Figure 4.

Model fitting by least-squares optimization. (A) Average excitatory postsynaptic current (EPSC, in black) together with a fit to an alpha synapse model (red) with a rate of 3.4 ms and amplitude of 13.7 pA. (B) A single unitary excitatory postsynaptic potential (EPSP, in black) was fitted to a bi-exponential function (red) with time constants of 7.1 and 161.7 ms, respectively. Cursors delimit the fitting regions. Measurements performed by Stimfit (e.g., baseline, peak) provide appropriate initial conditions to the fitting algorithm. (C) User interface allowing detailed manipulation of both initial parameters to the fitting model, and fitting options of the fitting algorithm. (D) Without the aid of the measurements to obtain adequate initial estimates, only a proportion of fits converges on the correct values (an SSE between trace and fit >0.001 is considered as a failure). If initial estimates are derived from the measurements (baseline, peak, rise time, and half-width), the fits converge correctly in all cases.