Data for "Overcoming the Memory Bottleneck in Auxiliary Field Quantum Monte Carlo
Simulations with Interpolative Separable Density Fitting."
=====================
In this dataset you will find:
1. The analysed data and plotting scripts necessary to reproduce the figures in the above
paper.
2. The raw data produced from qmcpack calculations:
- convergence/ (with respect to ISDF rank parameter c):
- 2x2x2/:
- DZ/
- TZ/
- 3x3x3/
- DZ/
- TZ/
- cold_curve/ (AFQMC results for cold curve)
- sparse/
- thc/
- cohesive_energy/
- solid/
- 2x2x2
- 3x3x3
- 4x4x4
- atom/
- cc-pvdz
- cc-pvtz
- cc-pvqz
The general procedure for running qmcpack using ISDF for the integrals is as follows:
1. Run a mean field calculation to generate the MO matrix. In All directories you should
find the checkpoint file generated by pyscf (
.dump) which will contain the necessary data.
The script to generate this data will be called kpoints.py.
2. Generate the supercell MOs and trial wavefunction from mean field dump file. This is
generated using the script dump_wfn.py which will be distributed with qmcpack with the
thc++ code in the future.
3. Run thc++ to generate the ISDF factorization. Input files for this step are located in
the calculation subdirectories (typically called input.json).
4. Run qmcpack using the xml input file (also located in the subdirectories).
5. Analyse the output (.scalar.dat) using analysis script.