Necessary and Sufficient Condition for Quadratic Functional to be Positive Definite/Dependent on N Functions
This article needs to be linked to other articles. You can help $\mathsf{Pr} \infty \mathsf{fWiki}$ by adding these links. To discuss this page in more detail, feel free to use the talk page. When this work has been completed, you may remove this instance of {{MissingLinks}} from the code. |
Theorem
Let $K$ be a (real) functional, such that:
- $\ds K \sqbrk {\mathbf h} = \int_a^b \paren {\mathbf h' \mathbf P \mathbf h' + \mathbf h \mathbf Q \mathbf h} \rd x$
where:
- $\mathbf h$ is an $N$-dimensional vector
- $\mathbf Q$ is a $N \times N$ matrix
- $\mathbf P$ is a $N\times N$ symmetric positive definite matrix.
Let $\closedint a b$ be such that it does not contain a point conjugate to $a$.
Then:
- $\forall \mathbf h: \map {\mathbf h} a = \map {\mathbf h} b = 0: K \sqbrk {\mathbf h} > 0$
if and only if the above holds.
This article, or a section of it, needs explaining. In particular: "the above" -- reference the specific statement, there is plenty of "above" and it may not be immediately clear exactly what You can help $\mathsf{Pr} \infty \mathsf{fWiki}$ by explaining it. To discuss this page in more detail, feel free to use the talk page. When this work has been completed, you may remove this instance of {{Explain}} from the code. |
Proof
Necessary Condition
Let $\mathbf W$ be an arbitrary differentiable symmetric matrix.
Then
\(\ds 0\) | \(=\) | \(\ds \int_a^b \map {\frac \d {\d x} } {\mathbf h \mathbf W \mathbf h} \rd x\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \int_a^b \mathbf h \mathbf W' \mathbf h \rd x + 2 \int_a^b \mathbf h' \mathbf W \mathbf h \rd x\) |
Suppose, $\mathbf W$ is such that:
- $\mathbf Q + \mathbf W' = \mathbf W \mathbf P^{-1} \mathbf W$
Then:
\(\ds \int_a^b \paren {\mathbf h' \mathbf P \mathbf h' + \mathbf h \mathbf Q \mathbf h} \rd x\) | \(=\) | \(\ds \int_a^b \paren {\mathbf h' \mathbf P \mathbf h' + 2 \mathbf h' \mathbf W \mathbf h + \mathbf h \mathbf Q \mathbf h + \mathbf h \mathbf W' \mathbf h} \rd x\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \int_a^b \paren {\mathbf h' \mathbf P \mathbf h' + 2 \mathbf h' \mathbf W \mathbf h + \mathbf h \mathbf W \mathbf P^{-1} \mathbf W \mathbf h} \rd x\) | ||||||||||||
\(\ds \) | \(=\) | \(\ds \int_a^b \paren {\mathbf P^{1/2} \mathbf h' + \mathbf P^{1/2} \mathbf h' \mathbf P^{-1/2} \mathbf W \mathbf h + \mathbf P^{-1/2} \mathbf W \mathbf h}^2 \rd x\) |
This article, or a section of it, needs explaining. In particular: Clarify the below -- what is being used to show what? You can help $\mathsf{Pr} \infty \mathsf{fWiki}$ by explaining it. To discuss this page in more detail, feel free to use the talk page. When this work has been completed, you may remove this instance of {{Explain}} from the code. |
Note that:
- $\mathbf P^{1/2} \mathbf h' + \mathbf P^{-1/2} \mathbf W \mathbf h \ne 0$
unless:
- $\forall x\in\closedint a b: \map {\mathbf h} x = 0$
However, this contradicts the absence of conjugate points.
Hence:
- $K > 0$
$\Box$
Sufficient Condition
Consider the following functional:
- $\ds \int_a^b \paren {K t + \mathbf h' \paren {1 - t} \mathbf h'} \rd x$
The corresponding Euler's equations are:
- $-\map {\dfrac \d {\d x} } {t \mathbf P \mathbf h' + \paren {1 - t} \mathbf h'} + t \mathbf Q \mathbf h = 0$
Suppose the interval $\closedint a b$ contains a point $\tilde a$ conjugate to $a$.
Then the determinant $\size {h_{ij} }$ vanishes.
Therefore there exists a linear combination of $h_i$ not identically equal to zero such that $\map {\mathbf h} {\tilde a} = 0$.
Furthermore, since the Euler's equations are continuous with respect to $t$, so is the solution of this equation.
Aiming for a contradiction, suppose $\tilde a = b$.
By lemma, $K$ vanishes.
This contradicts the positive definiteness of $K$.
Therefore, $\tilde a \ne b$.
Thus, for $t = 1$ the conjugate point may only reside in $\openint a b$.
This needs considerable tedious hard slog to complete it. In particular: to be continued To discuss this page in more detail, feel free to use the talk page. When this work has been completed, you may remove this instance of {{Finish}} from the code.If you would welcome a second opinion as to whether your work is correct, add a call to {{Proofread}} the page. |
Sources
- 1963: I.M. Gelfand and S.V. Fomin: Calculus of Variations ... (previous) ... (next): $\S 5.29$: Generalization to n Unknown Functions