R help - Jun 2010 - Help with iteration using while loop

Hello everyone,

I am trying to use while loop to iterate a function until convergence. But I
am having problem when I try to use a fixed number of iterations.
Say I want to use maximum iteration of 150. If the value don't converge
within maximum iteration, show warning of no convergence.

Currently I don't have non- convergence problem so I think my code works
fine. But in future I may encounter such problem that are likely to not
converge easily.

Below is my function that is working when I don't provide maximum iteration.

iter<- function (Fpi, Time, tolerance){
S = 22.4
Ts = 0.499
Ti = 0.25
K = 0.044
r<- 1.5
M = Ts- Ti
Ks = 0.044
     Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
       while((Fpt-Fpi) > tolerance) {
         Fpi = Fpt
             Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
     Fp0 = Fpt
    }
 return(Fpt)
}
x<- iter(Fpi = 0.224, Time = 0.2, tolerance = 0.000001)

But I want do something like this ( conceptually)
for( i in 2:itermax) {
  Fpt[i] = K*Time + M*S*log(1+ Fpi/(M*S))
if((Fpt[i]- Fpt[i-1])<= tolerance) break
print(Fpt[i]
}
something like this.

any kind of help is highly appreciated.
thank you





-- 
Acharya, Subodh

	[[alternative HTML version deleted]]

Code inline:

On Fri, Jun 4, 2010 at 1:30 PM, Subodh Acharya <shoebodh at gmail.com>
wrote:
> Hello everyone,
>
> I am trying to use while loop to iterate a function until convergence. But
I
> am having problem when I try to use a fixed number of iterations.
> Say I want to use maximum iteration of 150. If the value don't converge
> within maximum iteration, show warning of no convergence.
>
> Currently I don't have non- convergence problem so I think my code
works
> fine. But in future I may encounter such problem that are likely to not
> converge easily.
>
> Below is my function that is working when I don't provide maximum
iteration.
>
> iter<- function (Fpi, Time, tolerance){
> S = 22.4
> Ts = 0.499
> Ti = 0.25
> K = 0.044
> r<- 1.5
> M = Ts- Ti
> Ks = 0.044
> ? ? Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
> ? ? ? while((Fpt-Fpi) > tolerance) {
> ? ? ? ? Fpi = Fpt
> ? ? ? ? ? ? Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
> ? ? Fp0 = Fpt
> ? ?}
> ?return(Fpt)
> }
> x<- iter(Fpi = 0.224, Time = 0.2, tolerance = 0.000001)
Add a counter variable and a second check in your `while` criteria:

...
...
iter <- 0
while (((Fpt - Fpi) > tolerance) && (iter < itermax)) {
  Fpi = Fpt
  Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
  Fp0 = Fpt
  iter <- iter + 1
}

-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
 | Memorial Sloan-Kettering Cancer Center
 | Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact

Hi Subodh,

Minor note: please keep replies on list so everyone benefits from
answers/questions.

Now:


On Fri, Jun 4, 2010 at 2:49 PM, Subodh Acharya <shoebodh at gmail.com>
wrote:
> Thanks a lot Steve,
> It worked. I appreciate. But I have another question, may be thats trivial.
> Say, it doesn't converge at itermax. I need to display error message
saying
> "values don't converge at at itermax". Is there another
statement that
> accompanies while for this?
Perhaps you can use "warning(...)". You can check whether (or not)
your  Fpt - Fpi is > tolerance after you loop. If it is, you know your
loop terminated because you hit itermax, and not because your algo
converged:
...
...
iter <- 0
while (((Fpt - Fpi) > tolerance) && (iter < itermax)) {
 ?Fpi = Fpt
 ?Fpt = K*Time + M*S*log(1+ Fpi/(M*S))
 ?Fp0 = Fpt
 ?iter <- iter + 1
}

if (Fpt - Fpi > tolerance) {
  warning("Algorithm reached itermax and did not converge")
}

...

-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
 | Memorial Sloan-Kettering Cancer Center
 | Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact