MultiOptPy 1.20.2__py3-none-any.whl

multioptpy/WFAnalyzer/MO_analysis.py

@@ -0,0 +1,104 @@
+import numpy as np
+import matplotlib.pyplot as plt
+# NRO analysis
+# ref. S. Ebisawa, M. Hasebe, T. Tsutsumi, T. Tsuneda, and T. Taketsugu, Phys. Chem. Chem. Phys. 24, 3532 (2022).
+
+
+class NROAnalysis:
+    def __init__(self, **config):
+        # Currently this method is only available for tblite.
+        # Visualization of NRO is not implemented.
+        # Additionally, calculation of 1st and 2nd derivatives of orbital energy is implemented. (ref. Phys. Chem. Chem. Phys., 2018, 20, 14211-14222)   
+        self.config = config
+        self.method = self.config["xtb"]
+        self.element_list = self.config["element_list"]
+        self.electric_charge_and_multiplicity = self.config["electric_charge_and_multiplicity"]
+        
+        self.numerical_delta = 0.0001
+        self.LAMBDA_list = []
+        self.first_deriv_orbital_ene_list = []
+        self.second_deriv_orbital_ene_list = []
+        self.file_directory = self.config["file_directory"]
+        return
+        
+    def save_results(self, energy_list, bias_energy_list):
+        # LAMBDA is singular value describing magnitude of change in a molecular orbitals when atoms are moved in a particular direction.
+      
+        num_list = np.array(range(len(self.LAMBDA_list)))
+        
+        self.LAMBDA_list = np.array(self.LAMBDA_list)
+        self.LAMBDA_list = (self.LAMBDA_list - np.min(self.LAMBDA_list)) / (np.max(self.LAMBDA_list) - np.min(self.LAMBDA_list))
+  
+        fig = plt.figure()
+        ax1 = fig.add_subplot(111)  
+        ax1.plot(num_list, energy_list, label='Energy [kcal/mol]',color="g")
+        ax1.plot(num_list, bias_energy_list, label='Bias Energy [kcal/mol]',color="b")
+        ax2 = ax1.twinx()
+        ax2.plot(num_list, self.LAMBDA_list,label='LAMBDA', color="r")
+        #ax2.set_yscale('log')
+        h1, l1 = ax1.get_legend_handles_labels()
+        h2, l2 = ax2.get_legend_handles_labels()
+        ax1.legend(h1+h2, l1+l2, loc='upper right')
+
+        ax1.set_xlabel('# iteration')
+        ax1.set_ylabel("Energy [kcal/mol]")
+        ax2.set_ylabel('LAMBDA')
+        #if np.max(self.LAMBDA_list[i]) > 1.0:
+        #    ax2.axis([num_list[0], num_list[-1], 0, 1.0])
+        plt.title("LAMBDA ")
+        plt.tight_layout()
+        plt.savefig(self.file_directory+"NRO_lambda_plot.png", format="png", dpi=300)
+        plt.close()
+        with open(self.file_directory+"NRO_lambda_plot.csv", "w") as f:
+            f.write("#ITR. , energy [kcal/mol], bias energy [kcal/mol], LAMBDA\n")
+            for j in range(len(self.LAMBDA_list)):
+                f.write(str(num_list[j])+","+str(energy_list[j]-energy_list[0])+","+str(bias_energy_list[j]-bias_energy_list[0])+","+str(self.LAMBDA_list[j])+"\n")
+                
+            
+        return
+    
+    
+    
+    def run(self, SP, geom_num_list, move_vector):#geom_num_list :Bohr
+        print("### Natural Reaction Orbital analysis ###")
+        print("# This method is only available for tblite (xTB).")
+        displacement_for_numerical_differential = self.numerical_delta * move_vector / np.linalg.norm(move_vector)
+        neutral_orbital_coefficients = SP.orbital_coefficients
+        overlap_matrix = SP.overlap_matrix
+        orbital_energy = SP.orbital_energies
+        #plus  positions, element_number_list, electric_charge_and_multiplicity, method
+        _, _, _ = SP.single_point_no_directory(geom_num_list+displacement_for_numerical_differential, self.element_list, self.electric_charge_and_multiplicity, self.method)
+        p_orbital_coefficients = SP.orbital_coefficients
+        p_orbita_energy = SP.orbital_energies
+        
+        #minus positions
+        _, _, _ = SP.single_point_no_directory(geom_num_list-displacement_for_numerical_differential, self.element_list, self.electric_charge_and_multiplicity, self.method)
+        m_orbital_coefficients = SP.orbital_coefficients
+        m_orbital_energy = SP.orbital_energies
+        
+        
+        first_derivative_orbital_coefficients = (p_orbital_coefficients - m_orbital_coefficients) / (self.numerical_delta * 2)
+        first_derivative_orbital_energy = (p_orbita_energy - m_orbital_energy) / (self.numerical_delta * 2)
+        second_derivative_orbital_energy = (p_orbita_energy + m_orbital_energy -2 * orbital_energy) / (self.numerical_delta ** 2)
+        first_response_matrix = np.dot(np.conjugate(neutral_orbital_coefficients.T), np.dot(overlap_matrix, first_derivative_orbital_coefficients))
+        L, LAMBDA, R = np.linalg.svd(first_response_matrix)
+        
+        N_R = np.conjugate(R.T)
+        N_L = np.conjugate(L.T)
+        #print(N_L, LAMBDA, N_R)
+        #print("LAMBDA:", LAMBDA)
+        sum_of_LAMBDA = np.sum(LAMBDA)
+        self.LAMBDA_list.append(sum_of_LAMBDA)
+        self.first_deriv_orbital_ene_list.append(first_derivative_orbital_energy)
+        self.second_deriv_orbital_ene_list.append(second_derivative_orbital_energy)
+        
+        # temporary save
+        with open(self.file_directory+"NRO_lambda_plot.csv", "a") as f:
+            f.write(str(sum_of_LAMBDA)+"\n")
+        with open(self.file_directory+"1st_derivative_orbital_energy_plot.csv", "a") as f:
+            f.write(str(",".join(list(map(str, first_derivative_orbital_energy.tolist()))))+"\n")
+        with open(self.file_directory+"2nd_derivative_orbital_energy_plot.csv", "a") as f:
+            f.write(str(",".join(list(map(str, second_derivative_orbital_energy.tolist()))))+"\n")
+        return
+    
+