It's likely that the loss function has a log() or 1/x and the finite
difference
approximations to gradients have added / subtracted a small number and caused a
singularity. Unfortunately, you'll need to dig into the fGarch code or write
your own
(ouch!). Or perhaps the fGarch package maintainer will be willing to look at the
package
if you provide him/her your problem.
This sort of thing is common enough that some of us have been looking into
better tools
for generating gradients. However, don't hold your breath. It's not easy
work.
JN
On 03/24/2011 07:00 AM, r-help-request at r-project.org
wrote:> Message: 85
> Date: Thu, 24 Mar 2011 17:03:41 +0800
> From: Luis Felipe Parra <felipe.parra at quantil.com.co>
> To: r-help <r-help at r-project.org>
> Subject: [R] Problems with predict in fGarch
> Message-ID:
> <AANLkTi=62Zki1tt71ArjeMRV07--Mg_6Ejh+70iUrFfN at mail.gmail.com>
> Content-Type: text/plain
>
> Hello. I am using fGarch to estimate the following model:
>
>
> Call:
> garchFit(formula = fmla, data = X[, i], trace = F)
> Mean and Variance Equation:
> data ~ arma(1, 1) + garch(1, 1)
>
> Conditional Distribution:
> norm
> Coefficient(s):
> mu ar1 ma1 omega alpha1 beta1
> -0.94934 1.00000 -0.23211 54.06402 0.45709 0.61738
> Std. Errors:
> based on Hessian
> Error Analysis:
> Estimate Std. Error t value Pr(>|t|)
> mu -0.949336 11.600072 -0.082 0.93477
> ar1 1.000000 0.005947 168.139 < 2e-16 ***
> ma1 -0.232111 0.068638 -3.382 0.00072 ***
> omega 54.064022 16.578735 3.261 0.00111 **
> alpha1 0.457087 0.093125 4.908 9.19e-07 ***
> beta1 0.617378 0.044561 13.855 < 2e-16 ***
> ---
> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05
'.' 0.1 ' ' 1
>
>
> when I use predict I am getting the following error:
>
> Error en optim(init[mask], armafn, method = optim.method, hessian = TRUE,
:
> non-finite finite-difference value [1]
>
>
> does anybody know what might be going on?
>
> Thank you
>
> Felipe Parra
>
> [[alternative HTML version deleted]]
>
>