What data types can be generated?

The available data types depend from the kind of wanted generation:

Generator type	Double	Vector (Double)	Matrix (Double)	Integer	Boolean	String	Map
Fixed values	Yes	Yes*	Yes*	Yes	Yes	Yes	No
Random values	Yes	Yes	Yes	Yes	Yes	No	No
Values from file interpolation	Yes	No	No	No	No	No	No
Values from file injection	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note: expected format for Map is (key: String, value: openfluid::core::Value)

Floating-point variables

*The non-scalar double generator can be used either with a single value for all, or a value for each cell of the container If a single value is given, a parameter varsize is required with expected dimension (eg "[3]" for a vector or "[2,4]" for a matrix).

An example for every time step and every unit of the given unit class, for a generator of vector of size 4 containing doubles:

"3" with varsize parameter being [4] will produce as variable value [3.0, 3.0, 3.0, 3.0]
"[1,4,2.2,3]" will produce as variable value [1.0, 4.0, 2.2, 3.0]

An option is available for the random generator, to be able to have identical or different values inside the container. Eg with a vector of 4 double:

same value activated: t0: [2.1, 2.1, 2.1, 2.1], t1: [43, 43, 43, 43]
same value disabled: t0: [4.5, 2.0, 22.1, 1.7], t1: [10.2, 2.3, 7,4, 9.9]

Boolean

Currently the random generator for boolean gives a balanced probability between true and false (50% each)

File formats for generators

Note: Currently, these files formats are used by interp and inject generators only.

Sources file

The sources file format is an XML based format which defines a list of sources files associated to an unique ID.
The sources must be defined in a section delimited by the <datasources> tag, inside an <openfluid> tag and must be structured following these rules:

Inside the <datasources> tag, there must be a set of <filesource> tags
Each <filesource> tag must bring an ID attribute giving the identifier of source, and a file attribute giving the name of the file containing the source of data. The files must be placed in the input directory of the simulation.

<?xml version="1.0" standalone="yes"?>
<openfluid>
 
  <datasources>
    <filesource ID="1" file="source1.dat" />
    <filesource ID="2" file="source2.dat" />    
  </datasources>
  
</openfluid>

Note: As a sources file is not part of the standard input dataset, it must not have a .fluidx extension. Using the .fluidx extension may lead to an unexpected behaviour (such as deletion of the sources file).
The recommended extension for this file is .xml (e.g. sources.xml)

An associated source data file is a two columns text file, containing a serie of values in time. The first column is the date using the ISO format YYYY-MM-DD'T'HH:MM:SS. The second column is the value itself.

1999-12-31T12:00:00 -1.0
1999-12-31T23:00:00 -5.0
2000-01-01T00:30:00 -15.0
2000-01-01T00:40:00 -5.0
2000-01-01T01:30:00 -15.0

Distribution file

A distribution file is a two column file associating a unit ID (first column) to a source ID (second column).

Type-specific generator

By default, the values injected by inject generators are Double type.
To inject data of a specific type, add the vartype attribute to the generator element with appropriate type as value.

Warning: String representation of values defined in source files must match the string representation of specified variable type.
See String representation of values

The available values for vartype depend from the openfluid::core::Value type:

OpenFLUID value	vartype
openfluid::core::StringValue	string
openfluid::core::BooleanValue	boolean
openfluid::core::IntegerValue	integer
openfluid::core::VectorValue	vector
openfluid::core::MatrixValue	matrix
openfluid::core::MapValue	map
openfluid::core::NullValue	null

Example of a boolean inject generator:

<generator varname="gas.atm.V.windspeed.bool" unitsclass="AU" vartype="boolean" method="inject" enabled="true">
    <param name="sources" value="sources.xml"/>
    <param name="distribution" value="distribution.dat"/>
</generator>

Table of Contents