nnmt.linear_stability._solve_chareq_lambertw_constant_delay¶
-
nnmt.linear_stability.
_solve_chareq_lambertw_constant_delay
(branch_nr, tau, delay, connectivity)[source]¶ Calcs the most critical eigenvalue for neural networks with constant delay.
Uses the Lambert W function to solve the characteristic equation for the respective system. This function can be applied to networks without (Eq. 1 in Helias et al. [2013]) and with (Eq. 1 in Senk et al. [2020]) a spatially structured connectivity. It computes the most critical temporal eigenvalue given by Eq. 8 in Helias et al. [2013] or Eq. 7 in Senk et al. [2020] respectively for any given branch number.
Parameters: - branch_nrint
Branch number.
- taunp.array
Time constants in s.
- delaynp.array
Delays in s.
- connectivitynp.array
Matrix defining the connectivity. For non-spatial networks, this is just the weight matrix. For spatial networks, this is an effective connectivity matrix; each element is the weight multiplied with the Fourier transform of the spatial profile at the wave number k for which the characteristic equation is to be evaluated.
Returns: - eigenvalnp.complex
Temporal eigenvalue solving the characteristic equation.