I am trying to run a nonlinear least squares regression using minpack.lm:nls.lm. The equation to be minimized is e = x - p1*t1 -p2*t2 - svdep , where t1 and t2 are changed so that the sum of squares of e is minimized.
gnfun <- function(x, p1, p2, t1, t2, svdep){
x - p1*t1 - p2*t2 - svdep
}
start <- list(t1=-0.1095389, t2=0.02329868)
gn2 <- nls.lm(par=start, fn=gnfun, x=X1, p1=p1.1, p2=p1.2,
svdep=Epsvd1)
Error in p1 * t1 : non-numeric argument to binary operator
X1
X1 X2 X3 X4 X5 X6
7.725156e-08 7.342344e-08 7.334688e-08 7.572031e-08 7.350000e-08 7.441875e-08
X7 X8 X9 X10
7.388281e-08 7.105000e-08 7.357656e-08 7.028438e-08
Epsvd1
[1] 3.210028e-05 3.238753e-05 3.160692e-05 3.270296e-05 3.625271e-05 3.167958e-05
[7] 3.667674e-05 3.317648e-05 3.574715e-05 3.335333e-05
p1.1
[1] 0.001156993 0.001159083 0.001158931 0.001162099 0.001160497 0.001158225
[7] 0.001157901 0.001157770 0.001161935 0.001163280
p1.2
[1] 0.005751636 0.005710543 0.005711749 0.005697742 0.005720252 0.005765593
[7] 0.005759443 0.005778480 0.005759381 0.005712900
class(p1.1)
[1] "numeric"
class(p1.2)
[1] "numeric"
> class(X1)
[1] "numeric"
> class(Epsvd1)
[1] "numeric"
I cannot figure out why I got the error 'non-numeric argument to binary operator' for p1*t1, even though p1 and t1 are both numeric .
I would appreciate any advice as to why I am getting this error message, and what I need to do in order to get this to run properly.
You need to remember that you are passing a list and need to pull parameters from that list using [[:
gnfun <- function(x, p1, p2, par=par, svdep){
x - p1*par[[1]] - p2*par[[2]] - svdep
}
start <- list(t1=-0.1095389, t2=0.02329868)
gn2 <- nls.lm(par=start, fn=gnfun, x=X1, p1=p1.1, p2=p1.2,
svdep=Epsvd1)
summary(gn2)
Parameters:
Estimate Std. Error t value Pr(>|t|)
t1 0.003322 0.092779 0.036 0.972
t2 -0.006510 0.018755 -0.347 0.737
Residual standard error: 2.019e-06 on 8 degrees of freedom
Number of iterations to termination: 2
Reason for termination: Relative error in the sum of squares is at most `ftol'.