Examples

In these section, we will use some typical examples to show you how to apply the Python binding to control the atomic eigenvalue problem solver, CT-HYB and HF-QMC quantum impurity solvers.

Case 1: JASMINE component (atomic eigenvalue problem solver)

In the following, we will use an example to show you how to use API to control the JASMINE code. When you want to run your Python code, you have to ensure that pyjasmine.so is in correct PATH. Or else the Python will complain that it can not find iQIST.

(1) Import pyjasmine

import pyjasmine

You have to ensure that the pyjasmine package is in the sys.path. For example, you can use the following code to modify sys.path

sys.path.append('../../src/tools/jasmine/')

(2) Configure the atomic eigenvalue problem solver

You have to setup the key parameters for the atomic eigenvalue problem solver, and write them down to the atom.config.in file. Now you can do that manually. On the other hand, we provide a powerful Python module to facilitate this work (see iqist/src/tools/hibiscus/script/u_atomic.py).

(3) Initialize the atomic eigenvalue problem solver

pyjasmine.cat_init_atomic() # there is no parameter for cat_init_atomic()

(4) Start the atomic eigenvalue problem solver

pyjasmine.cat_exec_atomic()

(5) Close the atomic eigenvalue problem solver

pyjasmine.cat_stop_atomic()

(6) Access the computational results

You have to write your own Python codes to access the results.

Case 2: CT-HYB quantum impurity solvers

In the following, we will use the AZALEA code as an example to show how to use API to control it. When you want to run your Python code, you have to ensure that pyiqist.so is in correct PATH. Or else the Python will complain that it can not find iQIST.

(1) Import MPI support

Here we can use the pyalps.mpi package or mpi4py package to provide the MPI support, such as,

import pyalps.mpi

or

from mpi4py import MPI

We recommend to use the mpi4py package since it is included in the scipy package already. The above code will also start the MPI running environment implicitly.

(2) Import pyiqist

import pyiqist

You have to ensure that the pyiqist package is in the sys.path. For example, you can use the following code to modify sys.path

sys.path.append('../../src/ctqmc/azalea/')

(3) Configure the CT-HYB quantum impurity solver

You have to setup the parameters for the CT-HYB quantum impurity solver, and write them down to the solver.ctqmc.in file. Now you can do that manually. On the other hand, we provide a powerful Python module to facilitate this work (see iqist/src/tools/hibiscus/script/u_ctqmc.py).

(4) Initialize the CT-HYB quantum impurity solver

pyiqist.cat_init_ctqmc(my_id, num_procs)

Here my_id means the rank for current process, and num_procs means number of processes. If you are using the mpi4py package to provide MPI support, then you can use the following code to init the CT-HYB quantum impurity solver.

comm = MPI.COMM_WORLD
pyiqist.cat_init_ctqmc(comm.rank, comm.size)

(5) setup hybf, symm, eimp, and ktau

For examples:

mfreq = 8193
norbs = 2
size_t = mfreq * norbs * norbs
hybf = numpy.zeros(size_t, dtype = numpy.complex, order = 'F')
...
pyiqist.cat_set_hybf(size_t, hybf)

NOTE:

1. We strongly recommend to select the Fortran rule to manage the numpy array memory.

2. This step is optional, because the CT-HYB quantum impurity solver will provide default values for hybf, symm, eimp, and ktau or read them from external disk files.

(6) Start the CT-HYB quantum impurity solver

pyiqist.cat_exec_ctqmc(i)

Here $i$ is the current iteration number.

(7) Retrieve the calculated results

For examples:

size_t = norbs
nmat = pyiqist.cat_get_nmat(size_t)
print nmat

size_t = mfreq * norbs * norbs
grnf = pyiqist.cat_get_grnf(size_t)
grnf = numpy.reshape(grnf, (mfreq, norbs, norbs), order = 'F')

NOTE:

You have to pay attention to the array order when you try to use numpy.reshape() to convert a 1-D array to a 3-D array.

(8) Close the CT-HYB quantum impurity solver

pyiqist.cat_stop_ctqmc()

Case 3: HF-QMC quantum impurity solver

In the following, we will use the DAISY code as an example to show how to use API to control it. When you want to run your Python code, you have to ensure that pydaisy.so is in correct PATH. Or else the Python will complain that it can not find iQIST.

(1) Import MPI support

Here we can use the pyalps.mpi package or mpi4py package to provide the MPI support, such as,

import pyalps.mpi

or

from mpi4py import MPI

We recommend to use the mpi4py package since it is included in the scipy package already. The above code will also start the MPI running environment implicitly.

(2) Import pydaisy

import pydaisy

You have to ensure that the pydaisy package is in the sys.path. For example, you can use the following code to modify sys.path

sys.path.append('../../src/hfqmc/daisy/')

(3) Configure the HF-QMC quantum impurity solver

You have to setup the parameters for the HF-QMC quantum impurity solver, and write them down to the solver.hfqmc.in file. Now you can do that manually. On the other hand, we provide a powerful Python module to facilitate this work (see iqist/src/tools/hibiscus/script/u_hfqmc.py).

(4) Initialize the HF-QMC quantum impurity solver

pydaisy.cat_init_hfqmc(my_id, num_procs)

Here my_id means the rank for current process, and num_procs means number of processes. If you are using the mpi4py package to provide MPI support, then you can use the following code to init the HF-QMC quantum impurity solver.

comm = MPI.COMM_WORLD
pydaisy.cat_init_hfqmc(comm.rank, comm.size)

(5) Setup wssf, symm, eimp, and ktau

For examples:

mfreq = 8193
norbs = 2
size_t = mfreq * norbs
wssf = numpy.zeros(size_t, dtype = numpy.complex, order = 'F')
...
pydaisy.cat_set_wssf(size_t, wssf)

NOTE:

1. We strongly recommend to select the Fortran rule to manage the numpy array memory.

2. This step is optional, because the HF-QMC quantum impurity solver will provide default values for wssf, symm, eimp, and ktau or read them from external disk files.

(6) Start the HF-QMC quantum impurity solver

pydaisy.cat_exec_hfqmc(i)

Here $i$ is the current iteration number.

(7) Retrieve the calculated results

For examples:

size_t = norbs
nmat = pydaisy.cat_get_nmat(size_t)
print nmat

size_t = mfreq * norbs
grnf = pydaisy.cat_get_grnf(size_t)
grnf = numpy.reshape(grnf, (mfreq, norbs), order = 'F')

NOTE:

You have to pay attention to the array order when you try to use numpy.reshape() to convert a 1-D array to a 2-D array.

(8) Close the HF-QMC quantum impurity solver

pydaisy.cat_stop_hfqmc()