OverflowError when setting negative synaptic weight PointNet

Hi!

I am trying to set up a network of point neuron model using PointNet, but I encountered an error when trying to set inhibitory synapses.
I tried to use this example on the BMTK GitHub, but I got the following error:

---------------------------------------------------------------------------
OverflowError                             Traceback (most recent call last)
Cell In[11], line 7
      4 configure.build_env()
      6 network = pointnet.PointNetwork.from_config(configure)
----> 7 sim = pointnet.PointSimulator.from_config(configure, network)
      8 sim.run()

File [/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/pointsimulator.py:340](http://localhost:8888/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/pointsimulator.py#line=339), in PointSimulator.from_config(cls, configure, graph, n_thread)
    337 graph.build_nodes()
    339 graph.io.log_info('Building recurrent connections')
--> 340 graph.build_recurrent_edges()
    342 io.log_info('Network created.')
    343 return network

File [/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/pointnetwork.py:175](http://localhost:8888/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/pointnetwork.py#line=174), in PointNetwork.build_recurrent_edges(self, force_resolution)
    172     return
    174 for edge_pop in recurrent_edge_pops:
--> 175     for edge in edge_pop.get_edges():
    176         nest_srcs = self.gid_map.get_nestids(edge_pop.source_nodes, edge.source_node_ids)
    177         nest_trgs = self.gid_map.get_nestids(edge_pop.target_nodes, edge.target_node_ids)

File [/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/core/sonata_reader/network_reader.py:191](http://localhost:8888/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/core/sonata_reader/network_reader.py#line=190), in SonataEdges.get_edges(self)
    189 edge_adaptor = self._edge_adaptors[edge_group.group_id]
    190 if edge_adaptor.batch_process:
--> 191     for edge in edge_adaptor.get_batches(edge_group):
    192         yield edge
    193 else:

File [/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/sonata_adaptors.py:396](http://localhost:8888/opt/anaconda3/envs/Test_BMTK/lib/python3.12/site-packages/bmtk/simulator/pointnet/sonata_adaptors.py#line=395), in PointEdgeAdaptor.get_batches(self, edge_group)
    393         raise Exception('Could not find syn_weight value')
    395     # caluclate weight
--> 396     type_params[edge_id]['weight'] = nsyns * syn_weight
    398 yield PointEdgeBatched(source_nids=grp_vals['src_nids'].values, target_nids=grp_vals['trg_nids'].values,
    399                        nest_params=type_params[edge_id])

OverflowError: Python integer -550 out of bounds for uint32

I tried with a custom network, but I had the same error. From what I understand, I cannot assign negative synaptic weight, but then, I don't know how to make a specific connection inhibitory.

Do you have any idea why I am getting this error and what I could do?

I am using BMTK version 1.1.2 in python 3.12.11.

Thank you for your help!

Best,

Julien

Start with simulator/pointnet/sonata_adaptors.py in PointEdgeAdaptor.get_batches, where the traceback shows the synaptic weight is calculated, then trace the edge values through simulator/core/sonata_reader/network_reader.py. Reproduce the failure with examples/ei_networks/pointnet/build_network.py and determine how negative weights should be represented. Done means the example and a custom inhibitory PointNet network initialize without the reported OverflowError.