ramess101
commented
6 years ago We now have a working basis_function class that generates a text file with the following format (example for n-alkanes):
sumr6_CH3 sumrlam_CH3 sumr6_CH2 sumrlam_CH2 sumr6_CH3CH2 sumrlam_CH3CH2
We have this array of basis functions (sumr6, sumrlam) for each snapshot. The energy is computed by simple matrix multiplication by the "Cmatrix" that contains:
C6_CH3 Clam_CH3 C6_CH2 Clam_CH2 C6_CH3CH2 Clam_CH3CH2
These figures show the very good agreement between the energies computed with basis functions and obtained from direct simulation. The red line is perfect agreement and the dashed lines are 5% error:
Zoomed in on the low U (vapor phase, low N) region:
To quantify the agreement, the average percent deviation is 1.35x10^-14 over the entire range of energies and it is 1.31x10^-14 for just the vapor region. This is sufficient evidence that our basis functions are working properly.