# Developers:Workshops:DMFT:Meeting Notes

From ALPS

- [#ALPS.2FDMFT_workshop.2C_December_7-9.2C_2006.2C_G.C3.B6ttingen ]

- [#Meeting_Notes ]

- [#Thursday_afternoon ]
- [#Friday_Morning ]
- [#Friday_Afternoon ]
- [#Saturday_Morning ]

## Contents

# ALPS/DMFT workshop, December 7-9, 2006, G$ouml;ttingen

## Meeting Notes

### Thursday afternoon

- In attendance:
- Frithjof Anders (arrival around 15:45)
- Sebastian Fuchs
- Emanuel Gull
- Andreas Honecker
- Hunpyo Lee
- Gang Li
- Munehisa Matsumoto
- Reinhard Noack (arrival shortly before 17:30)
- Robert Peters
- Michael Potthoff
- Thomas Pruschke
- Ulrich Schollw$ouml;ck

- Schedule
- start at 15:00
- Review of existing ALPS DMFT framework by Emanuel Gull
- Issues:
- Discretization of energy/time axis

Cases:

- G(tau_i), tau_i=i beta/N
- G(omega_i) for some given (non-equidistant) omega_i
- (a_n,b_n), n=0, 1, ... for continued fraction expansion

proposed solution: map<double, complex>, meaning a tag in XML

- Interface for Hamiltonian solvers
- Interface for complex ordered phases
- Self-energy functionals by Michael Potthoff (not urgent)
- Analytic continuation (maximum entropy etc.) (not urgent)
- Measurements of two-particle Green's functions

### Friday Morning

- Start at 10:30, in attendance: same as Thurs. afternoon
- General ALPS talk for those that are new to ALPS (Reinhard)
- Comments/Discussion:
- ALPS 2.0: More flexible scheduler appropriate for DMFT (and parallel tempering in QMC)
- Band structure interface: contact other groups (e.g., O Anderson, www.abinitio.org, ...)
- Documentation: migrate more fully from old html web pages, make link more visible?

- DMFT-QMC blackboard discussion (Thomas P.)
- construction of effective Hamiltonian (for DMRG, NRG, ED)
- Bottom line for DMFT-loop of ALPS: loop: Solver->sigma(z)->ALPS->G_0(z)->Solver
- Numerical stability of DMFT loop - precision of numbers, rounding errors
- ALPS box: should contain stabilization (e.g., mixing old $amp; new solutions, spin averaging)
- What is saved at each loop interation?
- sigma(z), G_0(z) data elements can be more general (matrices) for DCA, CDMFT, ...
- Clusters/multiple orbitals generated from model/lattice library descriptions

### Friday Afternoon

- Talks in SFB Workshop
- Michael P.: Nonperturbative Conserving Appriximations for Systems of Strongly Correlated Electrons
- Uli S.: Recent Trends in the DMRG
- Emanuel G.: Continuous Time Solvers for DMFT
- Frithjof A.: DMFT and Numerical RG at Finite Temperature: Prospects and Challenges

### Saturday Morning

- Start at 09:30
- In attendance:
- Frithjof Anders
- Sebastian Fuchs
- Emanuel Gull
- Andreas Honecker
- Hunpyo Lee
- Gang Li
- Munehisa Matsumoto
- Reinhard Noack
- Robert Peters
- Thomas Pruschke

- Analytic continuation (stochastic sampling) status: Sebastian F.
- Hilbert transform for arbitrary D.O.S. (e.g., from LDA)
- Tetrahedron integration
- Poles, interpolation
- -> Fortran 77 code from Anisimov group -> ALPSification?
- Multiband case: Green functions matrices; cases:
- H(k) diagonal in orbital indices
- H(k)-> diagonal, sigma(z) not diagonal

- Immediate Goal: flexible data formats, initial support of usual (simple) D.O.S.'s
- Self-consisitency condition depends on model, e.g., periodic Anderson model
- Specification of ordered phases: fits into general multiorbital scheme, e.g., magnetic supercell?
- issue: need information about lattice structure beyond D.O.S. -> adapt input to Hilbert transform

- Generalization of XML definitions for Solver->Hilbert transforms
- VECTOR->MAP
- tags for spin, orbital, type of order, e.g., paramagnetic

- Closing of workshop at 11:30

Thanks to Reinhard for a major fraction of these notes.