problem with solving system of differential equations

Hello everyone,

I've been trying to solve a system of second order differential equations for a few days, but I cant solve them using any functions in matlab. I tried using solve , dsolve , fsolve, vpasolve, bvp4c (this one gives a wrong answer) ...

clc
clear all

x1 = 2;
y1 = 1;
r1 = 1;

m1 = 0.5;


% equations
syms x(t) y(t) ;

p1 = 0.65;
eqn1 = diff(x,2) - p1*(sqrt(m1 + r1^2 + (y-y1)^2))  == 0;
eqn2 = diff(y,2) - p1*(sqrt(m1 + r1^2 + (x-x1)^2))  == 0;

conds = [y(0)==0, x(0)==0, y(2)==5, x(2)==5];

eqns = [eqn1, eqn2];

sol = dsolve(eqns, conds)

I'm new to matlab and I searched a lot to find an answer but none worked on my problem. Any help would be appreciated.

Thank you in advance.

Answer 1

[Alan Weiss]

If you are willing to have a numerical solution, rather than a symbolic solution, then using fsolve along with ode45 can work. The technique is to solve the ODE from an initial guess using ode45, then change the initial guess using fsolve to satisfy the boundary conditions. Here is some code I knocked together. I assume that the ODE is given in the form [x xdot y ydot]. The initial conditions are x(0) = 0, y(0) = 0. The final conditions are x(2) = 5, y(2) = 5.

myx = fsolve(@myval,[1;1]) % Obtain the solution

Equation solved.

fsolve completed because the vector of function values is near zero
as measured by the value of the function tolerance, and
the problem appears regular as measured by the gradient.

myx = 2×1

1.4648 1.3795

% Examine the end point
y0 = [0;myx(1);0;myx(2)]; % initial poiint for equation
sol = ode45(@odefn,[0 2],y0);
yend = deval(sol,2)

yend = 4×1

5.0000 4.1178 5.0000 3.7295

function dydt = odefn(~,y)
% y(t) = [x dx y dy]
x1 = 2;
y1 = 1;
r1 = 1;
m1 = 0.5;
p1 = 0.65;
dydt = zeros(4,1); % allocate y
dydt(1) = y(2);
dydt(3) = y(4);
dydt(2) = p1*(sqrt(m1 + r1^2 + (y(3)-y1)^2));
dydt(4) = p1*(sqrt(m1 + r1^2 + (y(1)-x1)^2));
end

function mout = myval(x)
y0 = [0;x(1);0;x(2)]; % initial poiint for equation
sol = ode45(@odefn,[0 2],y0);
yend = deval(sol,2);
mout = [yend(1) - 5;yend(3) - 5];
end

Alan Weiss

MATLAB mathematical toolbox documentation

[Anil]

does that ode45 inside the myval, helps to find the sationary solution of nonlinear equations through fsolve? Could you please comment on that. Thanks.

[Alan Weiss]

The code works by taking a numerical solution of the ODE starting from [0,0] and evaluating it at time 2. It tries to make the values at time 2 equal to [5,5]; that is why in myval(x) I subtract [5,5] from the final point, so that fsolve can try to make the function equal to [0,0].

Alan Weiss

MATLAB mathematical toolbox documentation

Answer 2

[Friedel Hartmann]

I think your problem are the second derivatives. Matlab can only solve differential equations where the leading term(s) are first order..

Check the literature or the documentation on how you can transform your two equations into a system of four differential equations which only contain first derivatives.

Answer 3

[Friedel Hartmann]

% Example: solve -u''(x) = sin(x)

syms up(x) u(x)

eqns = [diff(up,x) == -sin(x), diff(u,x) - up == 0]; % -up' = sin(x) u' = up

S = dsolve(eqns,up(1) == 0,u(0) == 0)

[ali hamidi]

Thank you for your answer, but unfortunately doing this did not change anything and i still can't get an answer. here's my code trying this :

clc
clear all
x1 = 2;
y1 = 1;
r1 = 1;
m1 = 0.5;
p1 = 0.65;
% equations
syms x(t) y(t) xt(t) yt(t);

% xt' = x"     yt' = y"
eqn1 = diff(xt,t) - p1*(sqrt(m1 + r1^2 + (y-y1)^2))  == 0;
eqn2 = diff(x,t) == xt;
eqn3 = diff(yt,t) - p1*(sqrt(m1 + r1^2 + (x-x1)^2))  == 0;
eqn4 = diff(y,t) == yt;
conds = [y(0)==0, x(0)==0, y(2)==5, x(2)==5];
eqns = [eqn1, eqn2, eqn3, eqn4];
sol = dsolve(eqns, conds)

%V = odeToVectorField(eqns)