Table 4.

The algorithm of sensitive parameter selection using the LASSO method

Input: (1) Experimental Data and Dynamical Model

(2) Vector of Normalized Parameter Typical Values (${\Phi }_0\in {\mathbb{R}}^{n_{\mathrm{\Theta }}}$), where the pre-normalization parameter vector is $\mathrm{\Theta }$

(3) Desired Optimal Number of Sensitive Parameters $\left(n_{\theta }^{*}\right)$

Output: (1) A Subset of the Sensitive Parameters ${\theta }^{\text{LASSO}}$

1. NumParams = 0

$2.\hspace{1em}\lambda$ = 0.13

3. while NumParams! = $n_{\theta }^{\mathit{*}}$ do

4.   NumParams = 0

5.   S (selection matrix) = zeros $\left(n_{\theta }^{\mathrm{*}},n_{\mathrm{\Theta }}\right)$

6.   Solve Eq. (C4)

7.   for i = 1: $n_{\mathrm{\Theta }}$ do

8.      if $\left(\mathrm{\Phi }\left(i\right)-{\mathrm{\Phi }}_0\left(i\right)\right)$ > 0.005 then

9.         NumParams = NumParams + 1

10.       end if

11.       if NumParams ≤ $n_{\theta }^{\mathrm{*}}$ then

12.          S(NumParams,i) = 1

13.       end if

14.  end for

15.  if NumParams! = $n_{\theta }^{\mathrm{*}}$ then

16.     $\lambda =\lambda +\frac{NumParams-n_{\theta }^{\mathrm{*}}}{i+n_{\theta }^{\mathrm{*}}}$

17.  end if

18. end while

19. ${\theta }^{\text{LASSO}}=S\mathrm{\Theta }$