Negative labels in ModeClusterState of nested partitions

graph-tool
1

Hi Prof. Peixoto,

My graph tool version is 2.59
OS is linux ubuntu 22

ModeClusterState results in some negative labels in nested partitions.
How to resolve this problem?

My minimal working code is:

import numpy as np 
import pickle 
import graph_tool.all as gt 

gt.seed_rng(100)
np.random.seed(100)

with open(f'bs.pkl', 'rb') as f:
    bs = pickle.load(f)

pmode = gt.ModeClusterState(bs, nested=True)
gt.mcmc_equilibrate(pmode, wait=1, mcmc_args=dict(niter=1, beta=np.inf))

print(pmode.bs)

Uploading the necessary partitions, bs:
bs.pkl (524.1 KB)

Uploading the output of my code, pmode.bs:
pmode_bs.txt (522.6 KB)

As we can see, some labels are -1.
Please suggest some solution.

Many thanks,
Govinda

2

This is totally normal.

A value of -1 means that a given group does not have an assignment in the layer above, since it’s empty in the layer below.

3

Thank you!

Now when I create a NestedBlockState afresh, I get the following error:

import numpy as np 
import pickle 
import graph_tool.all as gt 

gt.seed_rng(100)
np.random.seed(100)

with open(f'bs.pkl', 'rb') as f:
    bs = pickle.load(f)

with open(f'graph.pkl', 'rb') as f:
    g = pickle.load(f)
    
pmode = gt.ModeClusterState(bs, nested=True)
gt.mcmc_equilibrate(pmode, wait=1, mcmc_args=dict(niter=1, beta=np.inf))

state = gt.NestedBlockState(g, bs=pmode.bs[0])

ERROR:

TypeError: _vt_copy() got an unexpected keyword argument 'value_type'

Is this expected?

4

This should be instead:

bs = nested_partition_clear_null(pmode.bs[0])
state = gt.NestedBlockState(g, bs=[g.new_vp("int", val=bs[0])] + bs[1:])
1 Like