Merge pull request #764 from Paul-St-Young/dat-h5

Add sofk estimator test and manual entry about how estimator works

manual/backflow_implementation.tex

@@ -1,5 +1,3 @@
-\newcommand{\bs}{\boldsymbol}
-\newcommand{\tr}{\text{tr}}
 \section{Slater-Backflow Wavefunction Implementation Details}
 
 For simplicity, consider $N$ identical fermions of the same spin (e.g. up electrons) at spatial locations $\{\bs{r}_1,\bs{r}_2,\dots,\bs{r}_{N}\}$. Then the Slater determinant can be written as

manual/developing.tex

@@ -6,4 +6,5 @@ \chapter{Development Guide}
 \input{coding_standards}
 
 \input{estimator_implementation}
+\input{estimator_manager}
 \input{backflow_implementation}

manual/qmcpack_manual.tex

@@ -164,6 +164,8 @@
 \newcommand{\overlap}[2]{\langle #1 \lvert #2 \rangle}
 \newcommand{\operator}[3]{\ket{#1} #2 \bra{#3}}
 \newcommand{\idop}{\hat{\mathbb{1}}}
+\newcommand{\bs}{\boldsymbol}
+\newcommand{\tr}{\text{tr}} % trace
 
 
 \begin{document}

tests/estimator/CMakeLists.txt

@@ -46,3 +46,33 @@ if (add_test)
     latdev_check.py
   )
 endif()
+
+set(sofk_python_reqs numpy;pandas;h5py)
+CHECK_PYTHON_REQS(sofk_python_reqs estimator-sofk add_test)
+
+if (add_test)
+  SIMPLE_RUN_AND_CHECK(estimator-sofk_pbyp-properties
+    "${CMAKE_SOURCE_DIR}/tests/estimator/sofk"
+    pbyp_dat-h5.xml
+    1 16
+    check_properties_h5dat.py dat-h5_pbyp
+  )
+  SIMPLE_RUN_AND_CHECK(estimator-sofk_pbyp-collectables
+    "${CMAKE_SOURCE_DIR}/tests/estimator/sofk"
+    pbyp_dat-h5.xml
+    1 16
+    check_collectables_h5dat.py dat-h5_pbyp
+  )
+  SIMPLE_RUN_AND_CHECK(estimator-sofk_allp-properties
+    "${CMAKE_SOURCE_DIR}/tests/estimator/sofk"
+    allp_dat-h5.xml
+    1 16
+    check_properties_h5dat.py dat-h5_allp
+  )
+  SIMPLE_RUN_AND_CHECK(estimator-sofk_allp-collectables
+    "${CMAKE_SOURCE_DIR}/tests/estimator/sofk"
+    allp_dat-h5.xml
+    1 16
+    check_collectables_h5dat.py dat-h5_allp
+  )
+endif()

tests/estimator/sofk/allp_dat-h5.xml

@@ -0,0 +1,91 @@
+<?xml version="1.0"?>
+<simulation>
+   <project id="dat-h5_allp" series="0">
+      <application name="qmcapp" role="molecu" class="serial" version="1.0"/>
+   </project>
+   <qmcsystem>
+      <simulationcell>
+         <parameter name="lattice" units="bohr">
+                  3.77945227        0.00000000        0.00000000
+                 -0.00000000        3.77945227        0.00000000
+                 -0.00000000       -0.00000000        3.77945227
+         </parameter>
+         <parameter name="bconds">
+            p p p
+         </parameter>
+         <parameter name="LR_dim_cutoff"       >    15                 </parameter>
+      </simulationcell>
+      <particleset name="e" random="yes">
+         <group name="u" size="1" mass="1.0">
+            <parameter name="charge"              >    -1                    </parameter>
+            <parameter name="mass"                >    1.0                   </parameter>
+         </group>
+         <group name="d" size="1" mass="1.0">
+            <parameter name="charge"              >    -1                    </parameter>
+            <parameter name="mass"                >    1.0                   </parameter>
+         </group>
+      </particleset>
+      <particleset name="ion0">
+         <group name="H" size="2" mass="1837.36221934">
+            <parameter name="charge"              >    1                     </parameter>
+            <parameter name="valence"             >    1                     </parameter>
+            <parameter name="atomicnumber"        >    1                     </parameter>
+            <parameter name="mass"                >    1837.36221934            </parameter>
+            <attrib name="position" datatype="posArray" condition="0">
+                     0.00000000        0.00000000        0.00000000
+                     1.88972614        1.88972614        1.88972614
+            </attrib>
+         </group>
+      </particleset>
+      <wavefunction name="psi0" target="e">
+         <determinantset type="einspline" href="pwscf.pwscf.h5" tilematrix="1 0 0 0 1 0 0 0 1" twistnum="0" source="ion0" meshfactor="1.0" precision="float">
+            <slaterdeterminant>
+               <determinant id="updet" size="1">
+                  <occupation mode="ground" spindataset="0"/>
+               </determinant>
+               <determinant id="downdet" size="1">
+                  <occupation mode="ground" spindataset="0"/>
+               </determinant>
+            </slaterdeterminant>
+         </determinantset>
+         <jastrow type="One-Body" name="J1" function="bspline" source="ion0" print="yes">
+            <correlation elementType="H" size="8" cusp="1.0">
+               <coefficients id="eH" type="Array">                  
+0.00206602038 -0.002841926986 0.0036266191 -0.001913930279 8.457152991e-06 
+0.0007380321824 3.635172529e-05 0.0001299635851
+               </coefficients>
+            </correlation>
+         </jastrow>
+         <jastrow type="Two-Body" name="J2" function="bspline" print="yes">
+            <correlation speciesA="u" speciesB="d" size="8">
+               <coefficients id="ud" type="Array">                  
+0.5954603818 0.5062051797 0.3746940461 0.2521010502 0.1440163317 0.07796688253 
+0.03804420551 0.01449320872
+               </coefficients>
+            </correlation>
+         </jastrow>
+      </wavefunction>
+      <hamiltonian name="h0" type="generic" target="e">
+         <pairpot type="coulomb" name="ElecElec" source="e" target="e"/>
+         <pairpot type="coulomb" name="IonIon" source="ion0" target="ion0"/>
+         <pairpot type="coulomb" name="ElecIon" source="ion0" target="e"/>
+         <estimator type="sk" name="sk" hdf5="no"/>
+         <estimator type="sk" name="h5sk" hdf5="yes"/>
+      </hamiltonian>
+   </qmcsystem>
+   <qmc method="vmc" move="byp">
+      <parameter name="walkers"             >    64              </parameter>
+      <parameter name="blocks"              >    16            </parameter>
+      <parameter name="steps"               >    2            </parameter>
+      <parameter name="subSteps"            >    2               </parameter>
+      <parameter name="timestep"            >    2.0             </parameter>
+      <parameter name="warmupSteps"         >    16             </parameter>
+   </qmc>
+   <qmc method="dmc" move="byp">
+      <parameter name="targetwalkers"       >    64              </parameter>
+      <parameter name="blocks"              >    16            </parameter>
+      <parameter name="steps"               >    1            </parameter>
+      <parameter name="timestep"            >    2.0             </parameter>
+      <parameter name="warmupSteps"         >    0             </parameter>
+   </qmc>
+</simulation>

tests/estimator/sofk/check_collectables_h5dat.py

@@ -0,0 +1,92 @@
+#!/usr/bin/env python
+import sys
+import os
+import h5py
+import numpy as np
+from check_properties_h5dat import read
+
+
+def get_last_sk(fdat,fh5):
+  """ extract S(k) at the longest k vector from scalar.dat and stat.h5
+
+  Args:
+    fdat (str): name of scalar.dat file
+    fh5  (str): name of stat.h5 file
+  Return:
+    tuple: (myy, h5y), S(k_max) at each block from scalar.dat and stat.h5
+  """
+
+  # get S(k) from scalar.dat
+  df = read(fdat)
+  sk_cols = [col for col in df.columns if col.startswith('sk')]
+  myy = df[sk_cols[-1]].values
+
+  # get S(k) from stat.h5
+  fp = h5py.File(fh5, 'r')
+  h5y = fp['h5sk/value'].value.T[-1]
+  fp.close()
+
+  return myy, h5y
+# end def
+
+
+def show_scalar_trace(data, seriesl):
+  import matplotlib.pyplot as plt
+  method_map = {0:'VMC',1:'DMC'}
+  fig,ax_arr = plt.subplots(1, 2, sharey=True)
+  ax_arr[0].set_ylabel('S(k->inf)')
+  iplot = 0
+  for iseries in seriesl:
+    ax = ax_arr[iplot]
+    ax.set_title(method_map[iseries])
+    ax.set_xlabel('block')
+    ax.set_ylim(0.3, 1.2)
+
+    entry = data[iseries]
+    daty  = entry['daty']
+    h5y   = entry['h5y']
+
+    sline = ax.plot(daty)
+    hline = ax.plot(h5y, ls='--', lw=2, alpha=0.8)
+
+    ax.legend(
+      handles = [sline[0], hline[0]]
+     ,labels  = ['scalar.dat', 'stat.h5']
+     ,loc=0
+    )
+
+    iplot += 1
+  # end for iseries
+  plt.show()
+
+
+if __name__ == '__main__':
+
+  prefix = sys.argv[1]
+  seriesl= [0,1]  # a list of series IDs to check
+
+  # check Properties v.s. Collectables
+  collectable_success_map = {}
+  data = {}
+  for iseries in seriesl:
+
+    # define files to read
+    fdat = '%s.s00%d.scalar.dat' % (prefix, iseries)
+    fh5  = '%s.s00%d.stat.h5' % (prefix, iseries)
+
+    daty, h5y = get_last_sk(fdat, fh5)
+    success   = np.allclose(daty, h5y, atol=0.1)
+    collectable_success_map[iseries] = success
+
+    # save data for plotting
+    data[iseries] = {'daty':daty, 'h5y':h5y}
+  # end for 
+  all_success = np.all( collectable_success_map.values() )
+
+  if all_success:
+    sys.exit(0)
+  else:
+    #show_scalar_trace(data, seriesl)
+    sys.exit(1)
+
+# end __main__

tests/estimator/sofk/check_properties_h5dat.py

@@ -0,0 +1,96 @@
+#!/usr/bin/env python
+import os
+import sys
+import h5py
+import numpy as np
+import pandas as pd
+
+
+def read(fdat):
+  """ read the scalar.dat file in table format readable by numpy.loadtxt.
+
+   The header line should start with '#' and contain column labels.
+
+  Args:
+    dat_fname (str): name of input file
+  Return:
+    pd.DataFrame: df containing the table of data
+  """
+  with open(fdat, 'r') as fp:
+    header = fp.readline()
+  # end with
+  cols = header.replace('#', '').split()
+  df = pd.read_table(fdat, sep='\s+', comment='#', header=None, names=cols)
+  return df
+# end def read
+
+
+def compare_columns_dat_h5(fdat, fh5):
+  """ compare mutual data columns in scalar.dat and stat.h5 files
+
+  Args:
+    fdat (str): name of scalar.dat file
+    fh5  (str): name of stat.h5 file
+  Return:
+    dict: a dictionary holding mutual columns names as key
+  """
+
+  # open database
+  df = read(fdat)
+  dat_cols = df.columns
+
+  fp = h5py.File(fh5,'r')
+  h5_cols = fp.keys()
+
+  # compare mutual columns in .dat v.s. .h5
+  agree_map = {}  # keep track of which columns agree
+  for col in h5_cols:
+    if col not in dat_cols:
+      continue
+
+    # check if col agree between .dat and .h5
+
+    # get .h5 values
+    h5_loc = os.path.join(col, 'value')
+    h5y  = fp[h5_loc].value[:,-1]
+
+    # get .dat values
+    daty = df.loc[:,col].values
+    agree_map[col] = np.allclose(h5y,daty)
+  # end for col
+
+  # close database
+  fp.close()
+
+  if len(agree_map) == 0:
+    raise RuntimeError('%s and %s have no mutual column' % (fdat, fh5))
+
+  return agree_map
+# end def
+
+
+if __name__ == '__main__':
+
+  prefix = sys.argv[1]
+  seriesl= [0,1]
+
+  # check Properties
+  series_success_map = {}
+  for iseries in seriesl:
+
+    # define files to read
+    fdat = '%s.s00%d.scalar.dat' % (prefix, iseries)
+    fh5  = '%s.s00%d.stat.h5' % (prefix, iseries)
+
+    agree_map = compare_columns_dat_h5(fdat, fh5)
+    success = np.all( agree_map.values() )
+    series_success_map[iseries] = success
+  # end for iseries
+
+  all_success = np.all( series_success_map.values() )
+  if all_success:
+    sys.exit(0)
+  else:
+    sys.exit(1)
+
+# end __main__