Let  A be a flat mXn matrix (m<n).

Let the QR decomposition of  A^{T} be

 A^{T} = \left \begin{bmatrix} Q_{1}&Q_{2} \end{bmatrix} \right \left \begin{bmatrix} R_{1}\\0 \end{bmatrix} \right

And let the singular-value decomposition of  R_{1} be

 R_{1} = \left \begin{bmatrix} U_{1}&U_{2} \end{bmatrix} \right \left \begin{bmatrix} \Sigma_{1}&0 \\ 0&0 \end{bmatrix} \right \left \begin{bmatrix} V_{1}&V_{2} \end{bmatrix} \right^{T}

Can you use the singular-value decomposition of  R_{1} to find the null space of  A ?


There's a bunch of things you already know from the decompositions:

 A^{T} = Q_{1}R_{1}

 R_{1} = U_{1}\Sigma_{1}V_{1}

 A^{T} = Q1_{1}U_{1}\Sigma_{1}V_{1}

 AQ_{2} = 0

 R_{1}V_{2} = 0

 R_{1}^{T}U_{2} = 0

 \text{null} (A^{T}) = \text{null} (R_{1})


But I can't seem to combine the above in any meaningful way.