If V is a finite dimensional vector space over the field F with dual space V* = Hom(V,F) . How to prove every ordered basis for V* is the dual basis for some basis for V ?
-
Definitions are your friends. What are the elements of the dual space? What makes a subset of $V^$ a dual basis* to some basis of $V$? – hardmath May 05 '16 at 12:48
2 Answers
Hint: given a basis $(\xi_1,\dots,\xi_n)$ of $V^*$, it has a dual basis in $V^{**}$; use the canonical isomorphism $V\to V^{**}$.
For any vector space $V$, define the map $\omega_V\colon V\to V^{**}$ by $\omega_V(x)=\hat{x}$ such that, for $\xi\in V^*$, $$ \hat{x}(\xi)=\xi(x) $$ is a well-defined linear map (verify). If $V$ is finite dimensional and $x\ne0$, there exists $\xi\in V^*$ such that $\hat{x}(\xi)=\xi(x)\ne0$ (just complete $x$ to a basis). Therefore $\omega_V$ is injective.
The existence of a dual basis entails that $\dim V=\dim V^*$ and therefore also $\dim V^*=\dim V^{**}$. Hence $\omega_V$ is an isomorphism.
In particular, if $\{\xi_1,\dots,\xi_n\}$ is a basis of $V^*$, there exists its dual basis in $V^{**}$ and we can assume it is of the form $\{\hat{x}_1,\dots,\hat{x}_n\}$ for some $x_1,\dots,x_n\in V$, because $\omega_V$ is an isomorphism.
Since we have chosen a dual basis, we have, by definition, $$ \hat{x}_i(\xi_j)=\delta_{ij} $$ (Kronecker delta), which amounts to saying that $$ \xi_j(x_i)=\delta_{ji} $$ and so $\{\xi_1,\dots,\xi_n\}$ is the dual basis of $\{x_1,\dots,x_n\}$.
- 244,946
-
-
-
-
@RipanJeet That for a finite dimensional vector space $V$ the map $x\mapsto \hat{x}$ where, for $\xi\in V^$, $\hat{x}(\xi)=\xi(x)$ is an isomorphism $V\to V^{*}$. – egreg May 05 '16 at 20:10
-
-
-
-
@RipanJeet This is basic theory of vector spaces: any linearly independent set can be extended to a basis. – egreg May 05 '16 at 21:31
-
-
-
@egreg In the last line where you wrote :$\hat x_i(\xi_j)=\delta_{ij}$", I have a question: By definition $\hat x_i(\xi_j)=\xi_j(x_i).$ But, $\xi_j(x_i)$ will only be equal to $\delta_{ij}$ if, $\xi_j$ is the jth coordinate function of $V$ wrt the basis ${x_1,x_2,...,x_n}$ which we are considering. – Thomas Finley Feb 04 '24 at 04:25
-
@ThomasFinley ${\hat{x}1,\dots,\hat{x}_n}$ is the dual basis of ${\xi_1,\dots,\xi_n}$ by assumption. Thus $\xi_j(x_i)=\delta{ji}$ because $\hat{x}_j(\xi_i)=0$. – egreg Feb 04 '24 at 08:49
Let $n$ be the dimension of $V$, and let $\{b^1,\ldots,b^n\}$ be an ordered basis for $V^*$. Let $\{e_1,\ldots,e_n\}$ be some basis for $V$, and let $\{e^1,\ldots,e^n\}$ be the dual basis for $V^*$, so $e^i(e_j)=\delta^i_j$. Then there exist coefficients $\{a_{ij}\}_{1\le i,j\le n}$ in the field over which the vector space is taken, such that $b^i=a_{ij}e^j$ for all $1\le i\le n$ (summation convention in force). Then $b^i(e_k)=a_{ij}e^j(e_k)=a_{ij}\delta^j_k=a_{ik}$. Since, for all $1\le i\le n$, there exists at least one $1\le k\le n$ such that $a_{ik}\neq0$, and this gives you the required basis for which $\{b^1,\ldots,b^n\}$ is the dual basis.
- 1,882
-
If {b^1, . . . ,b^n } is the required dial basis then shouldn't it be equal to the kronecker's delta when applied .. here its giving aik's which could be particular constants . – Shona May 05 '16 at 16:21
-
I didn't finish it; my comment about one of the coefficients being non-zero was meant to imply that you can divide by it, thus getting a Kronecker delta – B. Pasternak May 05 '16 at 16:27
-
Means that some basis which makes the ordered basis , a dial basis is ek/aik – Shona May 05 '16 at 16:37