# Linear Algebra/Orthogonal Projection Onto a Line/Solutions

## Solutions

This exercise is recommended for all readers.
Problem 1

Project the first vector orthogonally onto the line spanned by the second vector.

1. ${\begin{pmatrix}2\\1\end{pmatrix}}$ , ${\begin{pmatrix}3\\-2\end{pmatrix}}$
2. ${\begin{pmatrix}2\\1\end{pmatrix}}$ , ${\begin{pmatrix}3\\0\end{pmatrix}}$
3. ${\begin{pmatrix}1\\1\\4\end{pmatrix}}$ , ${\begin{pmatrix}1\\2\\-1\end{pmatrix}}$
4. ${\begin{pmatrix}1\\1\\4\end{pmatrix}}$ , ${\begin{pmatrix}3\\3\\12\end{pmatrix}}$

Each is a straightforward application of the formula from Definition 1.1.

1. $\displaystyle {\frac {{\begin{pmatrix}2\\1\end{pmatrix}}\cdot {\begin{pmatrix}3\\-2\end{pmatrix}}}{{\begin{pmatrix}3\\-2\end{pmatrix}}\cdot {\begin{pmatrix}3\\-2\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\-2\end{pmatrix}}={\frac {4}{13}}\cdot {\begin{pmatrix}3\\-2\end{pmatrix}}={\begin{pmatrix}12/13\\-8/13\end{pmatrix}}$
2. $\displaystyle {\frac {{\begin{pmatrix}2\\1\end{pmatrix}}\cdot {\begin{pmatrix}3\\0\end{pmatrix}}}{{\begin{pmatrix}3\\0\end{pmatrix}}\cdot {\begin{pmatrix}3\\0\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\0\end{pmatrix}}={\frac {2}{3}}\cdot {\begin{pmatrix}3\\0\end{pmatrix}}={\begin{pmatrix}2\\0\end{pmatrix}}$
3. $\displaystyle {\frac {{\begin{pmatrix}1\\1\\4\end{pmatrix}}\cdot {\begin{pmatrix}1\\2\\-1\end{pmatrix}}}{{\begin{pmatrix}1\\2\\-1\end{pmatrix}}\cdot {\begin{pmatrix}1\\2\\-1\end{pmatrix}}}}\cdot {\begin{pmatrix}1\\2\\-1\end{pmatrix}}={\frac {-1}{6}}\cdot {\begin{pmatrix}1\\2\\-1\end{pmatrix}}={\begin{pmatrix}-1/6\\-1/3\\1/6\end{pmatrix}}$
4. $\displaystyle {\frac {{\begin{pmatrix}1\\1\\4\end{pmatrix}}\cdot {\begin{pmatrix}3\\3\\12\end{pmatrix}}}{{\begin{pmatrix}3\\3\\12\end{pmatrix}}\cdot {\begin{pmatrix}3\\3\\12\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\3\\12\end{pmatrix}}={\frac {1}{3}}\cdot {\begin{pmatrix}3\\3\\12\end{pmatrix}}={\begin{pmatrix}1\\1\\4\end{pmatrix}}$
This exercise is recommended for all readers.
Problem 2

Project the vector orthogonally onto the line.

1. ${\begin{pmatrix}2\\-1\\4\end{pmatrix}},\{c{\begin{pmatrix}-3\\1\\-3\end{pmatrix}}\,{\big |}\,c\in \mathbb {R} \}$
2. ${\begin{pmatrix}-1\\-1\end{pmatrix}}$ , the line $y=3x$
1. $\displaystyle {\frac {{\begin{pmatrix}2\\-1\\4\end{pmatrix}}\cdot {\begin{pmatrix}-3\\1\\-3\end{pmatrix}}}{{\begin{pmatrix}-3\\1\\-3\end{pmatrix}}\cdot {\begin{pmatrix}-3\\1\\-3\end{pmatrix}}}}\cdot {\begin{pmatrix}-3\\1\\-3\end{pmatrix}}={\frac {-19}{19}}\cdot {\begin{pmatrix}-3\\1\\-3\end{pmatrix}}={\begin{pmatrix}3\\-1\\3\end{pmatrix}}$
2. Writing the line as $\{c\cdot {\begin{pmatrix}1\\3\end{pmatrix}}\,{\big |}\,c\in \mathbb {R} \}$  gives this projection.
${\frac {{\begin{pmatrix}-1\\-1\end{pmatrix}}\cdot {\begin{pmatrix}1\\3\end{pmatrix}}}{{\begin{pmatrix}1\\3\end{pmatrix}}\cdot {\begin{pmatrix}1\\3\end{pmatrix}}}}\cdot {\begin{pmatrix}1\\3\end{pmatrix}}={\frac {-4}{10}}\cdot {\begin{pmatrix}1\\3\end{pmatrix}}={\begin{pmatrix}-2/5\\-6/5\end{pmatrix}}$
Problem 3

Although the development of Definition 1.1 is guided by the pictures, we are not restricted to spaces that we can draw. In $\mathbb {R} ^{4}$  project this vector onto this line.

${\vec {v}}={\begin{pmatrix}1\\2\\1\\3\end{pmatrix}}\qquad \ell =\{c\cdot {\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}\,{\big |}\,c\in \mathbb {R} \}$

$\displaystyle {\frac {{\begin{pmatrix}1\\2\\1\\3\end{pmatrix}}\cdot {\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}}{{\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}\cdot {\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}}}\cdot {\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}={\frac {3}{4}}\cdot {\begin{pmatrix}-1\\1\\-1\\1\end{pmatrix}}={\begin{pmatrix}-3/4\\3/4\\-3/4\\3/4\end{pmatrix}}$

This exercise is recommended for all readers.
Problem 4

Definition 1.1 uses two vectors ${\vec {s}}$  and ${\vec {v}}$ . Consider the transformation of $\mathbb {R} ^{2}$  resulting from fixing

${\vec {s}}={\begin{pmatrix}3\\1\end{pmatrix}}$

and projecting ${\vec {v}}$  onto the line that is the span of ${\vec {s}}$ . Apply it to these vectors.

1. ${\begin{pmatrix}1\\2\end{pmatrix}}$
2. ${\begin{pmatrix}0\\4\end{pmatrix}}$

Show that in general the projection tranformation is this.

${\begin{pmatrix}x_{1}\\x_{2}\end{pmatrix}}\mapsto {\begin{pmatrix}(x_{1}+3x_{2})/10\\(3x_{1}+9x_{2})/10\end{pmatrix}}$

Express the action of this transformation with a matrix.

1. $\displaystyle {\frac {{\begin{pmatrix}1\\2\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}{{\begin{pmatrix}3\\1\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\frac {1}{2}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\begin{pmatrix}3/2\\1/2\end{pmatrix}}$
2. $\displaystyle {\frac {{\begin{pmatrix}0\\4\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}{{\begin{pmatrix}3\\1\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\frac {2}{5}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\begin{pmatrix}6/5\\2/5\end{pmatrix}}$

In general the projection is this.

${\frac {{\begin{pmatrix}x_{1}\\x_{2}\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}{{\begin{pmatrix}3\\1\end{pmatrix}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\frac {3x_{1}+x_{2}}{10}}\cdot {\begin{pmatrix}3\\1\end{pmatrix}}={\begin{pmatrix}(9x_{1}+3x_{2})/10\\(3x_{1}+x_{2})/10\end{pmatrix}}$

The appropriate matrix is this.

${\begin{pmatrix}9/10&3/10\\3/10&1/10\end{pmatrix}}$
Problem 5

Example 1.5 suggests that projection breaks ${\vec {v}}$  into two parts, ${\mbox{proj}}_{[{\vec {s}}\,]}({{\vec {v}}\,})$  and ${\vec {v}}-{\mbox{proj}}_{[{\vec {s}}\,]}({{\vec {v}}\,})$ , that are "not interacting". Recall that the two are orthogonal. Show that any two nonzero orthogonal vectors make up a linearly independent set.

Suppose that ${\vec {v}}_{1}$  and ${\vec {v}}_{2}$  are nonzero and orthogonal. Consider the linear relationship $c_{1}{\vec {v}}_{1}+c_{2}{\vec {v}}_{2}={\vec {0}}$ . Take the dot product of both sides of the equation with ${\vec {v}}_{1}$  to get that

${\vec {v}}_{1}\cdot (c_{1}{\vec {v}}_{1}+c_{2}{\vec {v}}_{2})=c_{1}\cdot ({\vec {v}}_{1}\cdot {\vec {v}}_{1})+c_{2}\cdot ({\vec {v}}_{1}\cdot {\vec {v}}_{2})=c_{1}\cdot ({\vec {v}}_{1}\cdot {\vec {v}}_{1})+c_{2}\cdot 0=c_{1}\cdot ({\vec {v}}_{1}\cdot {\vec {v}}_{1})$

is equal to ${\vec {v}}_{1}\cdot {\vec {0}}={\vec {0}}$ . With the assumption that ${\vec {v}}_{1}$  is nonzero, this gives that $c_{1}$  is zero. Showing that $c_{2}$  is zero is similar.

Problem 6
1. What is the orthogonal projection of ${\vec {v}}$  onto a line if ${\vec {v}}$  is a member of that line?
2. Show that if ${\vec {v}}$  is not a member of the line then the set $\{{\vec {v}},{\vec {v}}-{\mbox{proj}}_{[{\vec {s}}\,]}({{\vec {v}}\,})\}$  is linearly independent.
1. If the vector ${\vec {v}}\,$  is in the line then the orthogonal projection is ${\vec {v}}$ . To verify this by calculation, note that since ${\vec {v}}\,$  is in the line we have that ${\vec {v}}=c_{\vec {v}}\cdot {\vec {s}}\,$  for some scalar $c_{\vec {v}}$ .
${\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}={\frac {c_{\vec {v}}\cdot {\vec {s}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}=c_{\vec {v}}\cdot {\frac {{\vec {s}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}=c_{\vec {v}}\cdot 1\cdot {\vec {s}}={\vec {v}}$
(Remark. If we assume that ${\vec {v}}\,$  is nonzero then the above is simplified on taking ${\vec {s}}\,$  to be ${\vec {v}}$ .)
2. Write $c_{\vec {p}}{\vec {s}}\,$  for the projection ${\mbox{proj}}_{[{\vec {s}}\,]}({\vec {v}})$ . Note that, by the assumption that ${\vec {v}}$  is not in the line, both ${\vec {v}}$  and ${\vec {v}}-c_{\vec {p}}{\vec {s}}\,$  are nonzero. Note also that if $c_{\vec {p}}\,$  is zero then we are actually considering the one-element set $\{{\vec {v}}\,\}$ , and with ${\vec {v}}$  nonzero, this set is necessarily linearly independent. Therefore, we are left considering the case that $c_{\vec {p}}$  is nonzero. Setting up a linear relationship
$a_{1}({\vec {v}})+a_{2}({\vec {v}}-c_{\vec {p}}{\vec {s}})={\vec {0}}$
leads to the equation $(a_{1}+a_{2})\cdot {\vec {v}}=a_{2}c_{\vec {p}}\cdot {\vec {s}}$ . Because ${\vec {v}}\,$  isn't in the line, the scalars $a_{1}+a_{2}$  and $a_{2}c_{\vec {p}}$  must both be zero. The $c_{\vec {p}}=0$  case is handled above, so the remaining case is that $a_{2}=0$ , and this gives that $a_{1}=0$  also. Hence the set is linearly independent.
Problem 7

Definition 1.1 requires that ${\vec {s}}\,$  be nonzero. Why? What is the right definition of the orthogonal projection of a vector onto the (degenerate) line spanned by the zero vector?

If ${\vec {s}}\,$  is the zero vector then the expression

${\mbox{proj}}_{[{\vec {s}}\,]}({\vec {v}})={\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}$

contains a division by zero, and so is undefined. As for the right definition, for the projection to lie in the span of the zero vector, it must be defined to be ${\vec {0}}$ .

Problem 8

Are all vectors the projection of some other vector onto some line?

Any vector in $\mathbb {R} ^{n}$  is the projection of some other onto a line, provided that the dimension $n$  is greater than one. (Clearly, any vector is the projection of itself into a line containing itself; the question is to produce some vector other than ${\vec {v}}$  that projects to ${\vec {v}}$ .)

Suppose that ${\vec {v}}\in \mathbb {R} ^{n}$  with $n>1$ . If ${\vec {v}}\neq {\vec {0}}$  then we consider the line $\ell =\{c{\vec {v}}\,{\big |}\,c\in \mathbb {R} \}$  and if ${\vec {v}}={\vec {0}}$  we take $\ell$  to be any (nondegenerate) line at all (actually, we needn't distinguish between these two cases— see the prior exercise). Let $v_{1},\dots ,v_{n}$  be the components of ${\vec {v}}$ ; since $n>1$ , there are at least two. If some $v_{i}$  is zero then the vector ${\vec {w}}={\vec {e}}_{i}$  is perpendicular to ${\vec {v}}$ . If none of the components is zero then the vector ${\vec {w}}\,$  whose components are $v_{2},-v_{1},0,\dots ,0$  is perpendicular to ${\vec {v}}$ . In either case, observe that ${\vec {v}}+{\vec {w}}$  does not equal ${\vec {v}}$ , and that ${\vec {v}}$  is the projection of ${\vec {v}}+{\vec {w}}$  onto $\ell$ .

${\frac {({\vec {v}}+{\vec {w}})\cdot {\vec {v}}}{{\vec {v}}\cdot {\vec {v}}}}\cdot {\vec {v}}={\bigl (}{\frac {{\vec {v}}\cdot {\vec {v}}}{{\vec {v}}\cdot {\vec {v}}}}+{\frac {{\vec {w}}\cdot {\vec {v}}}{{\vec {v}}\cdot {\vec {v}}}}{\bigr )}\cdot {\vec {v}}={\frac {{\vec {v}}\cdot {\vec {v}}}{{\vec {v}}\cdot {\vec {v}}}}\cdot {\vec {v}}={\vec {v}}$

We can dispose of the remaining $n=0$  and $n=1$  cases. The dimension $n=0$  case is the trivial vector space, here there is only one vector and so it cannot be expressed as the projection of a different vector. In the dimension $n=1$  case there is only one (nondegenerate) line, and every vector is in it, hence every vector is the projection only of itself.

This exercise is recommended for all readers.
Problem 9

Show that the projection of ${\vec {v}}$  onto the line spanned by ${\vec {s}}$  has length equal to the absolute value of the number ${\vec {v}}\cdot {\vec {s}}$  divided by the length of the vector ${\vec {s}}$ .

The proof is simply a calculation.

$|{\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}\,|=|{\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}|\cdot |{\vec {s}}\,|={\frac {|{\vec {v}}\cdot {\vec {s}}\,|}{|{\vec {s}}\,|^{2}}}\cdot |{\vec {s}}\,|={\frac {|{\vec {v}}\cdot {\vec {s}}\,|}{|{\vec {s}}\,|}}$
Problem 10

Find the formula for the distance from a point to a line.

Because the projection of ${\vec {v}}$  onto the line spanned by ${\vec {s}}$  is

${\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}$

the distance squared from the point to the line is this (a vector dotted with itself ${\vec {w}}\cdot {\vec {w}}$  is written ${\vec {w}}^{2}$ ).

${\begin{array}{rl}|{\vec {v}}-{\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}\,|^{2}&={\vec {v}}\cdot {\vec {v}}-{\vec {v}}\cdot ({\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}})-({\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}\,)\cdot {\vec {v}}+({\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}\,)^{2}\\&={\vec {v}}\cdot {\vec {v}}-2\cdot ({\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}})\cdot {\vec {v}}\cdot {\vec {s}}+({\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}})\cdot {\vec {s}}\cdot {\vec {s}}\\&={\frac {({\vec {v}}\cdot {\vec {v}}\,)\cdot ({\vec {s}}\cdot {\vec {s}}\,)-2\cdot ({\vec {v}}\cdot {\vec {s}}\,)^{2}+({\vec {v}}\cdot {\vec {s}}\,)^{2}}{{\vec {s}}\cdot {\vec {s}}}}\\&={\frac {({\vec {v}}\cdot {\vec {v}}\,)({\vec {s}}\cdot {\vec {s}}\,)-({\vec {v}}\cdot {\vec {s}}\,)^{2}}{{\vec {s}}\cdot {\vec {s}}}}\end{array}}$
Problem 11

Find the scalar $c$  such that $(cs_{1},cs_{2})$  is a minimum distance from the point $(v_{1},v_{2})$  by using calculus (i.e., consider the distance function, set the first derivative equal to zero, and solve). Generalize to $\mathbb {R} ^{n}$ .

Because square root is a strictly increasing function, we can minimize $d(c)=(cs_{1}-v_{1})^{2}+(cs_{2}-v_{2})^{2}$  instead of the square root of $d$ . The derivative is $dd/dc=2(cs_{1}-v_{1})\cdot s_{1}+2(cs_{2}-v_{2})\cdot s_{2}$ . Setting it equal to zero $2(cs_{1}-v_{1})\cdot s_{1}+2(cs_{2}-v_{2})\cdot s_{2}=c\cdot (2s_{1}^{2}+2s_{2}^{2})-(2v_{1}s_{1}+2v_{2}s_{2})=0$  gives the only critical point.

$c={\frac {v_{1}s_{1}+v_{2}s_{2}}{{s_{1}}^{2}+{s_{2}}^{2}}}={\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}$

Now the second derivative with respect to $c$

${\frac {d^{2}\,d}{dc^{2}}}=2{s_{1}}^{2}+2{s_{2}}^{2}$

is strictly positive (as long as neither $s_{1}$  nor $s_{2}$  is zero, in which case the question is trivial) and so the critical point is a minimum.

The generalization to $\mathbb {R} ^{n}$  is straightforward. Consider $d_{n}(c)=(cs_{1}-v_{1})^{2}+\dots +(cs_{n}-v_{n})^{2}$ , take the derivative, etc.

This exercise is recommended for all readers.
Problem 12

Prove that the orthogonal projection of a vector onto a line is shorter than the vector.

The Cauchy-Schwarz inequality $|{\vec {v}}\cdot {\vec {s}}\,|\leq |{\vec {v}}\,|\cdot |{\vec {s}}\,|$  gives that this fraction

$|{\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}\cdot {\vec {s}}\,|=|{\frac {{\vec {v}}\cdot {\vec {s}}}{{\vec {s}}\cdot {\vec {s}}}}|\cdot |{\vec {s}}\,|={\frac {|{\vec {v}}\cdot {\vec {s}}\,|}{|{\vec {s}}\,|^{2}}}\cdot |{\vec {s}}\,|={\frac {|{\vec {v}}\cdot {\vec {s}}\,|}{|{\vec {s}}\,|}}$

when divided by $|{\vec {v}}\,|$  is less than or equal to one. That is, $|{\vec {v}}\,|$  is larger than or equal to the fraction.

This exercise is recommended for all readers.
Problem 13

Show that the definition of orthogonal projection onto a line does not depend on the spanning vector: if ${\vec {s}}$  is a nonzero multiple of ${\vec {q}}$  then $({\vec {v}}\cdot {\vec {s}}/{\vec {s}}\cdot {\vec {s}}\,)\cdot {\vec {s}}$  equals $({\vec {v}}\cdot {\vec {q}}/{\vec {q}}\cdot {\vec {q}}\,)\cdot {\vec {q}}$ .

Write $c{\vec {s}}$  for ${\vec {q}}$ , and calculate: $({\vec {v}}\cdot c{\vec {s}}/c{\vec {s}}\cdot c{\vec {s}}\,)\cdot c{\vec {s}}=({\vec {v}}\cdot {\vec {s}}/{\vec {s}}\cdot {\vec {s}}\,)\cdot {\vec {s}}$ .

This exercise is recommended for all readers.
Problem 14

Consider the function mapping to plane to itself that takes a vector to its projection onto the line $y=x$ . These two each show that the map is linear, the first one in a way that is bound to the coordinates (that is, it fixes a basis and then computes) and the second in a way that is more conceptual.

1. Produce a matrix that describes the function's action.
2. Show also that this map can be obtained by first rotating everything in the plane $\pi /4$  radians clockwise, then projecting onto the $x$ -axis, and then rotating $\pi /4$  radians counterclockwise.
1. Fixing
${\vec {s}}={\begin{pmatrix}1\\1\end{pmatrix}}$
as the vector whose span is the line, the formula gives this action,
${\begin{pmatrix}x\\y\end{pmatrix}}\mapsto {\frac {{\begin{pmatrix}x\\y\end{pmatrix}}\cdot {\begin{pmatrix}1\\1\end{pmatrix}}}{{\begin{pmatrix}1\\1\end{pmatrix}}\cdot {\begin{pmatrix}1\\1\end{pmatrix}}}}\cdot {\begin{pmatrix}1\\1\end{pmatrix}}={\frac {x+y}{2}}\cdot {\begin{pmatrix}1\\1\end{pmatrix}}={\begin{pmatrix}(x+y)/2\\(x+y)/2\end{pmatrix}}$
which is the effect of this matrix.
${\begin{pmatrix}1/2&1/2\\1/2&1/2\end{pmatrix}}$
2. Rotating the entire plane $\pi /4$  radians clockwise brings the $y=x$  line to lie on the $x$ -axis. Now projecting and then rotating back has the desired effect.
Problem 15

For ${\vec {a}},{\vec {b}}\in \mathbb {R} ^{n}$  let ${\vec {v}}_{1}$  be the projection of ${\vec {a}}$  onto the line spanned by ${\vec {b}}$ , let ${\vec {v}}_{2}$  be the projection of ${\vec {v}}_{1}$  onto the line spanned by ${\vec {a}}$ , let ${\vec {v}}_{3}$  be the projection of ${\vec {v}}_{2}$  onto the line spanned by ${\vec {b}}$ , etc., back and forth between the spans of ${\vec {a}}$  and ${\vec {b}}$ . That is, ${\vec {v}}_{i+1}$  is the projection of ${\vec {v}}_{i}$  onto the span of ${\vec {a}}$  if $i+1$  is even, and onto the span of ${\vec {b}}$  if $i+1$  is odd. Must that sequence of vectors eventually settle down— must there be a sufficiently large $i$  such that ${\vec {v}}_{i+2}$  equals ${\vec {v}}_{i}$  and ${\vec {v}}_{i+3}$  equals ${\vec {v}}_{i+1}$ ? If so, what is the earliest such $i$ ?

${\vec {a}}={\begin{pmatrix}1\\0\end{pmatrix}}\qquad {\vec {b}}={\begin{pmatrix}1\\1\end{pmatrix}}$
${\vec {v}}_{1}={\begin{pmatrix}1/2\\1/2\end{pmatrix}},\quad {\vec {v}}_{2}={\begin{pmatrix}1/2\\0\end{pmatrix}},\quad {\vec {v}}_{3}={\begin{pmatrix}1/4\\1/4\end{pmatrix}},\quad \ldots$