Reduction of order
ode internal name "reduction_of_order"
This is second order ode where on solution is known. The second solution is found using
reduction of order.
Example 1
Solve
\[ y^{\prime \prime }+p\left ( x\right ) y^{\prime }+q\left ( x\right ) y=0 \]
Given that one solution is known to be \(y_{1}\). We start by assuming the second solution is
\(y_{2}=y_{1}u\left ( x\right ) \) where \(u\left ( x\right ) \) is to be determined. Hence
\begin{align*} y_{2}^{\prime } & =y_{1}^{\prime }u+y_{1}u^{\prime }\\ y_{2}^{\prime \prime } & =y_{1}^{\prime \prime }u+y_{1}^{\prime }u^{\prime }+y_{1}^{\prime }u^{\prime }+y_{1}u^{\prime \prime }\\ & =y_{1}^{\prime \prime }u+2y_{1}^{\prime }u^{\prime }+y_{1}u^{\prime \prime }\end{align*}
Substituting in the given ODE gives (since \(y_{2}\) is a solution, then it also satisfies the ode)
\[ \left ( y_{1}^{\prime \prime }u+2y_{1}^{\prime }u^{\prime }+y_{1}u^{\prime \prime }\right ) +p\left ( y_{1}^{\prime }u+y_{1}u^{\prime }\right ) +qy_{1}u=0 \]
And
now we collect on \(u\) and all its derivatives. The above becomes
\[ u\left ( y_{1}^{\prime \prime }+py_{1}^{\prime }+qy_{1}\right ) +u^{\prime }\left ( 2y_{1}^{\prime }+py_{1}\right ) +y_{1}u^{\prime \prime }=0 \]
But \(y_{1}^{\prime \prime }+py_{1}^{\prime }+qy_{1}=0\). The above becomes
\[ u^{\prime }\left ( 2y_{1}^{\prime }+py_{1}\right ) +y_{1}u^{\prime \prime }=0 \]
Ok,
you migth ask, what did we accomplish in all of this? Since we eneded up with just
another second order ode. But here is the main point of this method. This new ode is
missing the \(u\) term. Therefore by letting \(u^{\prime }=v\) we can make the above ode become first
order ode
\[ v\left ( 2y_{1}^{\prime }+py_{1}\right ) +y_{1}v^{\prime }=0 \]
SInce \(y_{1}\) is given, the above first order ode is now solved for \(v\), and once \(v\)
is known, then \(u\) is found by integrating \(u^{\prime }=v\) and once \(u\) is found then \(y_{2}\) is found from
\(y_{2}=y_{1}u\left ( x\right ) \).
The above ode can be written as
\[ v^{\prime }+\left ( 2\frac {y_{1}^{\prime }}{y_{1}}+p\right ) v=0 \]
Hence it is linear first order ode. The integrating factor is
\begin{align*} \mu & =e^{\int 2\frac {y_{1}^{\prime }}{y_{1}}+pdx}\\ & =e^{\int \frac {2}{y_{1}}\frac {dy_{1}}{dx}dx+\int pdx}\\ & =e^{\int \frac {2}{y_{1}}dy_{1}+\int pdx}\\ & =e^{2\ln y_{1}+\int pdx}\\ & =e^{2\ln y_{1}}e^{\int pdx}\\ & =y_{1}^{2}e^{\int pdx}\end{align*}
Therefore
\begin{align} d\left ( v\mu \right ) & =0\nonumber \\ v\mu & =c_{1}\nonumber \\ v & =c_{1}\frac {e^{-\int pdx}}{y_{1}^{2}} \tag {1}\end{align}
Since \(u^{\prime }=v\) then we have
\[ \frac {du}{dx}=v \]
Integrating
\[ u=\int vdx+c_{2}\]
Here we are free to let \(c_{2}=0\). Therefore
\begin{equation} u=\int vdx \tag {2}\end{equation}
Therefore
\begin{align} y_{2} & =y_{1}u\nonumber \\ & =y_{1}\int vdx\nonumber \\ & =y_{1}\int \left ( c_{1}\frac {e^{-\int pdx}}{y_{1}^{2}}\right ) dx\nonumber \\ & =c_{1}y_{1}\int \left ( \frac {e^{-\int pdx}}{y_{1}^{2}}\right ) dx \tag {3}\end{align}
And the solution is
\[ y=c_{1}y_{2}+c_{2}y_{1}\]
The following example shows how the above can be applied to a
concrete problem.
Example 2
Solve
\begin{align*} x^{2}y^{\prime \prime }+xy^{\prime }-9y & =0\\ y_{1} & =x^{3}\end{align*}
Putting the ode in normal form, it becomes
\[ y^{\prime \prime }+\frac {1}{x}y^{\prime }-\frac {9}{x^{2}}y=0 \]
Hence \(p=\frac {1}{x},q=-\frac {9}{x^{2}}\). Using EQ (1)
\begin{align*} v & =c_{1}\frac {e^{-\int pdx}}{y_{1}^{2}}\\ & =c_{1}\frac {e^{-\int \frac {1}{x}dx}}{x^{6}}\\ & =\frac {c_{1}}{x^{6}}e^{-\ln x}\\ & =c_{1}\frac {1}{x^{7}}\end{align*}
EQ (2) becomes
\begin{align*} u & =\int vdx\\ & =\int c_{1}x^{-7}dx\\ & =c_{1}\frac {x^{-6}}{-6}\\ & =c_{1}x^{-6}\end{align*}
(last step above just rewrites the constant). Hence the second solution is
\begin{align*} y_{2} & =y_{1}u\\ & =x^{3}\left ( c_{1}x^{-6}\right ) \\ & =c_{1}x^{-3}\end{align*}
Therefore the solution is
\begin{align*} y & =c_{3}y_{2}+c_{4}y_{1}\\ & =c_{1}\frac {1}{x^{3}}+c_{2}x^{3}\end{align*}
Where in last step above, constants were merged and renamed.