multiple time series cross correlation

Hi,
I've different time series, suppose TS1, TS2, TS3, TS4. And I try to
combine TS2, TS3 and TS4 to explain TS1 in the "best possible way".
When I have two time series, for example TS1 and TS2, I can compute
crosscorrelation CCF(TS1, TS2) and take for example the lag L where
CCF is maximum (or other minimum or with maximum module), then
consider for example an increment in TS2 at (T-L) explains an
increment in TS1 in T.

I'm trying to get L1, L2 and L3 lags such that CCF(TS, LAG(TS2, L1) +
LAG(TS2, L2) + LAG(TS3, L3)) is maximum. Is it right? is something
like a combinatorial optimization problem? (TSs are with discrete
domain).
In a principle I think to get Lj such that CCF(TS,LAG(TSj,Lj) is
maximum and then use the Lj's lags. But, may be not the same solution,
or not?

Any help or references (books, text and web pages) about this problem
would be appreciated!

Thanks in advance!
.

Relevant Pages

  • Win 2003 - Terminal Service Manager
    ... Terminal Services Manager (on TS1) to view users etc, I can see TS2 in the ... I should be able to put both server in my Favourites folder so I can ...
    (microsoft.public.windows.terminal_services)
  • Win 2003 TS Manager - You could not be authenticated on this server
    ... We have two terminal servers running 2003, TS1 and TS2, when I go into ... Terminal Services Manager (on TS1) to view users etc, I can see TS2 in the ... Ive rebooted and tried logging on a Admin or as myself (im part of the admin ...
    (microsoft.public.windows.server.general)
  • Win 2003 TS Manager You could not be authenticated on this server
    ... We have two terminal servers running 2003, TS1 and TS2, when I go into ... Terminal Services Manager to view users etc, I can see TS2 in the left plain, ... Ive rebooted and tried logging on a Admin or as myself (im part of the admin ...
    (microsoft.public.win2000.termserv.apps)