Datatypes

Model input and outputs are described in model YAML files by several properties, including type. The type assigned to an input/output determines how yggdrasil will serialize data that is passed to/from the input/output from/to other models and how data objects are mapped between native data types in the different programming languages (see here for the mappings). Simple types (e.g. float, int) can be specified by strings, while more complex datatypes can be described directly in the YAML file using a JSON Schema (See below). When the YAML specification files are read in these datatypes are validated against the yggdrasil metaschema.

Primary Datatypes

yggdrasil supports all of the following basic data types which include the core JSON data types, as well as several additional datatypes that yggdrasil defines for convenience. Simple datatypes can be specified via strings, while collection datatypes require a schema that can be used to validate data being passed (dynamic arrays/objects are not yet supported).

Simple Datatypes

Type ========- boolean	Description -================================================================================= true or false
integer	Integer numbers
null
number	Integer or decimal numbers
string	Arrays of characters

Collection Datatypes

Type =======- array	Description -================================================================================= Container for sequences of values
object	Container for mappings of key and value pairs

yggdrasil Datatypes

Type =========- class	Description -================================================================================= Python class
function	Python function
instance	Python class instance
ndarray	N-dimensional arrays of memory contiguous arrays of scalars
obj	ObjWavefront 3D structure
ply	Ply 3D structure
scalar	Scalar quantities that can have units, variables precision, or an encoding (for unicode scalars)
schema	JSON schema

Information on adding new datatypes can be found here