mirror of https://github.com/QMCPACK/qmcpack.git
d77f861fb8
|
||
|---|---|---|
| .. | ||
| C.BFD.xml | ||
| C2-H5.wfj.xml | ||
| C2-XML-traditional.wfj.xml | ||
| C2-XML.wfj.xml | ||
| C2.h5 | ||
| C2.structure.xml | ||
| C2_TZ_PP_H5_REF_MSDJ.s001.scalar.dat | ||
| C2_TZ_PP_LONG_H5_MSDJ.s001.scalar.dat | ||
| C2_TZ_PP_LONG_XML_MSDJ.s001.scalar.dat | ||
| C2_TZ_PP_SHORT_H5_MSDJ.s001.scalar.dat | ||
| C2_TZ_PP_SHORT_XML_MSDJ.s001.scalar.dat | ||
| C2_TZ_PP_XML_REF_MSDJ.s001.scalar.dat | ||
| README | ||
| qmc_ref_h5_msdj.out | ||
| qmc_ref_h5_msdj.xml | ||
| qmc_ref_xml_msdj.out | ||
| qmc_ref_xml_msdj.xml | ||
| vmc_long_pp_msdj-H5.out | ||
| vmc_long_pp_msdj-XML.out | ||
| vmc_short_pp_msdj-H5.out | ||
| vmc_short_pp_msdj-XML.out | ||
README
#C2_PP test targets the Multideterminant route of the code. The wavefunction data is stored in a XML format AND HDF5 Format. Both files should be run
Energy from QP:-10.9015995458600 Ha
0- The trial wavefunction was generated using quantum Package as described in tests/molecules/C2_pp/QuantumPackage/script.sh
This will generate 202 determinants.
The wavefunction is then dumped from QP as follow:
> qp_run save_for_qmcpack C2.ezfio > C2
The C2 file is a txt format recognized by convert4qmc. Also, a fully QMCPACK compatible HDF5 is now produced by QP (named QMC.h5). It contains all information to run.
For Both routes, a Jastrow (J1, J2 and J3) optimization is performed outside of this test. Jastrow parameters are provided in the wfj.xml files.
VMC energy is computed to evaluate the local energy and variance.
The long VMC reference run used to extrapolate the short run reference used the following VMC blocks:
<qmc method="vmc" move="pbyp" checkpoint="-1">
<estimator name="LocalEnergy" hdf5="no"/>
<parameter name="samplesperthread">1</parameter>
<parameter name="stepsbetweensamples">100</parameter>
<parameter name="substeps">5</parameter>
<parameter name="warmupSteps">20</parameter>
<parameter name="blocks">65500</parameter>
<parameter name="timestep">0.5</parameter>
<parameter name="usedrift">no</parameter>
</qmc>
However, In order to keep the size of the files small in the reference directory, an equivalent VMC block was used using 1 sample per thread and 4096 threads
<qmc method="vmc" move="pbyp" checkpoint="-1">
<estimator name="LocalEnergy" hdf5="no"/>
<parameter name="samplesperthread">1</parameter>
<parameter name="stepsbetweensamples">32800</parameter>
<parameter name="substeps">5</parameter>
<parameter name="warmupSteps">20</parameter>
<parameter name="blocks">200</parameter>
<parameter name="timestep">0.5</parameter>
<parameter name="usedrift">no</parameter>
</qmc>
1- XML route.
Inputs (C2.structure.xml and C2.wfj-XML.xml were generated from the text file (C2) from Quantum_Package)
> convert4qmc -QP C2 -production
> mv C2.wfj.xml C2.wfj-XML.xml
Input file name is qmc_ref_xml_msdj.xml
The long VMC run leads to the following statistics:
LocalEnergy = -11.059755 +/- 0.000013
Variance = 0.16251 +/- 0.00046
Kinetic = 7.72378 +/- 0.00016
LocalPotential = -18.78354 +/- 0.00016
ElecElec = 12.142072 +/- 0.000083
LocalECP = -39.43847 +/- 0.00024
NonLocalECP = 1.69867 +/- 0.00011
IonIon = 6.81 +/- 0.00
LocalEnergy_sq = 122.48071 +/- 0.00055
BlockWeight = 16384.00 +/- 0.00
BlockCPU = 0.009885908 +/- 0.000000061
AcceptRatio = 0.4704575 +/- 0.0000027
Efficiency = 2004402.19 +/- 0.00
TotalTime = 647.37 +/- 0.00
TotalSamples = 1072889856 +/- 0
The equivalent run leads to the following statistics:
LocalEnergy = -11.059771 +/- 0.000020
Variance = 0.16235 +/- 0.00027
Kinetic = 7.72351 +/- 0.00026
LocalPotential = -18.78328 +/- 0.00025
ElecElec = 12.14210 +/- 0.00013
LocalECP = -39.43834 +/- 0.00038
NonLocalECP = 1.69877 +/- 0.00017
IonIon = 6.81 +/- 0.00
LocalEnergy_sq = 122.48089 +/- 0.00051
BlockWeight = 2686976.00 +/- 0.00
BlockCPU = 1.633350 +/- 0.000015
AcceptRatio = 0.4704575 +/- 0.0000043
Efficiency = 12259.38 +/- 0.00
TotalTime = 277.67 +/- 0.00
TotalSamples = 456785920 +/- 0
2- HDF5 route
Input file name is qmc_ref_h5_msdj.xml
The long VMC run leads to the following statistics:
LocalEnergy = -11.059757 +/- 0.000013
Variance = 0.16251 +/- 0.00043
Kinetic = 7.72336 +/- 0.00016
LocalPotential = -18.78311 +/- 0.00016
ElecElec = 12.142051 +/- 0.000086
LocalECP = -39.43812 +/- 0.00025
NonLocalECP = 1.69877 +/- 0.00011
IonIon = 6.81 +/- 0.00
LocalEnergy_sq = 122.48076 +/- 0.00050
BlockWeight = 16384.00 +/- 0.00
BlockCPU = 0.009946695 +/- 0.000000061
AcceptRatio = 0.4704592 +/- 0.0000028
Efficiency = 2047542.99 +/- 0.00
TotalTime = 651.51 +/- 0.00
TotalSamples = 1073152000 +/- 0
The equivalent run leads to the following statistics:
LocalEnergy = -11.059791 +/- 0.000019
Variance = 0.16245 +/- 0.00034
Kinetic = 7.72404 +/- 0.00024
LocalPotential = -18.78383 +/- 0.00026
ElecElec = 12.14216 +/- 0.00012
LocalECP = -39.43875 +/- 0.00035
NonLocalECP = 1.69857 +/- 0.00017
IonIon = 6.81 +/- 0.00
LocalEnergy_sq = 122.48142 +/- 0.00052
BlockWeight = 2686976.00 +/- 0.00
BlockCPU = 1.635521 +/- 0.000015
AcceptRatio = 0.4704577 +/- 0.0000047
Efficiency = 13010.96 +/- 0.00
TotalTime = 282.95 +/- 0.00
TotalSamples = 464846848 +/- 0