fstcomp(1) |
RPN UTILITY |
fstcomp(1) |
fstcomp version 7.2 (February 2005)
fstcomp [-a -b -l -as -bs - af -bf -li -nd -ne -d -n -va -vb -nt -n1 -n2 -n3 -nn -x]
Utility used to compare the contents of two RPN standard files (record by record). The comparison is done by looking at the records in file A sequentially and for each record, look in file B for the equivalent record before making a comparison of the values stored within this field. The output of the comparison report is through standard output but maybe redirected given a filename in the "-l" key.
| Definitions of keys in the command line | |
| key | description |
| -a filea |
filename of one of the two input RPN standard files for comparison. Default filename is "a" |
| -b fileb |
filename of the other input RPN standard file for comparison. Default filename is "b" |
| -l listing | file given for the stdout listing |
| -as | obsolete; do not use |
| -bs | obsolete; do not use |
| -af | obsolete; do not use |
| -bf | obsolete; do not use |
| -li | the exponent of the maximum relative error tolerance. Example: "-li =-3" means that FSTCOMP will tolerate differences not greater than 0.001 The default value is E-07 (.0000001) |
| -packerr | the multiplier (usually 1) on the tolerance criterion (which is the compaction error) for the comparison. Example: Tolerance criterion for R16 (16bit compaction) = (Max - Min) / 2**16 If the difference between two values are smaller than (tolerance)*(packerr), it will be tagged as no difference. |
| -nd | To ignore the "date validity" parameter when searching for the equivalent record in file "B" to the record in file "A". (datev=dateo+npas*deet) |
| -ne | To ignore the "etiket" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -d | To print more information than just the standard comparison |
| -n | Outputs no boxes around the message EXDB, EXFIN, at the beginning and at the end of the execution respectively. |
| -va filea |
filename of one of the two input RPN random standard files for comparison and will print a table of contents of this file in a "voir" style. If the key is mentioned, the default filename would be "va" |
| -vb fileb |
filename of the other input RPN random standard files for comparison and will print a table of contents of this file in a "voir" style. If the key is mentioned, the default filename would be "vb" |
| -nt | To ignore the "typvar" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -n1 | To ignore the "ip1" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -n2 | To ignore the "ip2" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -n3 | To ignore the "ip3" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -nn | To ignore the "nomvar" parameter when searching for the equivalent record in file "B" to the record in file "A". |
| -x | To specify that the "datyp" of the records in the files are of type "X" (ie: X32) - raw binary mode. Normally the "datyp" is in real "R" or IEEE "E" style representation. |
| Description of Report Heading | ||
| Title | Description | |
| CLEA | key of record found in file A | |
| CLEB | key of record found in file B | |
| NOM | nomvar - variable name of field | |
| ETIKET | etiket - label of run/data | |
| IP1 | ip1 - code of model level (see convip) | |
| IP2 | ip2 - hour of forecast | |
| IP3 | ip3 | |
| E-REL-MAX | maximum relative error => |1.0-A/B| where values A and B are compared. B must be non-zero so A and B could be switched for the purpose of calculation | |
| E-REL-MOY | average relative error. See E-REL-MAX | |
| VAR-A | variance of field in file A | |
| C-COR | correlation (1.000 if fields are equal) | |
| MOY-A | mean of field in file A | |
| BIAIS | bias (0.000 if fields are equal) | |
| E-MAX | maximum absolute error (0.000 if fields are equal) | |
| E-MOY | average absolute error (0.000 if fields are equal) | |
May 3rd, 2007, by Vivian Lee