Converting a sequence of solutions into a HAR File

The Convert SEQ to HAR file command in the TimeSeries menu performs a similar function to the Load Sequence command which displays a sequence of solutions in ViewSOL. However, Convert SEQ to HAR file instead stores the solutions in a HAR file which you can (optionally) view immediately in ViewHAR. Each variable in the HAR file has a new "period" index which tracks the years.

Possible advantages of this option are:

•Variables with up to 6 dimensions can be seen (limit with ViewSOL is 3).

•The HAR file could be read by another program -- probably written in TABLO language to further process results.

•You may prefer the ViewHAR interface.

Some disadvantages are:

•You need to choose in advance whether you wish to see Base, Policy, or Difference solutions, and whether year-on-year or cumulative results are stored (ViewSOL lets you switch between these options on the fly).

•The HAR file does not store closure information.

A dialog lets you choose:

•whether year-on-year or cumulative results are stored.

•whether Base, Policy, or Difference solutions are stored.

•whether variables are sorted alphabetically (the integer headers reflect the order in which variables appear in the model TAB file).

•whether to ignore higher-dimension variables (this may be needed to save memory).

•whether to combine macro variables into a single header.

•the scheme for labelling periods

•whether to use a Header Mapping file to select certain variables, and to chose headers for them.

•the name and location of the output file (the default output filename indicates whether results are Base, Policy or Difference, and whether Cumulative or Year-on-year).

There is a command-line program, SEQ2HAR, with similar functionality.

Within ViewHAR, the Contents page contains a "<-100" column which counts, for percent change variables, how many values are less than -100. This may help you notice % change variables which changed sign. To discover exactly which elements changed sign, right-click the variable from the Contents screen and select the Properties option -- at the bottom of the Properties window, the [first 50] elements<-100 will be listed. The Properties window also tells if a variable is percent or ordinary change.

Warning: This command only works with SL4 files created with GEMPACK Release 9.0 or later. If the SEQ file lists an SL4 file made with an earlier GEMPACK version, you will probably get an error message like:

not a HAR SL4 file (might be an old-style, non-HAR, SL4 file)

However, ViewSOL can still open such files using the Time Series...Load Sequence command..

Fine print: ViewHAR will apply its "SOL view" viewing options (eg, no row/column totals, a "<-100" column in Contents). This behaviour is triggered if the first line in the File History starts with "#SolView". If the header long names end in "(c)" or "(p)" ViewHAR may infer that a variable is respectively an ordinary or percent change.

URL of this topic: www.copsmodels.com/webhelp/viewsol/hc_seqtohar.htm

Link to full GEMPACK Manual

Link to GEMPACK homepage