GraphMath

QRF, Gram-Schmidt, part 2

From R to QR: triangular structure, deformation and uniqueness

What does the R part of QR factorization represent?

This chapter develops the structure of matrix R, shows why upper-triangular form appears in QR factorization and explains how QR separates deformation from rotation or reflection. It then presents geometric views of the factorization in square and tall cases, interprets the hierarchy encoded by R and ends with the uniqueness of thin QR factorization.

Key ideas

After Gram-Schmidt produces Q, the matrix R records how the original columns are assembled from the orthonormal directions. Its upper-triangular form reveals both algebraic dependence and geometric order.

  • R = QᵀA, and zeros below the diagonal appear because each new orthogonal direction is orthogonal to the earlier columns
  • The diagonal entries of R are chosen nonnegative, which fixes sign ambiguity and supports uniqueness of thin QR
  • In A = QR, R gives hierarchical deformation while Q provides rotation or reflection
  • The same factorization can also be read as a basis interpretation: Q defines a new orthonormal basis and R gives coordinates in that basis
  • The hierarchy of R explains both forward transformation of space and backward solution of triangular systems

Together with part 1, this chapter completes the structural and geometric picture of QR factorization.

Why is R upper triangular?

Because each orthogonal direction is built from the current column after removing components along earlier directions. That makes later basis vectors orthogonal to earlier columns, which creates zeros below the diagonal in QᵀA = R.

Related chapters

Chapter contents

The chapter is available as a PDF. Page links below are best-effort: most browsers support them but some viewers may ignore the page hint.

Topic Pages
Structure of matrix R 1–2
Geometric view of QR factorization 2–5
QRF of a 2×2 matrix (Wikipedia image) 6
QRF of a 3×3 matrix (Wikipedia image) 7
Alternative geometric view 8
QRF of a 3×2 matrix 9–10
'Hierarchy' of R 10–11
Uniqueness of QR factorization 12

Why is thin QR factorization unique?

Each orthogonal direction could be flipped in sign, but requiring nonnegative diagonal entries in R removes that ambiguity. Under that convention, thin QR gives a unique orthonormal basis for the column space together with a unique upper-triangular factor.

Was this chapter helpful?

Quick feedback helps us improve the site.