Description Usage Arguments Details Value Examples
Generates a tagged list representing starting values for coefficients (fixed and random effects) estimated by TMB
1 | Param_Fn(DataList)
|
DataList |
Tagged list generated by DataFn() |
This function tries to generate logical starting values. For some problems, or for speed, you might start from better-informed values.
This function generates a tagged list, where each slot corresponds to starting values for the following coefficients:
beta1_t |
Scaling encounter probability up or down for each year |
gamma1_j |
Effect of covariate j on encounter probability |
logetaE1 |
Magnitude of marginal variance in spatial variation in encounter probability |
logetaO1 |
Magnitude of marginal variance in spatiotemporal variation in encounter probability |
logkappa1 |
Distance for correlation in encounter probability |
logsigmaV1 |
standard deviation of vessel effects on encounter probability |
logsigmaVT1 |
standard deviation of vessel-year effects on encounter probability |
nu1_v |
Vessel effects for encounter probability |
nu1_vt |
Vessel-year effects for encounter probability |
Omegainput1_s |
Residual spatial variation in encounter probability that is constant for all years |
Epsiloninput1_st |
Residual spatial variation in encounter probability that varies among year |
The preceeding parameters are then repreated for positive catch rates, e.g., beta2_t scales positive catch rates up or down for each year. There are also the following parameters:
ln_H_input |
Parameters representing geometric anisotropy |
logSigmaM |
Parameters representing the distribution of residual errors |
Many coefficients are either random (and hence estimated during innner optimization) or 'turned-off' using the tagged list generated by Map_Fn
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | ## The function is currently defined as
function (DataList)
{
Return = list(ln_H_input = c(0, 0), beta1_t = qlogis(tapply(ifelse(DataList$b_i >
0, 1, 0), INDEX = DataList$t_i, FUN = mean)), gamma1_j = rep(0,
DataList$n_j), lambda1_k = rep(0, DataList$n_k), logetaE1 = 0,
logetaO1 = 0, logkappa1 = 0, logsigmaV1 = log(1), logsigmaVT1 = log(1),
nu1_v = rep(0, DataList$n_v), nu1_vt = matrix(0, nrow = DataList$n_v,
ncol = DataList$n_t), Omegainput1_s = rep(0, DataList$n_s),
Epsiloninput1_st = matrix(0, nrow = DataList$n_s, ncol = DataList$n_t),
beta2_t = log(tapply(ifelse(DataList$b_i > 0, DataList$b_i/DataList$a_i,
NA), INDEX = DataList$t_i, FUN = mean, na.rm = TRUE)),
gamma2_j = rep(0, DataList$n_j), lambda2_k = rep(0, DataList$n_k),
logetaE2 = 0, logetaO2 = 0, logkappa2 = 0, logsigmaV2 = log(1),
logsigmaVT2 = log(1), logSigmaM = c(log(5), qlogis(0.8),
log(2), log(5)), nu2_v = rep(0, DataList$n_v), nu2_vt = matrix(0,
nrow = DataList$n_v, ncol = DataList$n_t), Omegainput2_s = rep(0,
DataList$n_s), Epsiloninput2_st = matrix(0, nrow = DataList$n_s,
ncol = DataList$n_t))
return(Return)
}
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.