Clarfication on coverage computation

The navigation package references https://ieeexplore.ieee.org/document/10104150. Therein, the confidence interval for computing the coverage of NEES is defined as:

The degree of freedom of the chi-squared distribution where N is the number of simulations (samples) if I am not mistaken. However, the implementation uses the dimension of the error vector as the degree of freedom:

navigation/R/stats.R

Line 62 in a0b127b

bounds <- qchisq(c(0, alpha), df = length(idx))

Thus, I am wondering which one is correct. Can somebody clarify my confusion?

Considering it again, then the implementation is correct, it should be the dimension of the error vector. The single-sided bound is computed for a single experiment, i.e. N=1, and, thus, the remaining degree of freedom is given by the dimension of the error vector.