# Linearization of Second Order Differential Equations

1. Jan 2, 2013

### Ruby Tyra

I'm having some difficulties figuring out how to linearize second order differential equations for a double pendulum.

I have an equation that is in the form of

$\theta_{1}''$$\normalsize = function$ [$\theta_{1}$,$\theta_{2}$,$\theta_{1}'$,$\theta_{2}'$]

(The original equation is found at http://www.myphysicslab.com/dbl_pendulum.html, the equations inside the orange rectangle.)

I was told to replace that function by a linear function of all four variables but I don't know where to start with that since the original equation is much more complex than the simple pendulum example we were given.

Thank you!

2. Jan 2, 2013

### Staff: Mentor

If f = f(x,y,z,w) and you want to linearize, use the chain rule:

$$f(x,y,z,w) = f(x_0,y_0,z_0,w_0)+\frac{\partial{f}}{\partial{x}}(x-x_0)+\frac{\partial{f}}{\partial{y}}(y-y_0)+\frac{\partial{f}}{\partial{z}}(z-z_0)+\frac{\partial{f}}{\partial{w}}(w-w_0)$$

where the partials are evaluated at $$x_0,y_0,z_0,w_0$$

3. Jan 3, 2013

### HallsofIvy

Staff Emeritus
Note that what Chestermiller is saying is essentially the same as replacing the function by a Taylor polynomial in all variables, then dropping all but the linear terms. And that, in turn, is the same as replacing the "surface" by its "tangent plane".

Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook