Preparing the PARAMS file
Here we provide a standard PARAMS file for CT-QMC calculation for two-particle Green’s functions for Nickel at =50 (eV) which should be run on at least 1000 cores for a day to produce high quality data over all energies.
OffDiagonal real Sig Sig.out Gf Gf.out mode GH SampleVertex 100 nOm 2 Delta Delta.inp cix actqmc.cix Nmax 3000 # Maximum perturbation order allowed nom 200 # Number of Matsubara frequency points sampled exe ctqmc # Name of the executable tsample 50 # How often to record measurements Ed [ -26.407760, -26.407519, -27.011761, -26.407397, -27.011682, -25.662041, -25.662114, -26.272431, -25.662012, -26.272482] # Ed: Eimp-Edc for PARAMS M 80000000 # Total number of Monte Carlo steps per core Ncout 2000000 # How often to print out info PChangeOrder 0.9 # Ratio between trial steps: add-remove-a-kink / move-a-kink mu 26.407760 # mu = -Eimp for PARAMS warmup 5000000 # Warmup number of QMC steps sderiv 0.02 # Maximum derivative mismatch accepted for tail concatenation aom 10 # Number of frequency points used to determin the value of sigma at nom U 4.0 J 0.3 nf0 8.0 beta 50.0
If you have run CTQMC with this PARAMS file, you have in your local directory a new file named tvertex.dat. It has stored the two-particle Green’s functions for Nickel. I shall explain to you the data format now. So now we want to have the local Green’s functions and the non-local Green’s functions. These three quantities are inputs to the BSE. So for that, we want to run lmfdmft in a speecial mode called gprt:mode=3 to produce the loca and non-local Green’s functions.
Producing local and non-local Green’s functions:
cp Sig.out sig.inp lmfdmft ni -vnk=12 --rs=1,0 --ldadc=28.95 -job=1 --gprt:mode=3:nom=100 --fullg
This produces two files named gloc.ni and gkloc.ni. Suscept.py, the script that solved the BSE and produces the dynamic momentum resolved Susceptibilities, reads them. It would also look for a file with the k-list and a file with the Q-list. The k-list is the same k-grid information on which gkloc.ni is written. And Q-list file contains the Q-grid on which you want to find the susceptibilities .
The input files and their paths are identified in the suscept.in. You can have the input files from here this link. The suscept.in file looks like below:
suscept.klist # filename with k-list suscept.qlist # filename with Qlist rmesh.dat # real axis mesh ./G_k1r # file with real axis k-dependent Gren's function ./G_local1r # file with real axis local Gren's function ./chi0_real. # name of the Bubble on real axis gkloc.ni # imaginary axis k-dependent Green's function gloc.ni # imaginary axis local Green's function ./tvertex.dat # ctqmc local vertex function 50 # inverse temperature for bose function in Sq(omega)
Now, copy all essential files to a directory for Susceptibility calculations and launch Suscept.py to solve BSE:
mkdir suscept cp *.klist *.qlist tvertex.dat gkloc.ni gloc.ni suscept.in suscept cd suscept Suscept.py suscept.in
You can find a good vertex from here and also other input files from this link this link:
mkdir good_vertex cp ni-sus-inp/* good_vertex cd good_vertex cp /home/cseg/scarf705/ni-suscept/tvertex.dat .
Here are our cuprates results published in this paper . You should be able to reproduce the results with the input files and the script. You can copy a directory with all converged results for NCO and run the script to see if the results match with what we show in our paper.
cp -r /home/cseg/scarf705/*.tar.gz .