Plotting Graphs¶

Visualizing Graphs can easily be automated using datastructures and methods from the visualization package. This notebook shows some features and flows on how to display and format graphs. Execute the cell below to read the example design and retrieve the circuit graph object for further customization.

In [ ]:

import netlist_carpentry

FILE_PATH = "../files/decentral_mux.v"
circuit = netlist_carpentry.read(FILE_PATH, top="decentral_mux")
circuit.top.optimize()
GRAPH = circuit.top.graph()

import netlist_carpentry FILE_PATH = "../files/decentral_mux.v" circuit = netlist_carpentry.read(FILE_PATH, top="decentral_mux") circuit.top.optimize() GRAPH = circuit.top.graph()

Within the visualization package, there is a class Plotting, which is intended to make visual depiction of graphs easier, providing a bunch of methods for customized graph representations. When instantiating the Plotting class, it accepts

• a graph, which will be formatted,
• optionally a Format object (will be explained later),
• optionally a default node color,
• optionally a default node size. Execute the cell below to create a basic, uncustomized plot from the graph first.

In [2]:

from netlist_carpentry.core.graph.visualization import Plotting

plot = Plotting(GRAPH)
plot.show()

from netlist_carpentry.core.graph.visualization import Plotting plot = Plotting(GRAPH) plot.show()

Using Plotting.set_labels_default, each node receives a label. For input or output port nodes, it is the name of the port. For instances, it is the instance type by default.

Info: Plotting.set_labels_default accepts a parameter show_instance_names, with which the plot will contain the instance names instead of the types (if the parameter is set to True). However, this may result in a very incomprehensible plot, as each primitive gate instance will receive a very generic name.

In [3]:

plot.set_labels_default()  # Alternative: plot.set_labels_default(show_instance_names=True)
plot.show()

plot.set_labels_default() # Alternative: plot.set_labels_default(show_instance_names=True) plot.show()

The Plotting class supports creating and managing reusable formats (more on this in the dedicated notebook). This all boils down to the Plotting.set_format and Plotting.format_node methods. With set_format(format_name, **format_parameters), a format template can be created or modified. The format_node(node_name, format_name) method then applies the corresponding format template to the given node. Execute the cell below to format a single node to a bright green color.

In [4]:

plot.set_format("green_format", node_color='#00FF00')
plot.format_node('SELECT_I', "green_format")
plot.show()

plot.set_format("green_format", node_color='#00FF00') plot.format_node('SELECT_I', "green_format") plot.show()

Using Plotting.format_nodes, nodes can be batch-formatted. Like the format_node method, it takes a format name, but the first parameter is a predicate function (can be a lambda) that takes two parameters:

• the node name,
• the node data dictionary, which can then be used to filter nodes according to their respective name or data.

Execute the cell below to color all nodes representing instances in cyan.

In [5]:

plot.set_format(name="cyan_format", node_color="cyan")
plot.format_nodes(lambda node_name, node_data: node_data["ntype"] == "INSTANCE", format_name="cyan_format")
plot.show()

plot.set_format(name="cyan_format", node_color="cyan") plot.format_nodes(lambda node_name, node_data: node_data["ntype"] == "INSTANCE", format_name="cyan_format") plot.show()

Nodes can also be of different sizes, also applicable via the set_format method. Execute the cell below to enlargen all nodes representing OR gate instances and make them orange as well.

Info: Notice how previous formatting is overwritten, making the previously cyan nodes now orange.

In [6]:

plot.set_format("large_nodes", node_color="#FF6600", node_size=2500)
plot.format_nodes(lambda node_name, data: data.get("nsubtype") == "§or", format_name="large_nodes")
plot.show()

plot.set_format("large_nodes", node_color="#FF6600", node_size=2500) plot.format_nodes(lambda node_name, data: data.get("nsubtype") == "§or", format_name="large_nodes") plot.show()

In an alternative case, all nodes with similar names could possibly be highlighted. This is also possible via the lambda function, now utilizing the node name instead of the node data. Execute the cell below to color all nodes with data (case-insensitive) in their names in dark blue.

In [9]:

plot.set_format("dark_blue", node_color="#0000AA")
plot.format_nodes(lambda node_name, data: "data" in node_name.lower(), format_name="dark_blue")
plot.show()

plot.set_format("dark_blue", node_color="#0000AA") plot.format_nodes(lambda node_name, data: "data" in node_name.lower(), format_name="dark_blue") plot.show()

Another useful function is Plotting.format_in_out which formats all inputs and outputs according to the given format names. Execute the cell below, to recreate a similar image to the one above, but with different coloring for the input and output port nodes.

In [8]:

plot.set_format(name="brown_format", node_color="brown")
plot.set_format(name="magenta_format", node_color="magenta")
plot.format_in_out(in_format="brown_format", out_format="magenta_format")
plot.show()

plot.set_format(name="brown_format", node_color="brown") plot.set_format(name="magenta_format", node_color="magenta") plot.format_in_out(in_format="brown_format", out_format="magenta_format") plot.show()