It can be shown numerically that when equations (2) and (3)
exhibit periodically firing action potentials,
For fast-slow systems, a useful procedure for understanding the dynamics is to consider the nullclines. The V-nullcline is the set in space for which , while the n-nullcline is the set in space for which . It is messy to obtain analytical expressions for the nullclines of equations (2) and (3), but they can be computed numerically using the fzero command in Matlab which finds solutions of the equation . To do this, download the following three programs. First, nullclines.m:
global vna vk vl gna gk gl c I v
vna=50;
vk=-77;
vl=-54.4;
gna=120;
gk=36;
gl=.3;
c=1;
I = 20;
for i=1:130
v = -80 + i + 0.01;
vv(i) = v;
nullv(i) = fzero('rhsV',0.5);
nulln(i) = fzero('rhsn',0.5);
end
figure(2)
hold on;
plot(vv,nullv,'b');
plot(vv,nulln,'r');
axis([-80 60 0 1]);
xlabel('V');
ylabel('n');
Next, rhsV.m, which is the righthand side of equation (2):
function r = rhsV(n)
global vna vk vl gna gk gl c I v
r = (I - gna*(m_inf(v))^3*(0.8-n)*(v-vna) - gk*n^4*(v-vk)-gl*(v-vl))/c;
function r = rhsn(n)
global vna vk vl gna gk gl c I v
r = an(v)*(1-n) - bn(v)*n;
Note that in these programs, the membrane voltage (the Matlab variable ``v'') is treated as a global variable. This is because the functions that fzero uses need to be functions of a single, scalar variable. After downloading these programs, type ``nullclines'' at the Matlab prompt. You might get a few messages saying that Matlab was unable to find some zeros - this is OK. One obtains the following figure:
Note that the point at which the nullclines intersect corresponds to a ``fixed point'' of equations (2) and (3). At this point, .
Using the functions rhsV.m and rhsn.m, verify that