The Distributed_Vector Class is used to describe a distributed vector (existing across all the processors) in the CÆSAR Code Package. A Distributed_Vector is part of the overall data structure strategy in CÆSAR, which is made up of the following classes: Base_Structure, Data_Index, Assembled_Vector, Distributed_Vector, Overlapped_Vector, and Collected_Array. A description of the overall data structure strategy can be found in the Data_Structures Module.
The form of a Distributed_Vector object is given by:
Vector ( [dim1, [dim2, [dim3,]]] Structure_Axis )or
Vector ( dim_ragged_right, Structure_Axis ) --> not implementedwhere Structure_Axis refers to the axis which is spread across the processors.
Distributed_Vector public procedures:
| Fundamental procedures | |||
| Initialize | Initializes a Distributed_Vector object. | ||
| Finalize | Finalizes a Distributed_Vector object. | ||
| Valid_State | Returns false iff a Distributed_Vector object is in an invalid state. | ||
| Initialized | Returns true iff a Distributed_Vector object has been initialized. | ||
| Operations | |||
| Assemble | Sets an Assembled_Vector object to a Distributed_Vector object by assembling the data on the IO PE (also has an assignment interface). | ||
| Distribute | Sets a Distributed_Vector object to an Assembled_Vector object by distributing the data from the IO PE to all the PEs (also has an assignment interface). | ||
| Get_Values | Gets the values from a Distributed_Vector object and returns them in a bare naked vector (also has an assignment interface). | ||
| Locus | Returns the locus of the Distributed_Vector object. | ||
| Name | Returns the name of the Distributed_Vector object. | ||
| Output | Writes out the Distributed_Vector object. | ||
| Set_Values | Sets the values of the Distributed_Vector object to a bare naked vector (also has an assignment interface). | ||
| Set_Version | Sets the version number of the Distributed_Vector object (also has an assignment interface). | ||
| Version | Returns the version number of the Distributed_Vector object. | ||
Distributed_Vector public defined type:
| Distributed_Vector type | |||
| Dimensionality | The number of dimensions that the index``vector'' has, including the dimension that is spread over the processors. ``Ragged_Right'' indices are signified by a Dimensionality of -1. (Ragged_Right is not yet implemented.) | ||
| Dimensions | The extents of the dimensions that the ``vector'' has, including the dimension that is spread over the processors, which is last. | ||
| Initialized | Initialization status. | ||
| Name | The name for this variable (especially useful in a vector of Distributed Vectors). | ||
| NValues_PE | Number of values on this PE. | ||
| NValues_Total | Total number of values in the entire vector (including all PEs). | ||
| NValues_Vector | A vector containing the number of values on each PE. | ||
| Structure | Basic data structure for the axis that is spread over the processors. | ||
| Values{n} | Values in the vector, with a different length on each PE. Values may have either 1, 2, 3, or 4 dimensions (n = 1, 2, 3, or 4), or be a ragged right array (n = RR). The last dimension is always the dimension to be spread across the processors. Only one of the variables will be allocated for a given object. Ragged right arrays have not been implemented yet. | ||
| Version | Version number which is incremented every time the vector is modified, or is synced with the version number of a data structure that it depends on when it is updated. | ||
The Distributed_Vector Class code listing contains additional documentation. The Distributed_Vector Class also contains a Unit Test Program.