DESL


DESL Home | Website Outline | Commands | Examples | Download | Contacts

( BE) Details concerning the DESL learn command.


(See example(s) of the use of the learn command.)
[Jump to the list of available options.]
General description of the learn command :

Determines the coefficients and possibly the most important terms contained in up to 100 polynomials for a set of data contained on up to 100 SIF files. The resulting polynomials can be written to an equation file for future use. Each polynomial is a least-squares solution to the complete data set. The maximum number of data points considered per polynomial is a function of the number of polynomial terms specified. A measure of how well the polynomial actually fit the original data can be reported. The degree to which the determined polynomials will predict values of dependent variables is a function of how "complete" the sets of independent variables used in the polynomials are. Two measures of the "confidence" levels of the matches can be reported.


The list of currently available options for the learn command is the following.

Any blinking names enclosed in brackets are required or are in a set of options, one of which must be specified. Any non-blinking names enclosed in brackets are optional or are in a set of options, only one of which can be specified.

In cases where the option name is one of a set each different set number is indicated by a red numeric set number superscript trailing the respectiveright bracket :

[altvars] [backward] [confidence] [cset] [elsek] [endifk] [eqfile] [eqtol] [files] [ifk] [lines] [neglect] [newfiles] [newfilex] [newterms] [nobias] [noop] [output] [pathfile] [paths] [polyterms] [prefix] [report] [setvoid] [tags] [vfile] [vfilen]

[Top][Bottom][Option list]

( BEaltvars) (Command :learn) Details for the altvars option.

[General syntax rules for this keyword.]

The presence of this option indicates that IF there is an active composite alternate SIF variable name file (as established via the "variables" DESL command) then this composite file will be used to attempt to assign an alternate SIF variable name if a SIF variable would otherwise not be found on the file(s) being processed.

[Top][Bottom][Option list]

( BEbackward) (Command :learn) Details for the backward option.

[General syntax rules for this keyword.]

Indicates that searching for SIF variables will be done from back to front in the SIF records. The default condition is to search from front to back.

[Top][Bottom][Option list]

( BEconfidence) (Command :learn) Details for the confidence option.

[General syntax rules for this keyword.]

If specified this option must be followed by 3-item groups composed of (1) a desired confidence percent, "xx", (2) the name of the "dci" computed variable, and (3) the name of the "dpi" computed variable. The three-item groups correspond to the polynomials specified or implied in a 1:1 manner. "dci" and "dpi" are, respectively, the calculated confidence interval and prediction interval, both of which depend on the standard error and are statistical measures related to the level of confidence one may place on the computed relationship (least squares fit) between the independent and dependent variables.

The confidence interval is defined as a band about an estimated mean value (the least squares fit) within which there is an "xx" percent confidence that the true mean value will be found. The "xx" percent confidence is referred to as the confidence level. The resulting "dci" parameter, which will be added to each SIF file produced if "newfiles" or "newfilex" is also specified, is an increment, to be interpreted as a positive and negative increment about the computed value of the dependent variable.

The prediction interval is defined as a band about an estimated mean value (least squares fit) within which there is an "xx" percent chance that the next measured data point be found. The resulting "dpi" parameter, which will be added to each SIF file produced if "newfiles" or "newfilex" is also specified, is an increment, to be interpreted as a positive and negative increment about the computed value of the dependent variable.

The confidence percent values specified must lie between 68.268 (1 sigma) and 99.994 (4 sigma). Specification of a confidence percent value outside of this range will force the use of 95 percent.

Example : confidence 95 dci1 dpi1

[Top][Bottom][Option list]

( BEcset) (Command :learn) Details for the cset option.

[General syntax rules for this keyword.]

This option is used to specify "file-specific" conditions which, IN ADDITION to any other GLOBAL conditions which ALSO may have been specified, are to be applied to each SIF record processed from the respective SIF file. Just like global conditions, file-specific conditions act to limit the applicability of the DESL function being executed by requiring that ALL applicable "and" conditions be TRUE simultaneously or that at least ONE "or" condition be TRUE.

The item immediately following "cset" is the file sequence number in the "files" list for which the following up-to-20 conditions will be applicable.

Each file-specific 3-item condition, like its global 3-item condition counterpart, is made up of (1) a SIF variable name, (2) a 2-character logical conjunction, and (3) a second SIF variable name or a constant.

Example : cset 3 alpha gt phi mach lt 1.2

Here, the user has indicated that, along with whatever other GLOBAL conditions may have been specified, the pertinent DESL function, for the THIRD file specified or implied by the "files" list, will have its application ALSO limited to records which satisfy BOTH of the two conditions :

"ALPHA gt PHI" "MACH lt 1.2".

All file-specific conditions are implicitly joined by an "and" conjunction; ie, by default, all file-specific conditions would have to be true SIMULTANEOUSLY for the applicable record to be retained.

If one or more file-specific conditions following the file sequence number is to be connected to other file-specific conditions for the same file number by an "or" conjunction, then the 2-character item "or" must immediately follow the file number.

Example : cset 2 or alpha lt 0 alpha gt 4

Here, the user has indicated that, along with whatever other GLOBAL conditions may have been specified, the pertinent DESL function, for the SECOND file specified or implied by the "files" list, will have its application ALSO limited to records in which the value of the SIF variable "alpha" satisfies at least ONE of the conditions :

"ALPHA lt 0" "ALPHA gt 4".

Additionally, if the file number following the "cset" option is negative, the absolute value of this negative file number is used for the actual file number and the associated "and" or "or" condition(s) are NOT used in the traditional sense of filtering the records read but are ONLY used to indicate when READING of the associated file is to STOP. This ability is useful in preventing the reading of a long file when it is known that the data required is relatively near the beginning of the file.

Example : cset -1 or run eq 3 alpha gt 24

Here, the user has indicated that when a record is encountered for which either or both of the two conditions

"RUN = 3" "ALPHA gt 24"

is/are TRUE, reading of the current SIF file should end.

[Top][Bottom][Option list]

( BEeqfile) (Command :learn) Details for the eqfile option.

[General syntax rules for this keyword.]

Indicates the name of the file to contain the Fortran-like equations and logic to be created to define all of the polynomial equations to be determined. The polynomial equations written to the equation file will be in terms of the independent variables which were specified in the "polyterms" argument list(s). The polynomial equations may be preceded by the definitions of the new terms as specified by the "newterms" argument list. Therefore, when this equation file is used in the future by the DESL "equation" function to operate on other SIF files, the named "polyterms" variables which are not defined by "newterms" and the "newterms" variables which are not defined by other "newterms" should exist on each of those SIF files. Only one "eqfile" file can be specified. If the named equation file already exists, it is overwritten.

Example : eqfile eq1

[Top][Bottom][Option list]

( BEeqtol) (Command :learn) Details for the eqtol option.

[General syntax rules for this keyword.]

Indicates the tolerance to be applied to any global or file-specific "eq" condition(s) which may have been specified. If the tolerance value specified is positive the value is in the units of the variable being used in the respective condition(s). If the tolerance value is negative the absolute value of the value is a PERCENT difference. The default value of tolerance is 0.

Example : eqtol .5

[Top][Bottom][Option list]

( BEfiles) (Command :learn) Details for the files option.

[General syntax rules for this keyword.]

Indicates the names of the files from which to extract data corresponding to the value(s) of the specified polynomial independent variables and, optionally, the variables used in the right-hand-side of the "newterms" argument list. This list of files can be implicitly extended via use of the <list argument which can appear among the specified file names. If one or more of the <list-type arguments are specified, where "list" is a currently defined file name list, the file names contained in the indicated file name list(s) will be included, at their respective point(s) of encounter, in the current file list. The referenced list(s) must have been produced by the "newfiles" and/or "newfilex" argument lists associated with one or more previously executed DESL commands. Each "list" name can be up to 15 characters long. File names can also include the {...} construct. This type specification within a file name implies that the list name given between the curly brackets ( {} ), is to be used to have each of its elements substituted in turn for the { ... }, inclusively, such that an implicit lengthening of the file list is accomplished, up to the maximum number of files allowed. Any time such a {...} usage occurs, all lists mentioned must exist; i.e., must have been previously defined in the current DESL session. Only one {...} construct can be included in a file name. Including more than one such construct will result in unpredictable results.

An example of this type of file name specification is the following :

t43r{runs}.sif

Here, the list "runs" will have its elements used, in turn, in place of the {runs} substring.

.

(Begin modifications on 052102)

Each file name, whether it includes the {...} or not, can be a maximum of 256 characters long. If a file name is more than 16 characters long it must be enclosed in question marks (?) .

(End modifications on 052102)

Up to 100 file names can be specified.

Example : files run10 run11 run12

[Top][Bottom][Option list]

( BEneglect) (Command :learn) Details for the neglect option.

[General syntax rules for this keyword.]

If specified, indicates that the best combination of specified polynomial terms will be retained. Here, the "best" combination of terms means the combination of terms which yields a value of "r-squared" percent which is closest to 100 percent; ie, closest to EXACTLY predicting the actual data. If "neglect" is NOT specified, all of the specified polynomial terms, via "polyterms", will be used in the determination of the coefficients for each of the desired polynomials.

Example : neglect

[Top][Bottom][Option list]

( BEnewfiles) (Command :learn) Details for the newfiles option.

[General syntax rules for this keyword.]

Indicates that one or more new files are to be produced and indicates the desired names for these new files. Each new file produced will contain the respective original SIF file's data (after condition(s) applied) plus the calculated polynomial variables and, optionally, the two "confidence" variables. Normally no newfiles are produced from "learn".

There is an understood 1:1 correspondence between files in the "files" list and files in the "newfiles" list. A new file name prefixed with the 4-character string "null" can be used as a place holder in the "newfiles" file names list. This type of name is converted to a blank string before it is used; ie, using a "null" prefixed name is equivalent to having not specified that name at all. By default, all files explicitly named via this "newfiles" list will be flagged as a file "to be RETAINED at cleanup. (See the "cleanup" command.) The default file name generated for a new file not explicitly named via "newfiles" or "newfilex" is the 3-character string "gen" followed by a unique numeric suffix. Up to 100 new files can be specified.

The list of the names of the new files actually produced can be saved via the use of the ">list"- or ">>list"- type argument, which can appear among the specified new file names. If a ">list"- or ">>list"-type argument is specified, where "list" is the specified file name list, the names of the new files which DESL actually produces during the execution of the current command will be saved in the specified file name list.

This file name list can later be referenced by the "files"/"xfiles" options to imply the file names contained therein. Up to 100 new file names can be saved in any one list. If the ">list" argument is specified the named list will be created if it does not already exist, or will overwrite a like-named list if one exists. If the ">>list" argument is specified the named list will be appended to if it exists or will be created if it does not exist. All file name lists are in existence for the current DESL session only. Each "list" name can be up to 15 characters long.

In all but a few noted cases, it is acceptable to use the ">list" or ">>list"-type argument without specifying a full complement of actual new file names. Any new file produced whose NAME is GENERATED by DESL is ALWAYS marked as a file to "PURGE at cleanup".

A suffix which may be in effect as established via the filesuffix DESL command will be added to any expressed or implied new file name. New file names can also include the {...} construct. This type specification within a file name implies that the list name given between the curly brackets ( {} ), is to be used to have each of its elements substituted in turn for the { ... }, inclusively, such that an implicit lengthening of the new file list is accomplished, up to the maximum number of new files allowed. Any time such a {...} usage occurs, all lists mentioned must exist; i.e., must have been previously defined in the current DESL session. Only one {...} construct can be included in a new file name. Including more than one such construct will result in unpredictable results.

An example of this type of new file name specification is the following :

t43r{runs}.sif

Here, the list "runs" will have its elements used, in turn, in place of the {runs} substring.

.

(Begin modifications on 052102)

Each new file name, whether it includes the {...} or not can be a maximum of 256 characters long. If a newfile name is more than 16 characters long it must be enclosed in question marks (?).

(End modifications on 052102)

Example : newfiles nf1 nf2 >nflist21

[Top][Bottom][Option list]

( BEnewfilex) (Command :learn) Details for the newfilex option.

[General syntax rules for this keyword.]

This option has exactly the same function as the option "newfiles" except that all new files created via a "newfilex" name are flagged as files to "PURGE at cleanup". (See "newfiles" option. See "cleanup" command.) Up to 100 new files can be specified.

Example : newfilex newr1 thru newr4 newr201

[Top][Bottom][Option list]

( BEnewterms) (Command :learn) Details for the newterms option.

[General syntax rules for this keyword.]

This option and argument list allows the definition of temporary terms which can be functions of SIF variables and/or other previously-defin new terms. These temporary terms can be referenced by one or more of the polynomials whose coefficients are to be determined. This option must be followed by one or more "sets" of renaming phrases. Each renaming phrase is composed of a left-hand side, and equal sign, and a right-hand side.

Example : "alphasqd = alpha^2". The left-hand side is an up-to 8-character temporary variable name to be defined by the phrase. The right-hand side is an up to 32-character string which MUST contain either (a) a single-variable operator (such as "sin", "cos") and its argument, or (b) an argument, an arithmetic operator, and another argument. An example of the type of string described in (a) above is : "sin(angle)". An example of the type of string described in (b) above is : "alpha*beta".

The set of single-variable operators is the following :



abs ..........absolute value
acos .........arc cosine, result in radians
aint .........integral part of number
anint ........rounded value of number
asin .........arc sine, result in radians
atan .........arc tangent, result in radians
cos ..........cosine, argument in radians
cosh .........hyperbolic cosine, argument in radians
exp ..........exponential (e to the ...)
log ..........natural log
log10 ........log to base 10
sin ..........sine, argument in radians
sinh .........hyperbolic sine, argument in radians
sqrt .........square root
tan ..........tangent, argument in radians
tanh .........hyperbolic tangent, argument in radians
sind .........sin, argument in degrees
cosd .........cosine, argument in degrees
tand .........tangent, argument in degrees
asind ........arc sine, result in degrees
acosd ........arc cosine, result in degrees
atand ........arc tangent, result in degrees

An example of the type of string described in (b) above is "cd*cl". The set of operator symbols which can be used in type (b) strings is the following :



^ ...........exponentiation; 2^3 = 8
* ...........multiplication
/ ...........division
+ ...........addition
- ...........subtraction

New terms thus defined can be used in any or all of the polynomials to be determined. Up to 200 new terms can be defined. The string on the right-hand side should contain no blanks.

If "newfiles" or "newfilex" is also specified, the left-hand side (LHS) of each renaming phrase (see above) is added to the SIF file being created unless the string "local" follows immediately after "newterms". All of the new terms which are contained in a "newterms" argument list, in which "local" follows the option, are, themselves, termed "local" and will therefore NOT be written to the SIF file(s) produced. Following another occurrence of "newterms", if "local" is NOT present, the new terms so defined are NOT local; ie, they WILL be written to the SIF file(s) produced.

Example : newterms local beta2 = beta*beta

[Top][Bottom][Option list]

( BEnobias) (Command :learn) Details for the nobias option.

[General syntax rules for this keyword.]

Normally, to try to account for circumstances in which the magnitude of the independent variable values is large, the independent variable values are "biased" by their mean value before the arithmetic is done to compute the polynomial's coefficients. The accuracy of the resulting polynomial coefficients is improved by biasing. If it is desired that no biasing be done in the process of determining the polynomial coefficients, the "nobias" option can be specified.

[Top][Bottom][Option list]

( BEnoop) (Command :learn) Details for the noop option.

[General syntax rules for this keyword.]

A dummy option which allows one or more GLOBAL conditions to follow. This option has NO OTHER FUNCTION.

[Top][Bottom][Option list]

( BEoutput) (Command :learn) Details for the output option.

[General syntax rules for this keyword.]

If specified the following string is the name of a file which will be written and will include echoes of any "report" output which was issued as the "learn" function executed. This output file will be created whether it exists or not.

Example : output outeq

[Top][Bottom][Option list]

( BEpathfile) (Command :learn) Details for the pathfile option.

[General syntax rules for this keyword.]

If specified, any up-to-16-charact string specified via the "paths" option and argument list can be translated to an up-to-64-charact string. If a specified "paths" string matches a string in the left column in the path file, then the (potentially longer) string in the right column in the same line in the path file is substituted for the original string. The rules for the syntax of a path file are the following :

  1. all lines are ASCII
  2. any line beginning with "* " is a comment and is ignored (asterisk + one or more blanks)
  3. blank lines are ignored
  4. only 2 items per line per translation : the first up to 16 characters and the second up to 64 characters. Any string longer than 16 characters MUST be enclosed in question-mark (?) delimiters.
  5. free-field interpretation, therefore embedded blanks require the "?" delimiters
  6. first item is name to be translated; must match VERBATIM with path name specified or implied via "paths" argument list
  7. second item is the corresponding path name translation

An example of such a file is the following :



*
*  Example path file for test xxx
*
*  item  1 = path to be translated : MUST
*             match VERBATIM with name given
*             specified/implied via "paths"
*
*  item  2 = resulting name translation
*
data56    ? \usr\home2\ people\test56\ ?
up  ?..\?
way-up  ? ..\..\..\ Temp46\?


Example : pathfile pathA

[Top][Bottom][Option list]

( BEpaths) (Command :learn) Details for the paths option.

[General syntax rules for this keyword.]

Indicates, in a 1:1 correspondence with the list of files specified, directory locations for the specified files. These locations supersede the current directory. If a file location thus specified is not empty ("null...") a search for the respective file will be made at the indicated location before a search is made at any other alternate locations which have been made effective via a previous execution of the "path" command. By default, all path names are blank. Up to 100 path names can be specified.

Example : paths ..\ ..\Main\ sub1\

[Top][Bottom][Option list]

( BEpolyterms) (Command :learn) Details for the polyterms option.

[General syntax rules for this keyword.]

Indicates (1) the name(s) of the dependent variable(s) and (2) the name(s) of the independent variables to use in the determination of the coefficients of a least-squares type polynomial. An example of a "polyterms" argument list is the following :

... polyterms cl alpha beta mach ...

Here, a polynomial to compute the dependent variable "cl" from existing values of the independent variables "alpha", "beta", and "mach" is desired. In general, values of the independent variables may exist because they are defined on the SIF files being processed or may exist as having been defined in a "newterms" argument list.

Another example of a "polyterms" argument list is given below. Here, the bracket construct is used to imply that more than one polynomial argument list - more than one polynomial - will be defined, each one being associated with a different dependent variable but the SAME set of independent variables.

... polyterms [ cp1 thru cp20 ] alpha mach config ...

Here, 20 polynomials to compute the dependent variables "cp1", "cp2", ... , "cp20" from existing values of the independent variables "alpha", "mach", and "config" are desired.

Either the square brackets OR the curly braces can be used to contain the set of dependent variables being defined. If the "learn" command is used within a macro then the curly braces MUST be used because the macro itself uses the square brackets as delimiters for optional macro arguments.

If "newfiles" or "newfilex" has also been specified the dependent variables to be defined can have their names changed. If no name change is specified, the calculated dependent variable values will overstore the existing dependent variable values in each of the respective files being processed. If a dependent variable is renamed, it is considered a new SIF variable and is added to each of the SIF records being processed. Dependent variables can be renamed by the specification of a left-hand side (LHS) when specifying the dependent variable name(s). An example follows :

... polyterms newcl = cl alpha beta mach ...

Here, each new file produced will contain the computed dependent variable "newcl" as well as the original dependent variable "cl". Similar renaming constructs can be specified for the bracketed multi-polynomial definitions :

... polyterms [ newcl = cl newcd = cd ] alpha mach ...
... polyterms [ newcp1 = cp1 thru cp20 ] alpha config ...
... polyterms [ newcp1 thru newcp20 = cp1 ] alpha config ...

In the first of these examples two newly-named computed dependent variables, "newcl" and "newcd", will be added to each SIF record processed. In the second of these examples the shorthand for the renaming of a series of SIF variables has been utilized. The computed dependent SIF variables "cp1", "cp2", "cp3", ... will be renamed "newcp1", "newcp2", "newcp3", ..., respectively. The third example accomplishes exactly the same result as the second; the implied series of names is, however, now on the right-hand side.

Up to 100 polynomials can be specified. The number of data points which will be considered in the determination of the coefficients of a given polynomial is a function of the number of polynomial terms specified. The maximum number of data points which will be considered is (25000/Nterms), where "Nterms" is the number of polynomial terms + 1.

ALL of the independent variables in ALL of the polynomials specified should exist in ALL of the SIF files specified or should be defined in a "newterms" argument list. If one or more of the independent variables does NOT exist in either of these two locations for a particular SIF file, then this SIF file is skipped in the current "learn" execution; ie, its data will NOT be considered when determining the polynomial coefficients.

Example : ... polyterms cl alpha beta mach ...

[Top][Bottom][Option list]

( BEprefix) (Command :learn) Details for the prefix option.

[General syntax rules for this keyword.]

If specified implies that the determined coefficients, as well as any determined bias values (see "nobias"), of each of the polynomials are to be added to each SIF file. The specified strings, in a 1:1 correspondence with the polynomials to be dealt with, as specified by the "polyterms" option/argument lists, and added in the order in which the "polyterms" option/arguments lists were encountered AND actually used in the final set of coefficients, indicate those respective sequences of characters to be prefixed onto each of the determined polynomial coefficient names and, possibly, bias names. The coefficient names are, as mentioned in the learn "report" which can be issued, are of the form "C0", C1", ... corresponding to the implied first term and the expressed second through Nth polynomial term of each polynomial. If the "nobias" option was not specified, the names of the bias quantities are of the form "B1", "B2", ... There is no "B0" type term corresponding to the "C0" term. Again, the ORDER of the coefficient and bias base names are the same as the order of the final set of coefficients (and bias values) determined and reported in the output presented via the "report" option.

Example : prefix P1 P2

[Top][Bottom][Option list]

( BEreport) (Command :learn) Details for the report option.

[General syntax rules for this keyword.]

Indicates the desired level of reporting to be done during the determination of the polynomial coefficients. A value of 0, the default, will produce NO report. A value of 1 will force a report of the determined polynomial, showing its coefficients and independent variable names and measures of the degree to which the determined polynomial calculates the original data. "r-squared", in this level of report, is defined as the percent of the sum of the squares of the deviations (original minus calculated values) of the dependent variable about y-bar which are explained by the model (determined polynomial). Here, y-bar is the mean value of the dependent variable. The closer the value is to 100 the better the ability of the computed polynomial to calculate the original dependent variable values. "max error", in this level of report, is the maximum absolute value of the deviation between original and calculated values of dependent variable.

The "dci" and "dpi" parameters are the "confidence interval" and the "prediction interval" as described in the "confidence" option discussion.

Example : report 1

[Top][Bottom][Option list]

( BEsetvoid) (Command :learn) Details for the setvoid option.

[General syntax rules for this keyword.]

A option which assigns a value to be used to replace any SIF variable "voided" as a result of a void file condition being satisfied along with a suffix match. The default "setvoid" value is -888.

Example : setvoid 880

[Top][Bottom][Option list]

( BEtags) (Command :learn) Details for the tags option.

[General syntax rules for this keyword.]

Indicates the parameters which, if present on the SIF file(s) being processed, are to have their corresponding values included in the "r-squared, maximum err" report. This is a report indicating where the maximum deviation occurred between the original and computed dependent variables. Up to 10 tag names can be specified. Tag values will in g14.7 format. be reported

Example : tags run alpha

[Top][Bottom][Option list]

( BEvfile) (Command :learn) Details for the vfile option.

[General syntax rules for this keyword.]

Indicates the name of the file containing the lines of (1) condition(s) and/or (2) condition(s) AND a suffix to use to check for (1) SIF records or (2) SIF variables to "void" (or keep, if "keep" is the first non-comment line in the void file) as each DATA record of each SIF file named in the "files" argument list is processed. "Voiding" in these two cases means two different things: (1) If all conditions on the same line are TRUE and a suffix has NOT also been specified, then the entire SIF record which has satisfied the conditions is ignored (or kept if "keep" seen). (2) If all conditions on the same line are TRUE and a suffix HAS been specified as the last item on a void file line, the SIF record is NOT ignored but, rather, the values of up to 200 SIF variables whose names end in the suffix specified are changed to the value associated with the specified or default "setvoid" value.

The structure of a valid void file must adhere to the following rules :

  1. Any line beginning with an asterisk followed by at least one blank space is considered a comment and is ignored.
  2. Blank lines are ignored.
  3. The string "keep" may be encountered as the first non-comment line.
  4. There may be an unlimited number of lines in a void file.
  5. From the point issued until "nulled", abbreviations, which can be included anywhere in the void file within a line containing conditions, can be used as a "shorthand" for any part of a condition. Abbreviations are defined in lines which BEGIN with at least the first 4 characters of the word "abbreviation" and are followed, on the same line, by PAIRS of items : the first in each pair IS the abbreviation and the second in each pair is the string to be substituted for an encountered abbreviation. Up to 1750 abbreviations can be defined. If strings to be substituted for abbreviations contain one of the 2-character logical operators (see below) any blanks which end this string will be ignored; ie, "conditions" should be constructed such that at least one blank will delimit items AFTER any substitutions have been made. Abbreviations can be activated and deactivated or replaced as the void file is processed. If an abbreviation is encountered which is the same as one which has already been defined, the new one replaces the old one unless the string to be substituted for the abbreviation is the string "null". In this case the existing abbreviation is eliminated.
  6. Each line which is not blank, not a comment, not "keep" and does not contain one or more abbreviations can contain up to 20 "conditions" and an optional suffix or it will be ignored.
  7. A "condition" is a three-item group of items which is of the form :

    name/value logical name/value

    where, "name/value" represents either a SIF name OR a constant value, and "logical" represents one of the allowable 9 1- or 2-character logical operators : eq, =, lt. <, le, ge, gt, >, or ne. An illegally formed condition will cause the line containing it to be skipped and cause a message to be issued to this effect.

  8. A suffix is any string which will be used to try to match the END of up to 200 SIF names corresponding to values in the pertinent SIF record.

Before the void file is used, by specifying "vfile" rather than "vfilen", and if the DESL session is purely interactive, the user will have the opportunity to edit the void file.

An EXAMPLE of a void file is the following :

* * a comment line * * assign "ple" and "re" abbreviations * abbr ple ?point le? re ?run =? run=10 point ge 101 point le 120 re 10 point ge 153 ple 156 un32 * * deassign "ple" abbreviation * abbr ple null re 12 point ge 101 point le 120 * *

An EXAMPLE of a "keep" file is the following :

* * a comment line * keep * * assign "ple" and "re" abbreviations * abbr ple ?point le? re ?run =? run=10 point ge 101 ple 120 re 12 point ge 101 point le 120 * *

The procedural sequence of (1) voiding, then (2) accounting for frequency via "dfreq", and then (3) averaging is always maintained.

Example : vfile void1

[Top][Bottom][Option list]

( BEvfilen) (Command :learn) Details for the vfilen option.

[General syntax rules for this keyword.]

This option has the same meaning as "vfile" except that the user is NOT given the opportunity to edit the void file before it is used.

Example : vfilen void1

[Top][Bottom][Option list]


DESL Home | Website Outline | Commands | Examples | Download | Contacts




ViGYAN, Inc.


DESL Manager at ViGYAN, Inc.

30 Research Drive
Hampton, VA 23666
Voice: (757) 865-1400
Toll Free: (800) 288-3998
FAX: (757) 865-8177

© 1998 ViGYAN, Inc.