A confusion about the method of undetermined coef. and what is the motivation behind the given vector space?

65 Views Asked by At

In the book of Differential Equations: A Dynamical Systems Approach: Ordinary Differential Equations by West&Hubbard, at page 73, it is states for the Method of Undetermined coefficients that

Consider the differential equation $x' = p(t)x + q(t)$. Suppose that $q(t)$ is an element of a finite-dimensional vector space $V$ of junctions $f$, closed under the operation

$$f \to f' - p(t) f$$ [...]

I'm really confused about the definiton of $V$ because, for example, the operation that is stated is a unary operation (I see only one input), so how can we define a vector space with a unary operation ? Moreover, I coulnd't understand how exactly show we determined when a given function $g$ is in $V$ or not. I mean as far as I understood, to check whether $g\in V$ or not, I need to check that $g' - p(t)g$ (where $p(t)$ is given in the ODE) is in $V$, but by authors' definition, to check whether $g' - p(t)g \in V$, I need to check whether $[g' - p(t)g]' - p(t)[g' - p(t)g] \in V$ or not..., so I really found this confusing and I'm not even sure that I understood the author correctly.

Therefore, I'm looking for some clarifications for this proposition.

Plus, what is the motivation behind this vector space ? I mean I couldn't see why do need "the guesses solutions" to a first order linear DEs to be in this space ?

Edit:

I have put a link of the book.

Edit 2:

Note that my main question is about the structure of $V$, and not about it's existance or something, so you can assume $V$ exists for the purpuse of the question.

1

There are 1 best solutions below

7
On

The "unary operation", as you called it, is not going to be one of the operations of the vector space. More precisely, you have to find a vector space $(V, +, \cdot)$ such that $$ f\in V \quad \Rightarrow\quad Df \in V, $$ where $Df:=f' -p(x)f.$ (Verifying that this is indeed the case amounts exactly to the task that you "find confusing", unfortunately).

Let me make a toy model. Redefine $D$ as $Df:=f'$. The vector space $V$ of real polynomials is closed$^{[1]}$ under the operation $D$, because if $f$ is a polynomial, its derivative is again a polynomial. Notice that this space is infinite dimensional. If $n\in\mathbb N$, the vector space of polynomials of degree up to $n$ is finite dimensional and closed under $D$.

In general, constructing such vector spaces is difficult. The framework is called spectral theory.


[1] This choice of terminology is unfortunate, according to me. I would have said that $V$ is invariant under $D$.