Visualize Task Graphs

Task graph can be visualized using dask.visualize and .visualize on the collections.

High level task graphs

Below code snippet shows High level task graphs visualization.

x = da.ones((15, 15), chunks=(5, 5))
 
y = x.sum()
 
y.dask.visualize()

It shows graph at higher level, without the underlying nodes and subgraphs.

Low level task graphs

Low level task graphs includes all the nodes and dependencies. It is used when we wan to understand and analyse the graph thoroughly.

dask.visualize(y)
 
# Or
 
y.visualize()

It can take arguments such as engine to used. There are two engine supported, graphvis and cytoscape. Graph layout can also be changed using argument rankdir with values like LR, RL, TB, BT.

If we want optimized graph, we can use optimize_graph argument.