PETAMAT

Many important classes of materials are strongly correlated. Most magnets and superconductors fall into this category. Although these are materials of great technological promise, they are poorly understood due to longranged spin and charge correlations, competing ground states, and their complex phase diagrams. The goal of this SciDAC project is to build a new MultiScale ManyBody (MSMB) formalism and a suite of codes which treat each length scale with an appropriate approximation. Short length scales, where correlations are strong, are treated explicitly using Quantum Monte Carlo (QUEST and HFQMC). Intermediate length scales are treated diagrammatically, using information from the QMC simulation of the cluster (ParquetMSMB and GammaMSMB). The longest length scales are treated with a Dynamical Mean Field (SELDA), which uses the diagrammatic calculation. These codes are developed by an interdisciplinary team of computational physicists, applied mathematicians and computer scientists. Work are in progress. Your comments and suggestions are greatly appreciated. Please contact Mark Jarrell (jarrellphysics@gmail.com) for further details and sending your comments. 

QUEST
(Quantum Electron Simulation Toolbox) is a Fortran 90/95 package
that implements the Determinant Quantum Monte Carlo method for simulation
of magnetic, superconducting, and metalinsulator transitions in model
Hamiltonians. Currently, the available code treats the repulsive Hubbard model
with the ability to treat general lattice geometries. The QUEST site
includes the source codes, the user's manual
for installation, as well as sample input and output results. Publications
related to QUEST are also available at the site. Future upgrades will
include codes for the attractive Hubbard and Holstein Hamiltonians. Poster and Summary paper 

HFQMC: The Hirsch Fye cluster solver is a Fortran 90/95 solver that implements the Hirsch Fye Quantum Monte Carlo method for Greens function update. More specifically the solver updates the HubbardStatonovich fields by sweeping sequentially through them and testing if changes are required. The Ising field is updated by the metropolis or heat bath algorithm. If changes are accepted, then the Greens function is updated appropriately. HFQMC is available from http://scicompforge.org/petamat/coderelease. The cluster solver, a sample driver, makefile and example input file are provided along with instructions for compiling and executing the code.
ParquetMSMB: This parquet code is a massively parallel Fortran 90/95 package that implements an iterative solution of the diagrammatic parquet equations. In this code, the fully irreducible vertex is approximated by the bare interaction, resulting in the socalled parquet approximation. For a given Hubbard model, the code requires as input the hoping integrals, the Coulomb interaction and the temperature. Upon convergence, it produces the selfenergy, the oneparticle Green's function and the different vertex functions that are essential for the understanding of the underlying physics (superconductivity, magnetism,...). The website includes the source codes, a makefile and a typical input file as well as the corresponding output files. 

GammaMSMB: In the twoparticle irreducible vertex (Gamma) multiscale manybody formalism, the large cluster irreducible vertices Gamma(k,k',q)'s are approximated by the small cluster Gamma's that are calculated explicitly with a Quantum Monte Carlo (QMC) technique. In this formalism, the full frequency and momentum dependence is taken into account. Additionally, since the irreducible vertices contain the correct short length physics, no ansatz is used in the calculation of the self energy.
SELDA is a Fortran 90 based package of programs that implements electronic structure calculation using a combination of local density approximation (LDA) and externally given k and frequencydependent selfenergy. It is based on the extension of the LDA code LMTART which implements linear muffin tin orbital (LMTO) method for electronic structure calculation. The prototype of this code which includes the source files, the manual and some example files is available together with the user friendly interface MINDLab which runs under Windows operating system and is designed specifically for students and nonexperts in the field. 

This project is supported through a SciDAC (Scientific Discovery through Advanced Computing) grant by the U.S. Department of Energy  Office of Science, Advanced Scientific Computing Research and the National Nuclear Security Agency under the contract number DEFC0206ER25793. 