Best Approximation - Linear Algebra

  1. What is the distance from (2,3)(2, 3) to the line spanned by (1,1)(1, 1)?
    22\frac{\sqrt{2}}{2}. The error (2,3)−(52,52)=(−12,12)(2, 3) - (\frac{5}{2}, \frac{5}{2}) = (-\frac{1}{2}, \frac{1}{2}) has length 12\sqrt{\frac{1}{2}}.
  2. If y\mathbf{y} already lies in the subspace WW, what is its best approximation in WW?
    y\mathbf{y} itself, with error 0\mathbf{0}.
  3. What does the Best Approximation Theorem say about y^=projWy\hat{\mathbf{y}} = \text{proj}_W\mathbf{y}?
    It is the point of WW closest to y\mathbf{y}: ∥y−y^∥<∥y−v∥\|\mathbf{y} - \hat{\mathbf{y}}\| < \|\mathbf{y} - \mathbf{v}\| for every other v\mathbf{v} in WW.
  4. ∥y∥=5\|\mathbf{y}\| = 5 and its best approximation in WW has length 33. What is the distance from y\mathbf{y} to WW?
    44. By the Pythagorean theorem, ∥y−y^∥2=25−9=16\|\mathbf{y} - \hat{\mathbf{y}}\|^2 = 25 - 9 = 16.
  5. Why must the basis be orthogonal in the formula y^=∑iy⋅uiui⋅uiui\hat{\mathbf{y}} = \sum_i \frac{\mathbf{y} \cdot \mathbf{u}_i}{\mathbf{u}_i \cdot \mathbf{u}_i}\mathbf{u}_i?
    For a non-orthogonal basis, the sum of one-dimensional projections generally leaves an error that is not orthogonal to WW, so it is not the closest point.
  6. What is the best approximation of y=(2,3)\mathbf{y} = (2, 3) by a vector on the line spanned by (1,1)(1, 1)?
    (52,52)(\frac{5}{2}, \frac{5}{2}). The projection is y⋅uu⋅uu=52(1,1)\frac{\mathbf{y} \cdot \mathbf{u}}{\mathbf{u} \cdot \mathbf{u}}\mathbf{u} = \frac{5}{2}(1, 1).
  7. What is the best approximation of y=(3,4)\mathbf{y} = (3, 4) by a vector on the line spanned by (1,0)(1, 0)?
    (3,0)(3, 0). The projection is 31(1,0)\frac{3}{1}(1, 0).
  8. How is the error y−y^\mathbf{y} - \hat{\mathbf{y}} of the best approximation related to the subspace WW?
    It is orthogonal to WW, so it lies in W⊥W^\perp.
  9. If y\mathbf{y} is orthogonal to every vector in WW, what is its best approximation in WW?
    The zero vector, since every projection coefficient y⋅ui\mathbf{y} \cdot \mathbf{u}_i is 00.
  10. What is the best approximation of (1,2,3)(1, 2, 3) in the xyxy-plane of R3\mathbb{R}^3, and how far away is it?
    (1,2,0)(1, 2, 0), at distance 33.
  11. If {u1,…,up}\{\mathbf{u}_1, \ldots, \mathbf{u}_p\} is an orthogonal basis of WW, what is the best approximation of y\mathbf{y} in WW?
    y^=∑iy⋅uiui⋅uiui\hat{\mathbf{y}} = \sum_i \frac{\mathbf{y} \cdot \mathbf{u}_i}{\mathbf{u}_i \cdot \mathbf{u}_i}\mathbf{u}_i.
  12. WW is spanned by the orthogonal vectors (1,0,1)(1, 0, 1) and (0,1,0)(0, 1, 0). What is the best approximation of (2,5,4)(2, 5, 4) in WW?
    (3,5,3)(3, 5, 3). It is 62(1,0,1)+51(0,1,0)\frac{6}{2}(1, 0, 1) + \frac{5}{1}(0, 1, 0).

Study these flashcards