Diviya Smith wrote
Hi there,
I would like to solve the following equation in R to estimate 'a'. I have
the amp, d, x and y.
amp*y^2 = 2*a*(1-a)*(-a*d+(1-a)*x)^2
test data:
amp = 0.2370 y=
0.0233 d=
0.002 x=
0.091
Can anyone suggest how I can set this up?
This should help you to get started.
library(nleqslv)
f <- function(a, amp=.2370,y=0.0233, d=0.002, x=0.091) {
amp*y^2 - ( 2*a*(1-a)*(-a*d+(1-a)*x)^2 )
}
# draw the function
curve(f, from=-0.2, to=1.5, col="blue")
abline(h=0, col="red")
# use nleqslv
# use two different starting values to see if we can solve for the two roots
# you can leave out the control=... argument
astart <- 1
nleqslv(astart,f, control=list(trace=1))
astart <- 0
nleqslv(astart,f, control=list(trace=1))
# use uniroot
# limit the lower/upper interval to the two roots
# ranges chosen from the plot so that function values at the endpoints have different signs
uniroot(f,c(-0.5,.1))
uniroot(f,c(0.5,1.5))
Berend