This function allows the selection of a custom S3 optimizer and additional arguments to be passed to it.
Usage
set_optimizer(
optimizer_name,
objective_name,
start_name,
lower_name,
upper_name,
optim_vals_name,
objective_vals_name,
...
)Arguments
- optimizer_name
a one-length character specifying the name of the optimizer to be used.
- objective_name
a one-length character with the name of the optimizer.
- start_name
a one-length character with the name of the start/initial values input argument.
- lower_name
a one-length character with the name of the lower bounds input argument.
- upper_name
a one-length character with the name of the upper bounds input argument.
- optim_vals_name
a one-length character with the name of the optimum values output argument.
- objective_vals_name
a one-length character with the name of the objective function name output argument.
- ...
Further arguments to be passed to the selected optimizer. Do not include lower bounds, upper bounds and start/initial values. It must be included in the
maxlogLregfunction definition
Value
a list containing the selected optimizer name and the additional, the names of the start values argument, upper and lower bounds, optimum output values, objective function output value and the arguments passed to it.
Examples
library(EstimationTools)
#--------------------------------------------------------------------------------
# Example 1: Estimation in simulated normal distribution (alternative
# implementation using `nlminb` as a custom optimizer)
n <- 1000
x <- runif(n = n, -5, 6)
y <- rnorm(n = n, mean = -2 + 3 * x, sd = exp(1 + 0.3* x))
norm_data <- data.frame(y = y, x = x)
formulas <- list(sd.fo = ~ x, mean.fo = ~ x)
support <- list(interval = c(-Inf, Inf), type = 'continuous')
# Default optimizer
norm_mod1 <- maxlogLreg(formulas, y_dist = y ~ dnorm, support = support,
data = norm_data,
link = list(over = "sd", fun = "log_link"))
summary(norm_mod1)
#> _______________________________________________________________
#> Optimization routine: nlminb
#> Standard Error calculation: Hessian from optim
#> _______________________________________________________________
#> AIC BIC
#> 5148.477 5168.108
#> _______________________________________________________________
#> Fixed effects for mean
#> ---------------------------------------------------------------
#> Estimate Std. Error Z value Pr(>|z|)
#> (Intercept) -1.781010 0.113784 -15.653 < 2.2e-16 ***
#> x 3.054806 0.030641 99.695 < 2.2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> _______________________________________________________________
#> Fixed effects for log(sd)
#> ---------------------------------------------------------------
#> Estimate Std. Error Z value Pr(>|z|)
#> (Intercept) 1.0120201 0.0226350 44.710 < 2.2e-16 ***
#> x 0.2896703 0.0073068 39.644 < 2.2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> _______________________________________________________________
#> Note: p-values valid under asymptotic normality of estimators
#> ---
# Use the default optimizer as a custom one
optimizer <- set_optimizer(
optimizer_name = "nlminb",
objective_name = "objective",
start_name = "start",
lower_name = "lower",
upper_name = "upper",
optim_vals_name = "par",
objective_vals_name = "objective"
)
norm_mod2 <- maxlogLreg(formulas, y_dist = y ~ dnorm, support = support,
data = norm_data,
link = list(over = "sd", fun = "log_link"),
optimizer = optimizer, start= NULL, lower = NULL,
upper = NULL)
summary(norm_mod2)
#> _______________________________________________________________
#> Optimization routine: nlminb Optimization routine: objective Optimization routine: list(lower_name = "lower", upper_name = "upper", start_name = "start") Optimization routine: list(optim_vals = "par", objective_vals = "objective") Optimization routine: NULL
#> Standard Error calculation: Hessian from optim
#> _______________________________________________________________
#> AIC BIC
#> -5132.477 -5112.846
#> _______________________________________________________________
#> Fixed effects for mean
#> ---------------------------------------------------------------
#> Estimate Std. Error Z value Pr(>|z|)
#> (Intercept) -1.781010 0.113784 -15.653 < 2.2e-16 ***
#> x 3.054806 0.030641 99.695 < 2.2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> _______________________________________________________________
#> Fixed effects for log(sd)
#> ---------------------------------------------------------------
#> Estimate Std. Error Z value Pr(>|z|)
#> (Intercept) 1.0120201 0.0226350 44.710 < 2.2e-16 ***
#> x 0.2896703 0.0073068 39.644 < 2.2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> _______________________________________________________________
#> Note: p-values valid under asymptotic normality of estimators
#> ---