{"id":707,"date":"2018-09-01T15:45:48","date_gmt":"2018-09-01T13:45:48","guid":{"rendered":"https:\/\/project.inria.fr\/damaris\/?page_id=707"},"modified":"2023-09-05T11:16:51","modified_gmt":"2023-09-05T09:16:51","slug":"paraview-connector","status":"publish","type":"page","link":"https:\/\/project.inria.fr\/damaris\/paraview-connector\/","title":{"rendered":"ParaView Connector"},"content":{"rendered":"

In order to benefit from ParaView connector that has been implemented in Damaris, it is necessary to build Damaris on a machine that ParaView Catalyst is already installed. You may use this link<\/a> to install ParaView in your machine. Please note that if you are going to have your simulation and visualization in two different nodes, it is recommended not to install ParaView GUI on simulation nodes. Also, please ensure that the version of Paraview that Damaris is compiled against matches the Paraview being used for visualization of the simulation exactly.<\/p>\n

Note 1:<\/strong> The first version of ParaView backend for Damaris (Damaris 1.3.0), has been developed and tested with ParaView 5.5.0. Damaris v1.3.1+ has been tested with 5.5.0 and 5.8.0 of Paraview.<\/p>\n

Note 2:<\/strong> In situ visualization with VisIt and ParaView in Damaris are quite similar. As a result, it is recommended to check the in situ visualization with VisIt documentation<\/a> before working through this tutorial.<\/p>\n

Example code<\/h4>\n

A working example that closely matches the following description is available in the \/examples\/paraview<\/em> directory, of either the git repository or installed if Damaris was configured with -DENABLE_EXAMPLES=ON and is named image.cpp<\/em>. It has an accompanying Catalyst script code (image.py<\/em>) and Damaris configuration file (image.xml<\/em>).<\/p>\n

ParaView Options<\/h4>\n

When using ParaView Catalyst in support of in situ data analysis and visualization without Damaris, the instrumented simulation starts loading a set of Python scripts. These scripts, include necessary steps to produce images, generate plots, and also extract derived information such as polygonal data in support of in situ processing and visualization. During the simulation execution, any analysis and visualization output is generated in synchronous fashion (i.e., while the simulation is running).<\/p>\n

The mentioned Python scripts are used in the same way within Damaris too. In this context, simulation processes pass the datasets to the dedicated core(s) using shared memory. Afterwards, the dedicated core(s) load the Python scripts and use the steps inside each script to process data.<\/p>\n

In the context of Damaris, ParaView should be enabled in the xml configuration files. The Python scripts should be introduced in the same file as well. To do that, it is necessary to create a <paraview> tag in the xml configuration file and mention the Python script files like this:<\/p>\n

<paraview>\n    <script>\/path\/to\/the\/first\/script1.py<\/script>\n    <script>\/path\/to\/the\/first\/script2.py<\/script>\n<\/paraview><\/pre>\n
Description of meshes, fields and variables<\/h4>\n
Definition of meshes, fields and variables is as same as the ones that were described for VisIt connector here<\/a>. We will describe them once more in this tutorial with an example. Please note that in the current version of ParaView backend for Damaris (version 1.3.0), only rectilinear<\/strong> and curvilinear<\/strong> meshes are supported.<\/p>\n
Example<\/h5>\n
Suppose that you are going to decompose a 3D array with 100x100x100 dimensions over some processes (let’s say the number of processes is size<\/em>). The decomposition will happen over the Z axis, as a result, each process will work on a 3D array with the dimensions of (100 , 100 , 100\/size). To do so, doing the following steps is necessary:<\/p>\n
1. Parameters <\/span><\/strong><\/h6>\n
To make the xml configuration file more flexible, it is better to define every important data item as an parameter. To do so, we can define the following parameters in the xml file:<\/p>\n
<parameter name=\"WIDTH\" type=\"int\" value=\"100\" \/>\n<parameter name=\"HEIGHT\" type=\"int\" value=\"100\" \/>\n<parameter name=\"DEPTH\" type=\"int\" value=\"100\" \/>\n<parameter name=\"size\" type=\"int\" value=\"2\" \/>\n<\/pre>\n
As the parameter names imply, WIDTH, HEIGHT and DEPTH are defined as the dimensions of the main data to be written. The parameter size<\/em> is used for passing the number of simulation processes (not dedicated ones) to Damaris. This parameter can be set in the simulation code and at runtime using API function damaris_parameter_set()<\/em> .<\/p>\n
2. Variables<\/span><\/strong><\/h6>\n
Variables are those data that are written to Damaris at the end of each iteration. In this case, we should define a variable with its attributes like this:<\/p>\n
<variable name=\"pressure\" type=\"scalar\" layout=\"cells\" mesh=\"mesh\" centering=\"zonal\" \/>\n<\/pre>\n
In this definition, pressure is the name of the variable, that is going to be written at each iteration. Scalar is the type of the data. In addition, layout and the mesh of this variable are defined as cells<\/em> and mesh<\/em>. These two items are discussed later in this document. Finally, the centering attribute determines that is variable should be visualized using internal volumes of the mesh (zonal) or points of the mesh (nodal).<\/p>\n
3. Layout<\/strong><\/h6>\n
The layout of a variable defines the structure of its data in memory<\/em>. The definition of each layout includes the number of its dimensions, extents along each direction, the data type, etc. For the mentioned variable, its layout is defined as:<\/p>\n
<layout name=\"cells\" type=\"double\" dimensions=\"DEPTH\/size,HEIGHT,WIDTH\"\nglobal=\"DEPTH,HEIGHT,WIDTH\" ghosts=\"0:0,0:0,0:0\" \/>\n<\/pre>\n