Generalizing x ⊺ A x = 0 theorems to x ⊺ A ( x ) x = 0 forms?Let x ∈ R n be any real vector and let A = A ⊺ ∈ R n × n be a real symmetric matrix. Then, x ⊺ A x = 0 ∀ x ⟹ A ≡ 0 Can we still say this if A = A ⊺ : R n → R n × n is a real symmetric matrix-valued function of x? I.e., x ⊺ A ( x ) x = 0 ∀ x ⟹ ? A ≡ 0 ∀ xAlso, is there a name for equations of the form f ( x ) = x ⊺ A ( x ) x with symmetric A? It is obviously not quadratic since A depends on x, but it still has a certain quadratic-form feel.

Question

Generalizing       x    ⊺    A  x  =  0 theorems to       x    ⊺    A  (  x  )  x  =  0 forms?Let   x  ∈            R        n   be any real vector and let      A  =      A    ⊺    ∈            R              n      ×      n           be a real symmetric matrix. Then,      x    ⊺    A  x  =  0           ∀  x          ⟹          A  ≡  0 Can we still say this if       A  =      A    ⊺    :            R        n    →            R              n      ×      n          is a real symmetric matrix-valued function of x? I.e.,      x    ⊺    A  (  x  )  x  =  0           ∀  x                        ⟹              ?          A  ≡  0           ∀  xAlso, is there a name for equations of the form   f  (  x  )  =      x    ⊺    A  (  x  )  x with symmetric A? It is obviously not quadratic since A depends on x, but it still has a certain quadratic-form feel.

Sean Sutton · Accepted Answer

Step 1Starting with what you already knowImagine a vector   y  =  A  x. If       x    T    y  =  0, then either   x  ⊥  y, or either y or x must be 0. So A must be the composition of a rotation by   π      /    2 about some axis in any number of dimensions and any scaling. Just imagine the composition of an arbitrary number of Givens Rotations. We can impose more stringent requirements, but we won&#039;t bother.A rotation other than the identity cannot be symmetric. A fine property of matrices representing rotations is that       A    T    =      A          −      1      . If   A  =      A    T   then   A  =  I which is not a rotation by   π      /    2 about anything.Step 2FineSo we&#039;ve shown that A cannot spawn   y  ⊥  x. Therefore, if       x    T    y  =  0 then   y  =  0 or   x  =  0. If   y  =  0 for any   x  ≠  0 then   A  =  0.Let&#039;s ignore for a moment that A is a function of x. If we have any A, produced by a drawing from a hat,Assume there is a function A(x) for which       x    T    (  t  )  A  (  x  ,  t  )  x  (  t  )  =  0 but   A  (  x  ,  t  )  ≠  0. By the above, we are claiming that either:1.   A  (  x  ,  t  )  x  (  t  )  ⊥  x          ∀          x      ,      t      or2,   A  (  x  ,  t  )  =  0We know that 1. must be false, because       A    T    =  A, so 2. must be true.

Brenda Jordan · Accepted Answer

Step 1Nevermind, found a counter-example. The matrix A just needs to have an eigenvector of x with eigenvalue 0, or put another way, x just has to be in the null-space of A. This is obvious to construct in 2D, but of course applies generally. Let,  x  :=      [                                        x            1                                                            x            2                                ]  Step 2Find a vector orthogonal to x and call it v.  v  (  x  )  :=      [                                        x            2                                                −                      x            1                                ]  Construct the following matrix by outer-product,  A  (  x  )  :=  v      v    T    =      [                                        x            2            2                                    −                      x            1                                x            2                                                −                      x            1                                x            2                                                x            1            2                                ]  This matrix is symmetric   ∀  x, is not the zero matrix, but still satisfies       x    T    A  x  =  0  ,        ∀  x.

Generalizing x^⊺ Ax=0 theorems to x^⊺A(x)x=0 forms?

Answered question

Answer & Explanation

New Questions in Algebra I