Numerically solve a 3rd order differential equation with 3 unknowns
5 次查看(过去 30 天)
显示 更早的评论
Hi, I have a 3rd order differential equation in y, with 3 unknowns a, b, c. To simplify imagine the equation is:
diff(y, t, 3) + a.*diff(y, t, 2) + b.*diff(y, t) + y.^(3/2) + c.*y == 0;
I know the discrete values of y and the time vector. I want to know tha values of a, b, c that solve this equation. Clearly there will be more combination of this values, but is there a way to solve it? (like with ode function or others).
One idea would be to express all the derivatives as function of y (like using finite differences) and then solve, is there a way to do it automatically on matlab?
采纳的回答
Torsten
2023-6-1
编辑:Torsten
2023-6-1
Use "gradient" three times to get approximations for y', y'' and y''' in the points of your t-vector.
Let diff_y, diff_y2, diff_y3 be these approximations as column vectors of length n.
Now form a matrix A as
A = [diff_y2,diff_y,y]
and a column vector v as
v = [-y.^(3/2) - diff_y3]
Then approximations for a, b and c can be obtained using
sol = A\v
with
a = sol(1)
b = sol(2)
c = sol(3)
12 个评论
Torsten
2023-6-3
For the lsqnonlin fitting procedure, you can try two ways:
The first is without using the ode integrator and generating y',y'' and y''' from your data by using the "gradient" function.
The second is with the ode integrator by integrating the differential equation for given parameters a, b and c and by comparing the result with your given data for y.
更多回答(0 个)
另请参阅
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!