Dynamically generate new values of variables using dyngen

[Parth Aks]

Hi,
I am working with cohort life tables and generating the values of lx (number of people alive) at age x, dx (number of deaths) at age x based on qx (probability of death before next birthday). I am using dyngen to dynamically update values and would like help/advice to speed it up.

Three variables: q, l, and d for 90 observations (rows).
Initial values are as follows:
q₀ to q₉₀ are external inputs.
l₀ or the starting value of lx is external input.
d₀ = l₀*q₀

Subsequent rows l₁ and d₁ are calculated as:
l₁=l₀-d₀
d₁=l₁*q₁

The process continues row by row until l₉₀.
I use the following Stata code (dyngen) in Stata MP 17:

dyngen {    
    update lx = lx[_n-1] - dx[_n-1] , missval(`intial_lx')
    update dx = lx[_n]     * qx[_n] , missval(`intial_dx')
}

My question is, is there a way to speed up this line by line calculation? I use this in a simulation and the dyngen is the only bottle-neck.

Thank you
Parth

[Parth Aks]

Solved my own problem using a solution posted to recursive calculations by Clyde Schechter and Carlo Lazzaro https://www.statalist.org/forums/for...ver-a-variable

cap drop B
g B=sum(ln(1-qx))
replace B = exp(B)
replace lx = round(B[_n-1]*lx[1],1) in 2/L