连老师您好，下面是我做的一个二重循环，但是我run的结果出来当i =1 和i =2的时候，A11的结果是一样的，也就是说，里面的循环出了问题，导致了A11的前2行的数值一模一样，琢磨了半天，也没弄清楚里面那个循环到底出了什么问题，导致这样的结果，我想可能是跟`i', `j'等标记的顺序有关？非常感谢！

* generating variables
gen g = gdp[_n] - gdp[_n-1]
gen g_percent = [gdp[_n] - gdp[_n-1]]/gdp[_n-1]*100
gen g_year = [(1 + g_percent/100)^4 - 1]*100
gen k = _n
* NOTE: i is the i-th sampling; j is the j time lagging periods; k is the k-th principal components.
*-----------------------------------------*

* k = 1, number of Principal Components is 1.

forvalues j = 1/4{
    * Ajk: j denotes the time lagging, k denotes k-th principal components
    mat A`j'1 = J(42,1,0)  
    forvalues i = 1/2{
   
    * selecting our sample
    preserve
keep if k < 42 - `j'

* PCA regression  
    * claiming that this is time series data
    encode date, generate(time)
    tsset time, quarterly

    * gi_year_forwardj: i denotes the i-th sampling; j denotes the j time lagging
    gen g`i'_year_forward`j' = F`j'.g_year   
    keep if g`i'_year_forward`j'!=.
   
    * principal component analysis
    display "applying PCA when i = `i', j = `j'"
    pca cy*, comp(1)
    * Draw the figure
    * screeplot

    * Prediction
    * cyf (ijk)
    * i denotes the i-th sampling; j: number of time lagging periods; k: k-th principal components
display "prediction of the scores when i = `i', j = `j'"
    predict cyf`i'`j'1, score

    * Regression
    display "regression of principal components when i = `i', j = `j'"
    regress g`i'_year_forward`j' cyf`i'`j'1

    * coefijk: i denotes the i-th sampling;
    * j deontes number of time lagging periods. k denotes k-th principal components   
    scalar coef`i'`j'1 = _b[cyf`i'`j'1]

    * Bij: i denotes the i-th sampling, j denotes the j time lagging
    * k denotes k-th principal components
    mat B`i'`j'1 = (coef`i'`j'1)

    * obsi: i denotes the i-th sampling
    * j denotes the j time lagging; k denotes k-th principal components
    scalar obs`i'`j'1 = _N
    keep if k == _N

    mkmat cyf`i'`j'1, mat(cyf`i'`j'1)
    * Gijk: i denotes the i-th sampling, j denotes the j time lagging
   
     * k denotes k-th principal components
    display "G is the i-th rows of the matrix Ajk when j = `j', k = `k'"
    mat G`i'`j'1 = cyf`i'`j'1*B`i'`j'1 + _b[_cons]
    * Ajk: j denotes the time lagging, k denotes k-th principal components
    display "A is the matrix when j = `j'"
    mat A`j'1[`i',1] = G`i'`j'1

restore

}  
  }