Hello Remy, I don't have any numbers of Hill-Tononi vs aeif_cond_exp, but given that Hill-Tononi has a 16-dimensional and the aeif_cond_exp a 4-dimensional state vector, there will probably be a noticeable difference. Part of the simplicity of the aeif_cond_exp is in the synapses, since the model only has a two synaptic time constants, one excitatory and one inhibitory. Adding NMDA and GABA_B with different time constant would add at least two dimensions, but probably four as you probably want different rise and decay constants (you could experiment with the aeif_cond_{alpha,beta}_multisynapse models). A crucial question is if want to "lump" the NMDA and GABA_B synapses in the same way as synapses in NEST generally are lumped: we treat all spikes arriving through the excitatory synaspe of a model neuron equal (and correspondingly for the inhibitory synapse). As long as synapse activation is linear, this is unproblematic. But if activation is non-linear (as may be the case for second-messenger synapses), this may no longer hold. Then, one would strictly speaking need to treat every incoming synapse individually, resulting in very large state vectors (O(10^3)) and correspondingly poor performance. Best, Hans Ekkehard -- Prof. Dr. Hans Ekkehard Plesser Head, Department of Data Science Faculty of Science and Technology Norwegian University of Life Sciences PO Box 5003, 1432 Aas, Norway Phone +47 6723 1560 Email hans.ekkehard.plesser(a)nmbu.no Home http://arken.nmbu.no/~plesser On 14/02/2022, 16:02, "cagnol(a)ksvi.mff.cuni.cz" <cagnol(a)ksvi.mff.cuni.cz> wrote: Hello, We would like to use neurons models with NMDA channels in our spiking neuron model. We're still unsure whether we will use a native neuron model in Nest or whether we will implement our own one in NESTML. My understanding is that the only model in Nest which does that is the Hill - Tononi model, which seems rather complex. How fast would you roughly expect a Hill - Tononi neuron network to run compared to a network made of aeif_cond_exp neurons? Do you know by any chance any example of adex NESTML models which implement NMDA and Gaba_B channels? Also, one last question not really related to the previous ones: Is there any way to model synaptic reliability in Nest? Thanks a lot, Remy, _______________________________________________ NEST Users mailing list -- users(a)nest-simulator.org To unsubscribe send an email to users-leave(a)nest-simulator.org