# Getting started with using iGraph for max flow

**URL:** <https://igraph.discourse.group/t/getting-started-with-using-igraph-for-max-flow/1672>\
**Category:** Usage\
**Tags:** Python\
**Created:** [23 September 2023 18:52 UTC](https://igraph.discourse.group/t/getting-started-with-using-igraph-for-max-flow/1672 "2023-09-23T18:52:21Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Rob](https://avatars.discourse-cdn.com/v4/letter/r/838e76/32.png) [@Rob](https://igraph.discourse.group/u/Rob)\
**Post date:** [23 September 2023 18:52 UTC](https://igraph.discourse.group/t/getting-started-with-using-igraph-for-max-flow/1672/1 "2023-09-23T18:52:21Z")

</div>

I’m wondering what the procedure is in iGraph for adding in edges with capacity one at a time. In NetworkX, to find max flow, I would do something like:

```auto
> G = nx.Graph()
> for edge in edges_with_capacity: # get edge info format like ('a', 'b', 5)
> G.add_edge(edge[0], edge[1], capacity=edge[2])
> max_flow = nx.maximum_flow_value(G, 'a', 'z')

```

1 . The issue is that in `edges_with_capacity` I will have repeated node names (that are the same node) or even edges (that I would want to update). In NetworkX, a new node is only added when adding an edge if the node name doesn’t already exist (otherwise the existing node is used for constructing the new edge) and adding a duplicate edge updates the existing edge, is this something I have to handle manually in iGraph?

2.When using the maxflow method of an iGraph graph, I think using the string node name for source/sink is not supported? If so, what’s the ‘correct way’ of extracting the source and sink node index given string names of the nodes, is it just g.vs[“name”].index(“name of node”)?

3.Similar to the above, what’s the ‘correct way’ to retrieve the edge capacity given just the string name of the to/from vertex?

---

<div class="post-metadata">

**Author:** ![tamas](https://yyz2.discourse-cdn.com/free1/user_avatar/igraph.discourse.group/tamas/32/1261_2.png) [@tamas](https://igraph.discourse.group/u/tamas)\
**Post date:** [27 September 2023 10:36 UTC](https://igraph.discourse.group/t/getting-started-with-using-igraph-for-max-flow/1672/2 "2023-09-27T10:36:41Z")

</div>

In igraph, adding an edge to a nonexistent vertex yields a `ValueError`:

```python
>>> g.add_edge("a", "b")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "[...]/igraph/basic.py", line 24, in _add_edge
    graph.add_edges([(source, target)])
  File "[...]/igraph/basic.py", line 42, in _add_edges
    res = GraphBase.add_edges(graph, es)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: no such vertex: 'a'

```

But it is easy to create a helper function:

```auto
def add_edge_by_vertex_name(g: Graph, src: str, dest: str) -> None:
    if src not in g.vs["name"]:
        g.add_vertex(src)
    if dest not in g.vs["name"]:
        g.add_vertex(dest)
    g.add_edge(src, dest)

```

Re updating an existing edge: you need to check manually whether the edge exists already or not as it is possible to add multiple edges between the same pair of vertices, so doing something like `g.add_edge("a", "b", capacity=123)` will add a new edge even if another edge exists already.

So, again, extending our helper function:

```auto
def update_weight(g: Graph, src: str, dest: str, **kwds) -> None:
    if src not in g.vs["name"]:
        g.add_vertex(src)
    if dest not in g.vs["name"]:
        g.add_vertex(dest)
    eid = g.get_eid(src, dest, error=False)
    if eid < 0:
        eid = g.add_edge(src, dest).eid
    for k, v in kwds.items():
        g.es[eid][k] = v

```

1. igraph supports referring to a node by its name in every place where a numeric vertex ID is accepted. Names are looked up from the `name` vertex attribute. If you find a place in igraph where a vertex ID works but a vertex name does not then it’s most likely a bug.

2. Something like this:

```python
eid = g.get_eid(source, target)
print(g.es[eid]["capacity"])

```

There are alternatives like `g.es.select(_source=source, _target=target)` but IMHO `g.get_eid()` is more explicit.