Differences

This shows you the differences between two versions of the page.

--- exercises:2019_conexs_newcastle:ex4 [2019/09/11 03:22] – [Part 1: PDOS of liquid water] abussy
+++ exercises:2019_conexs_newcastle:ex4 [2020/08/21 10:15] (current) – external edit 127.0.0.1
@@ Line 3: / Line 3: @@
 In this exercise, we will look at a more complicated example than in Exercise 1, namely liquid water. We start by computing the PDOS and compare it the single H$_2$O molecules. This time we will have many more orbitals so in order to visualize our results, we will apply a Gaussian broadening to the results. After, we will do a short Born-Oppenheimer MD run and learn how the basic input and output.
-======Part 1: PDOS of liquid water=====
+======Part 1: PDOS and spectra broadening of liquid water=====
  In this part of the exerise we will compute the PDOS. Our model water will contain 64 H$_2$O molecules in a 12 Å box under periodic boundary conditions. The input is very similar to the first exercise, but since we now have a much more coordinates, it is convinient to import them from a separate file. Below is a an example input you can use to perform a single-point calculation extracting the PDOS.
@@ Line 298: / Line 298: @@
 &END TOPOLOGY
 </code>
-where we specify the filename, format, and that we should not divide our atoms into molecules. When you have finished running the above input, try to use the following python script to convolve your spectra with Gaussian functions. Note that you might have to change the parameters to get pleasing results.
+where we specify the filename, format, and that we should not divide our atoms into molecules. When you have finished running the above input, try to use the following python script to convolve your spectra with Gaussian functions. You do this by first loading Python3 by typing
+<code>
+module load Anaconda3/5.0.1
+</code>
+and then typing
+<code>
+python pdos_conv.py
+</code>
+Note that you might have to change the parameters at the top of the file to get pleasing results. If a new window with the plot does not open, it is also saved in the current directory as "pdos.png".
 <code - pdos_conv.py>
 import numpy as np
-import matplotlib.pyplot as plt
 import os
+import matplotlib as mpl
+mpl.use('Agg')
+import matplotlib.pyplot as plt
 # Parameters
-FWHM = 0.3   # Full width half maximum
+FWHM = 2.5   # Full width half maximum
 Emin=-16.    # Starting energy
 Emax= 4.     # -----
-DE = 0.01    # Energy intervals
+DE = 0.01    # Fineness of the energies to convolve to
 dE=np.arange(Emin,Emax,DE)
 E_shift = 0
 filename = 'liquid_water-k1-1.pdos'
-###
 def plot_pdos(txtfile):
@@ Line 359: / Line 367: @@
     plt.show()
 def convolve(E,PDOS):
@@ Line 372: / Line 379: @@
 plot_pdos(filename)
+</code>
+======Part 2: MD of liquid water=====
+We will now attempt to do a short Born-Oppenheimer MD run on our system. The majority of the input from Part 1 can be reused while changing the run type to
+<code>
+RUN_TYPE MD
+</code>
+in the &GLOBAL section as well as adding the section
+<code>
+&MOTION
+  &MD
+    ENSEMBLE NVT
+    STEPS 15
+    TIMESTEP 1
+    TEMPERATURE 300.0
+    &THERMOSTAT
+      REGION MASSIVE
+      TYPE CSVR
+      &CSVR
+        TIMECON 5
+      &END CSVR
+    &END THERMOSTAT
+  &END MD
+&END MOTION
+</code>
+The parameters in the MD section has been selected to finish in a few minutes, but feel free to experiment with different values and get a feel for how they work. To not print too much, also modify the &PDOS section to
+<code>
+&PDOS
+  FILENAME ./liquid_water
+  APPEND
+  &EACH
+    MD 3
+  &END
+&END PDOS
+</code>
+which means that we only print the PDOS every third step, as well as appending the spectra instead of overwriting them.
+When you have finished the calculation, you should have created some new files, including
+    * liquid_water-1.ener - containing information about energies, temperature, and timings
+    * liquid_water-pos-1.xyz - containing your trajectory
+Examine liquid_water-1.ener and make sure that the temperature and total energy remains reasonably stable during the run. As a final exercise, use a slightly modified python code from Part 1 to see how the PDOS spectrum changes with time.
+<code - pdos_conv_md.py>
+import numpy as np
+import os
+import matplotlib as mpl
+mpl.use('Agg')
+import matplotlib.pyplot as plt
+# Parameters
+FWHM = 0.4   # Full width half maximum
+Emin=-16.    # Starting energy
+Emax= 4.     # -----
+DE = 0.01    # Energy intervals
+dE=np.arange(Emin,Emax,DE)
+E_shift = 0
+filename = 'liquid_water-k1-1.pdos'
+def plot_pdos(txtfile):
+    if not os.path.isfile(txtfile):
+        print("Warning: File does not exist!")
+        return np.zeros(np.shape(dE)[0])
+    PDOS = np.asarray(open(txtfile).read().split())
+    if 'f' in PDOS:
+        start = 26
+        Norb = 4+3
+    elif 'd' in PDOS:
+        start = 25
+        Norb = 3+3
+    elif 'p' in PDOS:
+        start = 24
+        Norb = 2+3
+    else:
+        start = 23
+        Norb = 1+3
+    L = int(PDOS[-Norb])
+    N = np.where(PDOS=='Projected')[0]-1
+    for i in range(N.shape[0]-1):
+        pdos = PDOS[N[i]:N[i+1]]
+        pdos = pdos[start:]
+        pdos = np.asarray(np.split(pdos,L))
+        E = pdos[:,1].astype(np.float)*27.2114 #a.u. -> eV
+        pdos = pdos[:,3:].astype(float)
+        if i == 0:
+            pdos_tot = np.zeros((pdos.shape))
+        pdos_tot += pdos / N.shape[0]
+        s, p = convolve(E,pdos)
+        if i == 0:
+            yshift = np.max(np.append(s,p))
+        plt.plot(dE+E_shift,s-i*yshift, c='blue',linestyle='-',linewidth=2.,alpha=0.7)
+        plt.plot(dE+E_shift,p-i*yshift, c='red',linestyle='-',linewidth=2.,alpha=0.7)
+    s, p = convolve(E,pdos_tot)
+    plt.text(-14,-(i+1.5)*yshift,'Sum')
+    plt.plot(dE+E_shift,s-(i+2)*yshift, c='blue',linestyle='-',linewidth=2.,alpha=0.7)
+    plt.plot(dE+E_shift,p-(i+2)*yshift, c='red',linestyle='-',linewidth=2.,alpha=0.7)
+    plt.yticks([])
+    plt.ylabel('PDOS (arbitrary units)')
+    plt.xlabel('Binding energy (eV)')
+    plt.xticks(np.arange(-15,5,5),-np.arange(-15,5,5))
+    plt.xlim([-16,3])
+    plt.title('FWHM = ' + str(FWHM) + ' eV')
+    plt.plot([100,101],[0,0], c='blue',linestyle='-',linewidth=2., label='O s')
+    plt.plot([100,101],[0,0], c='red',linestyle='-',linewidth=2., label = 'O p')
+    plt.legend(loc=1)
+    plt.savefig('pdos.png')
+    plt.show()
+def convolve(E,PDOS):
+    sigma = 1 / (2 * np.sqrt(2 * np.log(2) ) ) * FWHM
+    conv = np.zeros((len(dE),2))
+    for j in range(conv.shape[1]):
+        W = PDOS[:,j]
+        for i,e in enumerate(E):
+            conv[:,j] += np.exp(-(dE-e)**2 / (2 * sigma**2))*W[i]
+    return conv[:,0], conv[:,1]
+plot_pdos(filename)
 </code>