0
$\begingroup$

What are the conditions under which the formula I've given below for an inhomogeneous reducible 2nd‐order ODE is valid?

I have the second‐order inhomogeneous ODE $$ y''(t) \;+\; h(t)\,y'(t)\;+\;q(t)\,y(t)\;=\;f(t), $$ together with one known solution $\phi_{1}(t)\\$ of its corresponding homogeneous equation $$ \phi_{1}''(t) \;+\; h(t)\,\phi_{1}'(t)\;+\; q(t)\,\phi_{1}(t)\;=\;0. $$ From this, I derived the following “general” solution by just doubly integrating the reduced inhomogeneous ODE: $$ y(t) \;=\;\phi_{1}(t)\;\int^{\,t} \exp\!\biggl(-\int^{\,\sigma}\Bigl[\tfrac{2\,\phi_{1}'(s)}{\phi_{1}(s)} \;+\;h(s)\Bigr]\,ds\biggr) \,\Biggl(\,\int^{\,\sigma} \frac{f(\theta)}{\phi_{1}(\theta)}\,\exp\!\Bigl(\,\int^{\,\theta}\bigl[\tfrac{2\,\phi_{1}'(u)}{\phi_{1}(u)} + h(u)\bigr]\,du\Bigr)d\theta\Biggr)\,d\sigma. $$

When I apply this formula to the specific example $y''(t)\;-\;\frac{t+2}{t}\,y'(t)\;+\;\frac{t+2}{t^{2}}\,y(t)\;=\;t,\quad t>0,\\$

where a known homogeneous solution is $\phi_{1}(t)=t\\$, I obtained the general solution (matching the solution from the standard method of reduction of order + variation of parameters).

However, my professor said that this integral formula is not universally valid and that it works here only because certain “required conditions” happen to be satisfied (I'm assuming in addition to the usual conditions of the functions being well behaved(continuous, integrable, etc.) that we use for the standard method as well)

Question: What exactly are the conditions under which the above double‐integral expression yields the full general solution? and why does one usually revert to the “standard” method of reduction of order followed by variation of parameters?

$\endgroup$

1 Answer 1

0
$\begingroup$

We start with the second-order inhomogeneous ODE:

$$ y''(t) + h(t)y'(t) + q(t)y(t) = f(t), $$

and assume that $\phi_1(t)$ is a solution of the corresponding homogeneous equation:

$$ \phi_1''(t) + h(t)\phi_1'(t) + q(t)\phi_1(t) = 0. $$

Let $\phi_2(t) = v(t)\phi_1(t)$ be the second solution of the homogeneous equation, where $v(t)$ is to be determined. Substituting $\phi_2(t)$ into the homogeneous equation gives:

$$ \frac{v''}{v'} + \left[2\frac{\phi_1'(t)}{\phi_1(t)} + h(t)\right] = 0. $$

Now, for the inhomogeneous equation, we use variation of parameters. Let $v_1(t)$ and $v_2(t)$ be functions such that:

$$ v_2'(t) = \frac{f(t)}{W(t)}\phi_1(t), \quad v_1'(t) = -\frac{f(t)}{W(t)}\phi_2(t), $$

where $W(t)$ is the Wronskian of $\phi_1(t)$ and $\phi_2(t)$. From the relation $v_1'(t) = -v(t)v_2'(t)$, we find:

$$ v_1(t) = \int -v(t)v_2'(t) \, dt. $$

The particular solution of the inhomogeneous equation is then given by:

$$ y_p(t) = v_1(t)\phi_1(t) + v_2(t)\phi_2(t). $$

Substituting $v_1(t)$ and $v_2(t)$, we write:

$$ y_p(t) = v_2(t)\phi_2(t) - \phi_1(t)\int v(t)v_2'(t) \, dt. $$

Using $\phi_2(t) = v(t)\phi_1(t)$, this simplifies to:

$$ y_p(t) = \phi_1(t)\int v'(t)v_2(t) \, dt. $$

Next, substitute $v_2(t) = \int \frac{f(\sigma)}{W(\sigma)}\phi_1(\sigma) \, d\sigma$ into the expression for $y_p(t)$:

$$ y_p(t) = \phi_1(t) \int v'(t) \int \frac{f(\sigma)}{W(\sigma)}\phi_1(\sigma) \, d\sigma \, dt. $$

Using the relation $W(\sigma)/\phi_1(\sigma) = v'(\sigma)\phi_1(\sigma)$, we further simplify:

$$ y_p(t) = \phi_1(t) \int v'(t) \int \frac{f(\sigma)}{v'(\sigma)\phi_1(\sigma)} \, d\sigma \, dt. $$

Rewriting the nested integrals using $\theta$ and $\sigma$ for clarity, the particular solution becomes:

$$ y_p(t) = \phi_1(t) \int_t \exp\left(-\int_\sigma \left[2\frac{\phi_1'(s)}{\phi_1(s)} + h(s)\right] ds\right) \left(\int_\sigma \frac{f(\theta)}{\phi_1(\theta)}\exp\left(\int_\theta \left[2\frac{\phi_1'(u)}{\phi_1(u)} + h(u)\right] du\right) d\theta\right) d\sigma. $$

This matches the desired form. The solution is dependent only on $\phi_1(t)$, $f(t)$, and the coefficients of the differential equation, expressed as a double-integral where $\phi_1(t)$ and $f(t)$ are explicitly involved.

In deriving the solution, we assumed that $\phi_1(t)$ is a well-behaved solution to the homogeneous equation, satisfying $\phi_1(t) \neq 0$ on the domain of interest. This ensures that the Wronskian $W(t)$ is non-zero, allowing us to construct $\phi_2(t)$ as $v(t)\phi_1(t)$ and proceed with variation of parameters. Additionally, we assumed that the coefficients $h(t)$, $q(t)$, and the inhomogeneous term $f(t)$ are continuous, which guarantees the applicability of the method.

The use of $W(t)/\phi_1(t) = v'(t)\phi_1(t)$ is valid as it follows directly from the definition of the Wronskian and the form $\phi_2 = v\phi_1$. Furthermore, the final nested integral form relies on the convergence of these integrals, which holds under the assumption that $f(t)$ and $\phi_1(t)$ are sufficiently smooth and well-behaved.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.