanding and our capacity to model turbulence.
Kontakt: Eberhard Bodenschatz
Dr. Dan Mordehai
Department of Mechanical Engineering,Technion, Haifa, Israel
Size-Dependent Mechanical Properties of Crystalline Nanoparticles
It is well established that materials can drastically change mechanical properties when their size is reduced to the nanoscale, mainly because of an increase in surface to volume ratio and of lowering the amount of defects in the lattice. Defect free crystalline nanostructures reach strengths which are two orders of magnitude higher than their bulk counterparts, since their deformation is controlled by dislocation nucleation from the surfaces. In this talk we examine how the size and shape nanoparticles affect the mechanical response to compression and indentation. Earlier experiments on Au nanoparticles showed that they become easier to indent as they are smaller, but a reduction of their size increases their strength under compression. Molecular Dynamics (MD) simulations show how the lateral dimensions give rise to size effect in indentation through the competition between dislocation storage and depletion on free surfaces. On the other hand, under compression, the size effect arises from a size-dependent dislocation nucleation criterion at the nanoparticle’s vertices. While other FCC metals show similar behavior, the size effect is suppressed in indentation of Fe nanoparticles (BCC), due to strong dislocation pining beneath the indent. However, the strength of Fe nanoparticles under compression is increasing as their size is reduced, due to the stress concentration at the nucleation site. The influence of the stress gradient can be controlled through the nanoparticles geometry. For instance, the size effect vanishes in Ni3Al intermetallic nanocubes under compression. An analysis of the dislocation evolution in Ni3Al nanoparticles shows that partial dislocations are nucleated at the vertices, shearing the nanoparticle with large complex stacking faults planes, regardless of their size. Understanding how material type and geometry governs the deformation at the nanoscale serves us as guidelines in designing and controlling mechanical properties of nanustructures.
MPIDS, Bunsenstraße 10, House 8, Lecture Hall
Dr. Michael Habeck
Institut f. Math. Stochastik, Göttingen
Bayesian approach to inverse statistical mechanics
Physical modeling of complex systems is typically based on an energy function that quantifies the interactions between parts of the system. Often there is no tractable physical theory that allows us to derive the interaction strengths from first principles. Rather the couplings have to be recovered from observed system configurations or, even more indirectly, from ensemble data. To recover interactions from indirect observations is a widespread problem in molecular physics, neuroscience, and biological data analysis. Examples include the determination of molecular interaction potentials from scattering curves and the estimation of coupling strengths of a spin glass (inverse Ising problem). Here I outline statistical approaches to tackle these inverse problems including maximum likelihood, Bayesian inference, and entropy maximization. I explain formal equivalences between these approaches and introduce a general Markov chain Monte Carlo algorithm that can be used to solve inverse problems in statistical mechanics. I will illustrate the method on inverse problems of varying complexity, including the estimation of a temperature, the inverse Ising problem, maximum entropy fitting, and the reconstruction of molecular interaction potentials.
Kontakt: David Hofmann, MPIDS, Dept. NLD