3
$\begingroup$

In Conway's A Course in Functional Analysis, the author stated the following Proposition $1.4$ at page $377.$

Let $X$ be a vector space over $\mathbb{R}.$ Given $f,f_1,f_2,...,f_n$ are linear functionals on $X.$ If $$\bigcap_{j=1}^n ker(f_j) \subseteq ker(f),$$ then there exist scalars $a_1,...,a_n$ such that $$f(x) = \sum_{j=1}^n a_j f_j.$$

The proof starts as follows:

It may be assumed that without loss of generality that for $1\leq k\leq n,$ $$\bigcap_{j\neq k}ker(f_j) \neq \bigcap_{j=1}^n ker(f_j).$$

Question: Why can we assume above? In general, we have $$\bigcap_{j\neq k} ker(f_j) \supseteq \bigcap_{j=1}^n ker(f_j).$$ So in the proof, we can safely assume that the inclusion is strict. If the inclusion is not strict, meaning the two intersections are equal, then kernel of $f_j$ is a subset of kernel of $f_i$ for some $i.$

$\endgroup$
1
  • 1
    $\begingroup$ If there is an equality, you can just forget about $f_k$ (because the assumption still holds for $f,f_1,\dots,f_{k-1},f_{k+1},\dots,f_n$) and obtain a linear combination where $f_k$ does not appear. $\endgroup$ Jan 25, 2018 at 9:48

1 Answer 1

3
$\begingroup$

I am sure you know the result which says if $f,g$ are linear functional with $ker(g)\subseteq ker(f)$ then $f$ is a constant multiple of $g$.

Suppose $f,g,h$ are such that $ker(g)\cap ker(h)\subset ker(f)$. Suppose we have $ker(g)\cap ker(h)=ker(g)$ then $ker(g)\subset ker(f)$ then by above fact we have $f=a g=ag+0h$. Similarly in the case when $ker(g)\cap ker(h)=ker(h)$.

You can generalise this for more than$3$ functions.

$\endgroup$
1
  • $\begingroup$ And if none of the suppositions is true?? How was this answer upvoted and accepted?! It needs at least more explanation... $\endgroup$
    – DonAntonio
    Mar 9, 2020 at 11:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .