How can I print the result using Python API? #215

gzzyyxh · 2024-11-21T09:43:41Z

Is there anything missing from the code shown on the website?

import os
import kahypar as kahypar

num_nodes = 7
num_nets = 4

hyperedge_indices = [0,2,6,9,12]
hyperedges = [0,2,0,1,3,4,3,4,6,2,5,6]

node_weights = [1,2,3,4,5,6,7]
edge_weights = [11,22,33,44]

k=2

hypergraph = kahypar.Hypergraph(num_nodes, num_nets, hyperedge_indices, hyperedges, k, edge_weights, node_weights)

context = kahypar.Context()
context.loadINIconfiguration("<path/to/config>/km1_kKaHyPar_sea20.ini")

context.setK(k)
context.setEpsilon(0.03)

kahypar.partition(hypergraph, context)

HOW can I print the partitioning result?

The text was updated successfully, but these errors were encountered:

SebastianSchlag · 2024-11-22T01:33:29Z

Hey @gzzyyxh ,

You can iterate over all the nodes of the hypergraph by using

kahypar/python/module.cpp

Line 164 in efa12a9

.def("nodes", [](Hypergraph &h) {

and then call

kahypar/python/module.cpp

Line 123 in efa12a9

.def("blockID", &Hypergraph::partID,

for each of the nodes.

SebastianSchlag · 2024-11-22T01:34:06Z

If you want to write out the actual partition file, you can add write-partition=1 to the .ini config file you use for partitioning. That should create a partition file named <hypergraph>.hgr.part<k>.epsilon<epislon>.seed<seed>.KaHyPar in the same folder as the actual hypergraph.

gzzyyxh · 2024-11-22T02:12:50Z

Hey@SebastianSchlag ,
Thank you for getting back to me. I'm sorry, but I don't understand how to use the code you mentioned. Could you provide a simple example? I added write-partition=1 to the km1_kKaHyPar_sea20.ini configuration file, but nothing is generated after running it.

N-Maas · 2024-11-22T11:51:33Z

A usage example would be this:

for node in hypergraph.nodes():
  blockID = hypergraph.blockID(node)
  print(f"Node {node} is mapped to block {blockID}")

gzzyyxh · 2024-11-23T13:32:46Z

Hi @N-Maas ,
Thanks so much. Last question:
I added two for-loop to print the result:

for node in hypergraph.nodes():
    blockID = hypergraph.blockID(node)
    print(f"Node {node} is mapped to block {blockID}")

for net in hypergraph.edges():
    blockID = hypergraph.blockID(net)
    print(f"Net {net} is mapped to block {blockID}")

And I got this output:

Objectives: 
Hyperedge Cut  (minimize) = 99 
SOED           (minimize) = 198 
(k-1)          (minimize) = 99 
Absorption     (maximize) = 64.1667 
Imbalance                 = 0 

Partition sizes and weights:  
|part 0 | = 3  w( 0 ) = 14 
|part 1 | = 4  w( 1 ) = 14 
...
Node 0 is mapped to block 1
Node 1 is mapped to block 0
Node 2 is mapped to block 1
Node 3 is mapped to block 1
Node 4 is mapped to block 0
Node 5 is mapped to block 1
Node 6 is mapped to block 0
Net 0 is mapped to block 1
Net 1 is mapped to block 0
Net 2 is mapped to block 1
Net 3 is mapped to block 1

Could you tell me what's SODE, (k-1), Absorption?
And, as I have learned, the net is said to be cut if it connects more than one part (block here), uncut otherwise. So what does "Net 1 is mapped to block 0" mean? Net 1 connects two blocks here. I think it should be outside, not in any blocks.

SebastianSchlag · 2024-11-24T18:02:50Z

Could you tell me what's SODE, (k-1), Absorption?

These are other quality metrics. SOED is the shorthand for the Sum-Of-External-Degrees metric, (k-1) is the shorthand for the connectivity (sometimes also called ($\lambda-1$)) metric.

And, as I have learned, the net is said to be cut if it connects more than one part (block here), uncut otherwise. So what does "Net 1 is mapped to block 0" mean? Net 1 connects two blocks here. I think it should be outside, not in any blocks.

The blockID method returns the block the vertex with the given id is assigned to. Nets are not assigned to any blocks. So this code:

for net in hypergraph.edges():
    blockID = hypergraph.blockID(net)
    print(f"Net {net} is mapped to block {blockID}")

actually just returns the block IDs of the first 4 vertices (with ids 0, 1, 2, and 3), since hypergraph.edges() just enumerates the ids of all the edges in the hypergraph.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

How can I print the result using Python API? #215

How can I print the result using Python API? #215

How can I print the result using Python API? #215

How can I print the result using Python API? #215

Comments