COMPETENCE NETWORK FOR TECHNICAL, SCIENTIFIC HIGH PERFORMANCE COMPUTING IN BAVARIA
High-Temperture Superconductivity and Scalability of Software Architecture for HPC
The general idea of this project consists in providing a new "local" concept for treating solid-state physics and material properties on the basis of high-performance computing (HPC). This idea may be viewed as proliferating the information contained in a microscopic cluster and its interacting electrons and ions to the infinite system size of the macroscopic solid. If this is possible then, for example, local binding concepts of chemistry can be used to systematically make contact with material science. This concept has indeed been worked out already to some extent in our Theory Group in connection with earlier KONWIHR Projects ("OOPCV" and "CUHE") in the last few years aiming at the microscopic theory, i. e. the fundamental understanding of High-Temperature Superconductivity (HTSC).
The numerical implementation of these so called "embedded cluster schemes" requires first the solution of the isolated cluster problem. For this we are routinely using Exact Diagonalization for extracting the ground-state and Quantum-Monte-Carlo (QMC) for obtaining the finite-temperature properties. Within a typical QMC algorithm, the calculation is carried out, with each core performing its own QMC procedure, including the so-called "warm-up phase", which is required to place the simulated system in thermal equilibrium. This creates a rather substantial waste of computer resources, which we want to avoid in this project. The idea is to combine many cores, which perform the QMC updating process, in one QMC step. The overall communication between the "combined" cores will be managed by tools like the Message-Passing-Interface (MPI).
The applications, which then can use substantially larger individual clusters, will center around two topics:
- We have recently obtained in KONWIHR-supported embedded cluster calculations the phase diagram of the high-Tc Superconductors (HTSC) on the basis of a model (Hubbard model) which reproduced salient features of the experiments (competition of Anti-ferromagnetism (AF) and superconductivity (SC)). Here, we aim at a deeper understanding of what drives superconductivity and what is the underlying pairing mechanism in this proto-type model of a so-called doped Mott Insulator. Is it spin-fluctuations being related to the AF phase or charge-fluctuations? What is the dynamics of the pairing interaction in the high-Tc cuprate superconductors? Is it retarded or non-retarded, i.e. instantaneous?
- The ongoing search for new high-temperature SC has recently resulted in yet another surprise: The discovery of SC in the iron-based compound LaOI-x Fx Fe P sparked a new direction to explore SC in a completely new class of materials. Very recently, the substitution of P by As and of La by Nd and Pr raised Tc to over 50 K with the corresponding material becoming the superconductor with the highest Tc among non-cuprate based solids. In this part (b), we aim at exploring and understanding the mechanism of SC in these and similar iron-based compounds, an understanding of which holds a lot of promise for systematically increasing Tc, maybe even beyond the records (approx. 130K) in the cuprate superconductors.