How does POLYMAT evaluate the norm of the deviation between data and model for the spectrum?

Each experimental curve c is known as a list of (Xi,Yi_e), for i = 1,n. For each Xi, the model prediction Yi_p of the corresponding property is evaluated by POLYMAT. The distance d between experimental curves and model properties is evaluated as
d = sum_on_all_curves_c {weight[c] * error[c]}
where the error[c] on curve c is evaluated as: error[c] = (1/n) * sum_on_all_points {[log(Yi_e) - log(Yi_p)]^2}, while weight[c] is a weight for that curve, as selected by the user. During iterative process, POLYMAT identifies parameters that minimize the above distance d.

Show Form
No comments yet. Be the first to add a comment!