loop ↔ vectorized ↔ direct [Two-Stage / GS Designs]

posted by ElMaestro  – Belgium?, 2017-08-20 17:22  – Posting: # 17726
Views: 23,483

Hi Hötzi,

» I don’t understand why you use vectors at all. What ’bout:
»

df <- Nps*2-2
» if (!Is.St2) {
»   q  <- qt(p=1-alpha1, df=df)
» } else {
»   df <- df-1
»   q  <- qt(p=1-alpha2, df=df)
» }



Simple, but I think I aready told you: qt is a computationally very expensive function. If we need "many" qt values then it makes much more sense in terms of computation to cache the well-defined ones.

Within the 1000000 iterations values from qt are needed many times, but each of them have defined ranges of DF's (up to the max number of subjects -3 [or 2:-)]) and alphas (two levels). Thus we make a cache of qt values prior to entering the iteration loop then we can just take a sample from the cache in stead of re-inventing the wheel for every call. Do the population of the arrays once, and rely on them during the iterations or whatever other functions that need them. The speedup can be massive both in R and in C.

Le tits now.

Best regards,
ElMaestro

Complete thread:

Activity
 Admin contact
20,249 posts in 4,261 threads, 1,397 registered users;
online 17 (1 registered, 16 guests [including 8 identified bots]).
Forum time (Europe/Vienna): 06:40 CET

Many people tend to look at programming styles and languages like religions:
if you belong to one, you cannot belong to others.
But this analogy is another fallacy.    Niklaus Wirth

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5