fosr error

[bernard-liew]

Dear authors,

I am experiencing some errors, and have some questions.

Background: my response is a 16 time point waveform. I have 28 subjects. so Y is a 24 x 16 matrix. My X is a scalar (factor) of 2 groups. My question is when in time do the two group's waveform different.

Error/warning:
fosr_mod = fosr (Y ~ X,
data = dat,
method = "GLS")
Error in S.[[i]] : subscript out of bounds

Alternatively, I have a warning when using OLS
fosr_mod = fosr (Y ~ X,
data = dat,
method = "OLS")
Finding optimal lambda by optimize()...
Warning message:
In optimize(cvfcn, c(0, maxlam), tol = 0.01) :
NA/Inf replaced by maximum positive value

Questions:

1. When using the permuations, how many perms should be done?
2. How does one get the bayesian CI for the coefficients?
3. when deciding lambda's, how does one decide the range of values.

Thank you for the kind response.

Regards,
Bernard

[simonvandekar]

I am getting the same error. The error is caused by a line in refund:::amc indexing the list S.[[i]][[1]]. It looks like, as programmed, amc is not not compatible with multiple penalty matrices.