二阶线性齐次ODE降阶法原理疑问：为何该方法可行？

Great question—this is a common point of confusion when first learning reduction of order for linear ODEs. Let's break down why this method works, and address why it doesn't conflict with what you know about linearity.

Why we use ( y_2(t) = v(t)y_1(t) ) #

First, recall the general form of a second-order linear homogeneous ODE:

y'' + p(t)y' + q(t)y = 0

We know ( y_1(t) ) is a solution, so it satisfies ( y_1'' + p(t)y_1' + q(t)y_1 = 0 ). The key idea here isn't to apply the superposition principle directly (which deals with constant multiples of solutions), but to construct a second linearly independent solution—since the solution space for this ODE is 2-dimensional, we just need one solution that isn't a constant multiple of ( y_1(t) ).

The magic happens when we substitute into the ODE #

Let's compute the derivatives of ( y_2 = v(t)y_1(t) ):

• First derivative: ( y_2' = v'y_1 + vy_1' )
• Second derivative: ( y_2'' = v''y_1 + 2v'y_1' + vy_1'' )

Substitute these into the original ODE:

(v''y_1 + 2v'y_1' + vy_1'') + p(t)(v'y_1 + vy_1') + q(t)(vy_1) = 0

Now group terms by the order of the derivative of ( v ):

v''y_1 + v'(2y_1' + p(t)y_1) + v(y_1'' + p(t)y_1' + q(t)y_1) = 0

Here's where linearity (and the fact that ( y_1 ) is a solution) comes in: the term with ( v ) is exactly ( v \times 0 = 0 ), because ( y_1 ) satisfies the ODE. That simplifies our equation drastically to:

v''y_1 + v'(2y_1' + p(t)y_1) = 0

Reduce to a first-order ODE #

Let ( w = v' ), so ( w' = v'' ). Now the equation becomes a first-order linear ODE in ( w ):

w'y_1 + w(2y_1' + p(t)y_1) = 0

This is separable. Rearrange terms to separate variables:

dw/w = - (2y_1'/y_1 + p(t)) dt

Integrate both sides to solve for ( w ), then integrate again to find ( v(t) ). The resulting ( v(t) ) will not be a constant (if it were, ( y_2 ) would just be a multiple of ( y_1 ), which is linearly dependent), so ( y_2 = v(t)y_1(t) ) is our second linearly independent solution.

Why this doesn't contradict linearity #

The superposition principle tells us that constant multiples and sums of solutions are also solutions—but this method isn't using superposition to create a solution. Instead, we're using the structure of the linear ODE to guess a form for a second solution and then solve for the unknown function ( v(t) ). Linearity helps simplify the equation (by eliminating the term with ( v )), but we're not relying on it to directly generate the solution.

In short: reduction of order works because substituting ( y_2 = v(t)y_1(t) ) leverages the linearity of the ODE to reduce the problem to a solvable first-order equation, not because we're applying superposition directly.

内容的提问来源于stack exchange，提问作者jeanluc