General solution using SPA (module_A_chisq.m)

Let element $x_i$ have distribution $p(x_i)=\chi^2(x_i; k_i,\rho^2_i)$, that is the chi-square distribution with $k_i$ degrees of freedom and scale $\rho^2_i$. A chi-square RV is created whenever the squares of independent Gaussian RVs of equal variance are added up. The distribution of the chi-square RV $x_i$ depends on two parameters, the degrees of freedom $k_i$ , which equals the number of Gaussians that were added, and the variance $\rho^2_i$, which is the assumed variance of the Gaussians. The mean of $x_i$ equals $k_i\rho^2_i$. A description of chi-square random variables is provided in Section 17.1.2.