Hi Arun,
Your constant and scale terms are very large. How big are your
TENT function betas?
While '-sconstr 2' contains 8.6 (standard GAM), '-sconstr 3'
does not contain 0.547. So the standard shape is not possible
to match. Note that some of the curves can peak over 100, so
scaling on the order of 1000 might be too much.
But nrand=500 might be a problem, too. I would suggest more
like 10000.
Consider running '1deval' to test possible outcomes. For
example assuming no time offset and a scalar of 1, this would
plot the curve of the standard "8.6, 0.547" model on a 0.1
second grid for 12 seconds:
1deval -num 120 -expr '(0.1*t)**8.6*exp(-t/5.47)' > ts.1D
1dplot ts.1D
Note that I multiplied the 0.1 factor out of the exponential
term...
- rick