id	summary	reporter	owner	description	type	status	priority	milestone	component	version	severity	resolution	keywords	cc
161	Estimating series with different datings produces an error but no warnings	jimarin	Víctor de Buen Remiro	"Hi list,

I tried to make a simple model with Estimate and by error I just put an output 
series in Daily and inputs in Weekly.

I received an error from the Estimate routine:


2004/09/27 16:59:34 : 
Empieza Carga de datos
Cargando datos de la serie output entre y2002m11d19 y y2004m08d30
Norma de la serie output diferenciada : 0.0317868070
Cargando datos de series input entre y2002m11d19 y y2004m08d30
Tiempo Carga de datos 0 segundos
2004/09/27 16:59:34 : 
Empieza Test de datos input para 1 variables ... 
Tiempo Test de datos input para 1 variables ...  0 segundos
 Variables	      : 3
 Datos	          : 643
 Interrupciones  : 0
 Restricciones   : 0
 Grado DIF	      : 8
 Grado AR	      : 0
 Grado MA	      : 8
 Jacobiano	      : Analytical
2004/09/27 16:59:34 : 
Empieza  Método de minimización cuadrática de Marquardt
  Tolerancia Absoluta   = 0.0005
  Tolerancia relativa   = 0.001
  Máximo de iteraciones = 100
  Tamaño de paso para diferenciales numéricas = 0.0001
  Dimensiones = 3x643
Iteración( 0)	norma : ? Tiempo : 0.031 segundos
ERROR: La evaluación de la función objetivo ha fallado.

ERROR: Marquardt falló en la iteración (1)

Tiempo  Método de minimización cuadrática de Marquardt 0 segundos



This is somewhat obscure...the analyst must find out that she or he was specifying 
something wrongly so the objective function could not be evaluated, perhaps due 
to unknown values in a series that come from a strange dating...(at least I put 
everything in Weekly and it worked).

 I think it would have been better to have received a warning about the different 
datings of the time series involved.


thanks to all

cheers

jima"	defect	closed	normal		Math	1.1.1	normal	fixed