advice on efficiency in doing loop

[charlie wong]

hello

I am trying to do the following loop which would take quite some time. I would be grateful if anyone can comment if there is any part of the code that can be modified to speed up to computation. Thanks!

unab x : x*
loc xcount = wordcount("`x'")
forvalues i=1/`xcount' {
loc xLx="`xLx' x`i' L1_x`i' "
}
putmata X=(`xLx') //X store a large number of variables from Stata
mata xcount=cols(X) 
mata store=J(xcount*xcount*xcount,7,.) //to store computation results from the mata loop below

mata
    loop=1
    for (a=1; a<=xcount; a++) { 
        for (b=1; b<=xcount ; b++) { 
            for (c=1; c<=xcount ; c++) { 
                
                if (a!=b & a!=c & b!=c & a>3 & b>3 & c>3) {
                
                    x1mean        = mean(select(X[,1], X[,a]:>X[,b] :& X[,a]:<X[,c] :& !(X[,a]:>X[,b] :& X[,a]:<X[,c]) :& X[,1]:<0.4 )) // compute conditional mean on X[,1]
                    
                    x2mean        = mean(select(X[,2], X[,a]:>X[,b] :& X[,a]:<X[,c] :& !(X[,a]:>X[,b] :& X[,a]:<X[,c]) :& X[,1]:<0.4 )) // compute conditional mean on X[,2]
                    
                    x3mean        = mean(select(X[,3], X[,a]:>X[,b] :& X[,a]:<X[,c] :& !(X[,a]:>X[,b] :& X[,a]:<X[,c]) :& X[,1]:<0.4 )) // compute conditional mean on X[,3]

                    freq        = colnonmissing(select(X[,1], X[,a]:>X[,b] :& X[,a]:<X[,c] :& !(X[,a]:>X[,b] :& X[,a]:<X[,c]) :& X[,1]:<0.4 )) // compute obs no. on selected X[,1] 
                    
                    store[loop,]=(a,b,c,freq,x1mean,x2mean,x3mean)
                    loop=loop+1
                }
                
                else 1
            }
        }
    }

    xa=store[,1]
    xb=store[,2]
    xc=store[,3]
    freq=store[,4]
    x1mean=store[,5]
    x2mean=store[,6]
    x3mean=store[,7]

end