randfixedsum - Sum of numbers?

Hi everyone, I've been using randfixedsum to simulate events occurring with a specified total time occurring across the simulated 8-hr day (28800 seconds) with specified bout range (e.g., 1 to 10 seconds) and number of events. It has been working brilliantly, but I can't get it to work for events that occur for a large proportion of the day (i.e., the total time the behaviour occurs for falls short of what it should). The assumptions of randfixedsum are not being violated (as far as I can see) and the code runs just fine, but the output isn't correct. I'm betting the answer is obvious, however I can't see it as I've spent a long time fiddling with different numbers! Any help would be greatly appreciated.

For "behaviour" occurring for 75% of total time (total time = 28800), bout range 1 to 10 seconds, 21600s of behaviour in total, 4320 events:

x = randfixedsum(4320,1,21600,1,10);

 tp = .5; tm = -.5;
 for k = 1:53
   t0 = (tp+tm)/2;
   du = round(x+t0); % Round with offset
   e = sum(du)-21600;
   if e > 0, tp = t0;
   elseif e < 0, tm = t0;
   else break % Break out when sum is correct
   end
 end
y = randfixedsum(4321,1,7200,0,7200);
tp = .5; tm = -.5;
 for k = 1:53
   t0 = (tp+tm)/2;
   on = round(y+t0); % Round with offset
   e = sum(on)-21600;
   if e > 0, tp = t0;
   elseif e < 0, tm = t0;
   else break % Break out when sum is correct
   end
 end
h=[0]
dura=[h;du]'
b = cumsum(on)'
f=cumsum(dura)
onset=b+f
st=onset(1:4320)
duration=dura(2:4321)
m=cumsum(duration)
t = m - duration + 1;  
s  = zeros(1,m(end));
s(t) = 1;
ii = cumsum(s);
out = (1:m(end)) - t(ii) + st(ii);
sort(out)'
out=unique(out)

Thank you, Rebecca

Answer 1

[Roger Stafford]

I notice that in your second for-loop you write

 e = sum(on)-21600;

even though y = randfixedsum(4321,1,7200,0,7200) is set up to obtain a sum of only 7200. You will never correct for this mismatch with your for-loop procedure. You will probably end up with more than two thousand seconds too much for a sum in 'on'. I think you meant to write

 e = sum(on)-7200;

If that isn't the source of the error I suggest you find the sums of both the x and y vectors when they are first created to see if these are as requested, with a reasonable allowance for round-off errors of course. If that checks out all right, then check your 'e' values as you emerge from each for-loop to see if at each of the two points it is exactly zero. I suspect one of these will uncover the error you are experiencing. If not, let us know.

[Rebecca]

That was it! Thanks very much, it was staring me in the face the whole time! :)