Fix conversion from heterogeneous to homogeneous connector when deleting last synapse of a type #566
Conversation
jakobj
force-pushed
the
fix565_sp_disables_connectors
branch
from
b74aff5
to
5eb1b47
Compare
| Synopsis: (issue-565) run -> NEST exits if test fails | ||
|
|
||
| Description: | ||
| This test provokes a switch from a heterogeneous connector to a homogeneous connector using the structural plasticity framework and asserts that the resulting homogeneous connector is correctly used. |
There was a problem hiding this comment.
This line could use a line break
|
|
||
| % Delete the first connection again, causing the heterogeneous | ||
| % connector to be turned into a homogeneous connector | ||
| [2] cvgidcollection [1] cvgidcollection << /rule /all_to_all >> << /model /static_synapse >> Disconnect_g_g_D_D |
There was a problem hiding this comment.
Same here, maybe add a line break after /all_to_all >>
| % The spike detector should have recorded three events | ||
| { | ||
| 3 GetStatus /n_events get 3 eq | ||
| } assert_or_die |
There was a problem hiding this comment.
I see from your example in the issue #565 how this ensures that everything works properly, but wouldn't it be much more fail-safe to add another simulation to this unit test where one neuron with the same parameters is just connected to the spike detector (without disconnecting a temporary connection) and then (after both simulations) to compare the actual spike times recorded by the spike detector?
| M_ERROR setverbosity | ||
|
|
||
| % Create neurons and spike detector | ||
| /iaf_neuron 2 << /I_e 400. >> Create /neurons Set |
There was a problem hiding this comment.
Use iaf_psc_alpha instead (see PR #526)
heplesser
added
ZC: Kernel
last synapse of a type
jakobj
force-pushed
the
fix565_sp_disables_connectors
branch
from
5eb1b47
to
1915f1c
Compare
|
Thanks @janhahne for the quick review. I have pushed a new version addressing all your comments. |
|
Dear @jakobj thanks for fixing this issue. I agree with the solution 👍 from my side. |
|
@heplesser @jakobj Yes, he did! Nice work 👍 |
See #565 for a discussion of the issue. This PR fixes that issue and provides a test.
I suggest @sdiazpier and @suku248 as reviewers.