Re: Non-linear curve fitting: y = Ae^(-t/tau) + C
- From: robert bristow-johnson <rbj@xxxxxxxxxxxxxxxxxxxx>
- Date: Tue, 20 Jun 2006 03:23:22 GMT
ya know, Markus, i spelled out a closed form expression for a least square
estimate for "C" in
y = Ae^(-t/tau) + C
with that C subtracted out from y, the remainder, Ae^(-t/tau), can be
logged, resulting in
log(y - C) = log(A) + (-1/tau)*t
which is a straight line and in which the standard least squares formulae
results in an estimate for log(A) and -1/tau from which A and tau can be
trivially computed.
you had a solution staring at you the whole time.
--
r b-j rbj@xxxxxxxxxxxxxxxxxxxx
"Imagination is more important than knowledge."
.
- References:
- Re: Non-linear curve fitting: y = Ae^(-t/tau) + C
- From: Markus Svilans
- Re: Non-linear curve fitting: y = Ae^(-t/tau) + C
- Prev by Date: Re: How does an inverter affect phase?
- Next by Date: Re: Energy Measurement
- Previous by thread: Re: Non-linear curve fitting: y = Ae^(-t/tau) + C
- Next by thread: Re: Non-linear curve fitting: y = Ae^(-t/tau) + C
- Index(es):
Relevant Pages
|
