Linear Independence Theorem for Vectors 2

Given a finitely generated vector space V, if we consider a set of generators $$ \{ \vec{v}_1, \vec{v}_2, ...,\vec{v}_n \} $$ and a set of linearly independent vectors belonging to V $$ \{ \vec{w}_1, \vec{w}_2, ..., \vec{w}_p \} $$, then $$ p \le n $$.

Essentially, the number of linearly independent vectors in V is always less than or equal to the number of vectors in a set of generators for V.

    The Proof

    For a real vector space V, let's consider a set of generators for V

    $$ \{ \vec{v}_1, \vec{v}_2, ...,\vec{v}_n \} $$

    and a set of linearly independent vectors belonging to V

    $$ \{ \vec{w}_1, \vec{w}_2, ..., \vec{w}_p \} $$

    Any vector w ∈ V can be generated by the linear combination of the generators {v1,v2,...,vn}

    $$ \vec{w}_1 = \lambda_1 \vec{v}_1 + \lambda_2 \vec{v}_2 + ... + \lambda_n \vec{v}_n $$

    By assumption, w1 is linearly independent.

    Therefore, at least one of the coefficients λ must be nonzero. Otherwise, it would be linearly dependent.

    We consider the last coefficient λn≠0.

    $$ \lambda_n \ne 0 $$

    Since λn≠0 is nonzero, we divide both sides of the equation by λn

    $$ \vec{w}_1 = \lambda_1 \vec{v}_1 + \lambda_2 \vec{v}_2 + ... + \lambda_{n-1} \vec{v}_{n-1} + \lambda_n \vec{v}_n $$

    $$ \frac{1}{ \lambda_n} \vec{w}_1 = \frac{\lambda_1}{ \lambda_n} \vec{v}_1 + \frac{\lambda_2}{ \lambda_n} \vec{v}_2 + ... + \frac{\lambda_{n-1}}{\lambda_n} \vec{v}_{n-1} + \frac{\lambda_n}{ \lambda_n} \vec{v}_n $$

    $$ \frac{1}{ \lambda_n} \vec{w}_1 = \frac{\lambda_1}{ \lambda_n} \vec{v}_1 + \frac{\lambda_2}{ \lambda_n} \vec{v}_2 + ... + \frac{\lambda_{n-1}}{\lambda_n} \vec{v}_{n-1} + \vec{v}_n $$

    Next, we derive the vector vn

    $$ \vec{v}_n = \frac{1}{ \lambda_n} \vec{w}_1 - \frac{\lambda_1}{ \lambda_n} \vec{v}_1 - \frac{\lambda_2}{ \lambda_n} \vec{v}_2 - ... - \frac{\lambda_{n-1}}{\lambda_n} \vec{v}_{n-1} $$

    We then replace the coefficients with another variable 1/λn1, λ1n2, λ2n3,..., λn-1nn,

    $$ \vec{v}_n = \alpha_1 \vec{w}_1 - \alpha_2 \vec{v}_1 - \alpha_3 \vec{v}_2 - ... - \alpha_n \vec{v}_{n-1} $$

    Thus, the set of vectors {w1,v1,v2,...,vn-1} is still a set of generators with the same number of vectors as before because if we replace vn= α1 w1 + α2 v1 + α3 v2 +...+ αnvn-1 in the linear combination of a generic vector v of the vector space V, we get another set of generators for V.

    $$ \vec{v} = \lambda_1 \vec{v}_1 + \lambda_2 \vec{v}_2 + ... + \lambda_n \vec{v}_n $$

    Knowing that vn= α1 w1 + α2 v1 + α3 v2 +...+ αnvn-1

    $$ \vec{v} = \lambda_1 \vec{v}_1 + \lambda_2 \vec{v}_2 + ... + \lambda_n ( \alpha_1 \vec{w}_1 - \alpha_2 \vec{v}_1 - \alpha_3 \vec{v}_2 - ... - \alpha_n \vec{v}_{n-1} ) $$

    $$ \vec{v} = \lambda_1 \vec{v}_1 + \lambda_2 \vec{v}_2 + ... + \lambda_n \alpha_1 \vec{w}_1 - \lambda_n \alpha_2 \vec{v}_1 - \lambda_n \alpha_3 \vec{v}_2 - ... - \lambda_n \alpha_n \vec{v}_{n-1} $$

    $$ \vec{v} = ( \lambda_1 - \lambda_n \alpha_2 ) \vec{v}_1 + ( \lambda_2 - \lambda_n \alpha_3 ) \vec{v}_2 + ... + (\lambda_{n-1} - \lambda_n \alpha_n ) \vec{v}_{n-1} + ( \lambda_n \alpha_1) \vec{w}_1 $$

    We then replace the coefficients with another variable λ1nα21, and so on.

    $$ \vec{v} = \beta_1 \vec{v}_1 + \beta_2 \vec{v}_2 + ... + \beta_{n-1} \vec{v}_{n-1} + \beta_n \vec{w}_1 $$

    Thus, the vector v is a combination of n vectors {w1,v1,v2,...,vn-1}

    Next, we repeat the same process, hypothesizing that βn-1≠0 and replacing vn-1 with w2

    We continue by replacing vn-2 with w3, and so on... until we replace v1 with wn

    Ultimately, once all the generator vectors are replaced, we obtain a set of generators {w1,w2,...,wn}

    $$ \vec{v} = \lambda_1 \vec{w}_1 + \lambda_2 \vec{w}_2 + ... + \lambda_{n-1} \lambda{w}_{n-1} + \lambda_n \vec{w}_n $$

    Given that the initially hypothesized linearly independent vectors were p vectors

    $$ \{ \vec{w}_1, \vec{w}_2, ...,\vec{w}_p \} $$

    If p>n, there would be vectors wn+1, wn+2, ... , wp that are not included in the set of generators {w1,w2,...,wn}

    $$ \{ \vec{w}_1, \vec{w}_2 , ... , \vec{w}_n, \color{red}{ \vec{w}_{n+1} }, \color{red}{ \vec{w}_{n+2} }, ..., \color{red}{ \vec{w}_{p} } \} $$

    Hence, I could generate the vector wn+1 using a linear combination of the generator vectors {w1,w2,...,wn}

    $$ \vec{w}_{n+1} = \lambda_1 \vec{w}_1 + \lambda_2 \vec{w}_2 + ... + \lambda_{n-1} \lambda{w}_{n-1} + \lambda_n \vec{w}_n $$

    The vector wn+1 is linearly dependent on {w1,w2,...,wn}

    However, this contradicts the initial assumption which states that the vectors {w1,w2,...,wp} are linearly independent.

    If p>n is not true, then the opposite must be true, i.e., p≤n.

    In conclusion, if the vector space V has p linearly independent vectors, then the number n of vectors in any generator of V must be equal to or greater than p.

    $$ n \le p $$

    And so on.

     
     

    Please feel free to point out any errors or typos, or share suggestions to improve these notes. English isn't my first language, so if you notice any mistakes, let me know, and I'll be sure to fix them.

    FacebookTwitterLinkedinLinkedin
    knowledge base

    Linear Dependency and Independence