Solving an ODE for a Complex Variable in Polar Coordinates

10 Ansichten (letzte 30 Tage)
I want to solve an ODE where the variable of interest is a complex number in terms of polar coordinates, . I have been using the example here where they have chosen . I roughly know what the dynamics of the system should look like (it is a well-known reduction of the Kuramoto model) and am almost certain that it should not have features that cause ode45() to have any problems. However, the systems seems to blow up. I believe it is because of how I have set up definitions or returned values in the Second Function File, but I am not sure. Does anyone have any suggestions? Thanks!
Second Function File and Call:
% Set up ICs and t, then solve
zv0 = [0.8; 3];
tspan = [0 100];
[t, zv] = ode45(@imaginaryODE, tspan, zv0);
% Plot
plot(t,zv(:,1))
function fv = imaginaryODE(t, zv)
% Construct z from argument and angle
z = abs(zv(1)).*exp(1i.*angle(zv(2)));
% Evaluate for the function defined in complexf.m
zp = complexf(t, z);
% Return argument and angle
fv = [abs(zp); angle(zp)];
end
First Function File:
function f = complexf(t, z)
% Constants
omega0 = 2.*pi/24;
gamma = 1/100;
k = 1;
% Function
f = (1i.*omega0 - gamma).*z + (k/2).*(z - (z^2).*conj(z));
end
  3 Kommentare
Cameron3332
Cameron3332 am 18 Mär. 2020
Hey Darova!
The original equation in is:
for constants:
,
,
for z in the polar representation:
.
If you have any other questions, feel free to ask. Thank you so much :)
darova
darova am 18 Mär. 2020
You code looks OK. I changed this
tspan = [0 2.45];
result

Melden Sie sich an, um zu kommentieren.

Akzeptierte Antwort

David Goodmanson
David Goodmanson am 19 Mär. 2020
Hi Cameron,
to solve the differential equation, ode needs real inputs so you could either use z = x+iy in the usual way or use
z = r exp(i psi) % fwd
r = abs(z) psi = angle(z) % reverse
The problem is with the usual 2pi ambiguity for angles, and Matlab uses -pi < psi <= pi. Angle(z) is discontinuous, which is bad for use in a differential equation.
The following code uses the same x and y approach as in the link you provided (although I changed a bunch of names). For this problem you get nice oscillatory behavior. If you want r and psi, you can always get them later, as in plots 2 and 3. Plot 2 illustrates the problem with angle, and unwrap in plot 3 takes care of it.
z0 = .8+3i;
xy0 = [real(z0); imag(z0)];
tspan = [0 100];
[t,xy] = ode45(@odefun, tspan, xy0);
figure(1)
x = xy(:,1);
y = xy(:,2);
plot(t,x,t,y); grid on
z = x+i*y;
r = abs(z);
psi = angle(z);
figure(2)
plot(t,r,t,psi); grid on
figure(3)
plot(t,10*r,t,unwrap(psi)); grid on
function Dxy = odefun(t,xy)
% usual ode function, [x;y] in; [dx/dt; dy/dt] out
z = xy(1) + i*xy(2);
g = fun(t,z);
Dxy = [real(g); imag(g)];
end
function g = fun(t,z)
% Define function that takes and returns complex values
omega0 = 2.*pi/24;
gamma = 1/100;
k = 1;
g = (i*omega0 - gamma)*z + (k/2)*(z - (z.^2).*conj(z));
end

Weitere Antworten (0)

Kategorien

Mehr zu Programming finden Sie in Help Center und File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by