HYDRA: A C++ repository from Molecular spintronics @UHH

HYDRA

HYDRA is a tool to compute hybridization functions using atom-centered basis sets. It can perform local decompositions of hybridization functions. A scientific discussion of this decomposition is given in https://doi.org/10.26434/chemrxiv.13655435.v1

The tool was developed at University of Hamburg by Dr. Marc Philipp Bahlke (in Fortran) and re-written in C++ by Michaela Schneeberger, for an improved performance.

############ Installation ############

Download and unzip the Eigen library (http://eigen.tuxfamily.org/)
Create 'build' directory (mkdir build)
For omp version:
a) cp source/main_omp.cpp ./build/main.cpp and cp source/Makefile_omp ./build/Makefile

To switch off omp, add the option -fopenmp to the compile command in the Makefile. This will result in a serial version.
b) cp source/InputOutput.cpp source/InputOutput.hpp ./build
c) cd ./build
d) Include the path to Eigen library in the Makefile & Edit Makefile according to your systems settings
e) type 'make' in folder with Hydra to compile it
For mpi version: (requires installation of Eigen according to INSTALL file in Eigen directory)
a) cd ./build
b) cmake -G "Unix Makefiles" ../source
c) make all

############ Usage ############

HYDRA requires:

-Input-file (filename=delta.global.in)
-Overlap matrix (filename=overlap)
-Hamiltonian matrix (filename=hamiltonian)
-Eigenvalues (filename=eigenvalues)
-MO coefficients (filename=coef)
-Basis set indices for local decomposition (filename=basis_array)

delta.global.in looks, e.g., as follows:

    3000                     ! nbas, total number of basis functions
    2948                     ! xbf1, index of 1st basis function of impurity 
    2952                     ! xbf2, index of last basis function of impurity
    1                        ! 1= real, 2= matsu; type of energy axis, matsu is Matsubara axis
    100                      ! beta, inverse temperature, only relevant for Matsubara axis
    1001                     ! points, number of data points for the hybridization function
    0.02                     ! step size
    0.1                      ! smear,  smearing or value of the imaginary off-set
    -3.959                   ! Fermi energy
    1                        ! unit (0 = hartree, 1 = eV)
    1                        ! diagonalize (true if 1), note that this is not always required, check if impurity block is already diagonal!
    0                        ! method, 0 = bath, 1 = use Lehmann, 2 = projector (not fully implemented yet)
    1                        ! 1 = calculate a decomposition (only for method 0, requiers basis_array input)

For Turbomole, the required matrices can be obtained by a tool which is provided in ./tools/turbomole. It is required to remove the file extensions in some cases, according to the required files mentioned above, e.g.:

eigenvalues.dat -> eigenvalues
hamiltonian.1 -> hamiltonian
etc.

Usage for other QC codes: Basically, you just need the required matrices as mentioned above in the exact same format (see ./examples).

########## Plotting environment (bath) orbitals and impurity orbitals ##########

Hydra outputs 'truncated' molden files (https://en.wikipedia.org/wiki/Molden). To make them readable by molden, it is required to add the 'header' of a full molden file to the truncated ones. This can be done by hand or by a python script provided in:

./tools/molden/hydra_2molden.py

This script needs the truncated molden files provided by hydra and in addition a full molden file which can, e.g. for Turbomole, created with the script tm2molden (part of Turbomole).

Notes/TODOs:

The impurity orbital indices are not allowed to start from 1. This is, e.g., the case when one of the impurity atoms is the 1st atom in the xyz/coord file and the s functions are included in the local impurity sub-space.

molspintron/HYDRA

HYDRA