Orthogonal Vectors in Rn Problem

In summary: F(t) &= \|a \|^2 + 2 \langle a,(b-a) \rangle t + \| b-a \|^2 t^2 \\&= t^2 \| b-a \|^2 + 2 t \langle a, b-a \rangle + \| a \|^2\end{align*}$$This is a parabola with a minimum at ##t = - \langle a, b-a \rangle / \| b-a \|^2.##In summary, to determine a vector c that lies on a line segment [a,b] and is perpendicular to (b-a), we can use the formula c = a + t_0 (
  • #1
Onezimo Cardoso
11
2

Homework Statement


Given ##a\neq b## vectors of ##\mathbb{R}^n##. Determine ##c## which lies in the line segment ##[a,b]=\{a+t(b-a) ; t \in [0,1]\}##, such that ##c \perp (b-a)##. Conclude that for all ##x \in [a,b]##, with ##x\neq c## it is true that ##|c|<|x|##.

Homework Equations


The first part of the question can be solved just using inner product definition.
I don't know how the second part can be solved. But I think could be useful the Cauchy-Schawarz inequality ##|<x,y>| \leq |x||y|## or maybe the cosine rule...

The Attempt at a Solution


If we suppose that ##c \in [a,b]## then we can write ##c## as:
$$c=a+t_0 (b-a)$$
where ##t_0 \in [0,1]##.
By the fact that ##c \perp (b-a)## we have:
$$<c,(b-a)> \quad = \quad0$$
$$\Rightarrow \quad <a+t_0 (b-a),(b-a)> \quad = \quad 0$$
$$\Rightarrow \quad <a,(b-a)>+<t_0 (b-a),(b-a)> \quad =\quad 0$$
$$\Rightarrow \quad <a,(b-a)>+t_0 |b-a|^2=0$$
$$\Rightarrow t_0 = - \frac{<a,(b-a)>}{|b-a|^2} $$

Then ##c## can be uniquely determined as ##c=a+t_0 (b-a)##, where ##t_0 = - \frac{<a,(b-a)>}{|b-a|^2}##.
$$\ldots$$
Now we need to prove that if we consider any other ##x \neq c## in the line segment ##[a,b]## then ##|c|<|x|##.

upload_2018-7-12_16-35-57.png
 

Attachments

  • upload_2018-7-12_16-35-57.png
    upload_2018-7-12_16-35-57.png
    11.9 KB · Views: 379
Last edited:
Physics news on Phys.org
  • #2
You can show that ##x = c + k(b-a)## for some ##k \ne 0##. Then calculate ##\langle x,x \rangle##.
 
  • Like
Likes Onezimo Cardoso
  • #3
Onezimo Cardoso said:

Homework Statement


Given ##a\neq b## vectors of ##\mathbb{R}^n##. Determine ##c## which lies in the line segment ##[a,b]=\{a+t(b-a) ; t \in [0,1]\}##, such that ##c \perp (b-a)##. Conclude that for all ##x \in [a,b]##, with ##x\neq c## it is true that ##|c|<|x|##.
The premise of the question is wrong: if ##a \neq b, \; a,b \in \mathbb{R}^n## it is not necessarily true that ##c \perp (b-a)## for some ##t \in [0,1].## For example, if ##a = (1,1)## and ##b = (2,1)##, we have ##b-a = (1,0)## and ##c = (1,1)+t(1,0) = (1+t,1)##. In order to have ##c \perp (b-a)## we need ##t = -1##, so the"perpendicular" ##c## is not on the segment from ##a## to ##b##; it is outside that segment, but still on the line through ##a## and ##b##.

However, if you allow values ##t < 0## and ##t > 1##---in other words, if you allow any ##t \in \mathbb{R}##---then the "minimization" result is true. The easiest way to get that is to minimize ##F(t) \equiv \| c \|^2##, which is a quadratic function of ##t##.
 
Last edited:
  • Like
Likes Onezimo Cardoso
  • #4
Ray Vickson said:
The premise of the question is wrong: if ##a \neq b, \; a,b \in \mathbb{R}^n## it is not necessarily true that ##c \perp (b-a)## for some ##t \in [0,1].## For example, if ##a = (1,1)## and ##b = (2,1)##, we have ##b-a = (1,0)## and ##c = (1,1)+t(1,0) = (1+t,1)##. In order to have ##c \perp (b-a)## we need ##t = -1##, so the"perpendicular" ##c## is not on the segment from ##a## to ##b##; it is outside that segment, but still on the line through ##a## and ##b##.

However, if you allow values ##t < 0## and ##t > 1##---in other words, if you allow any ##t \in \mathbb{R}##---then the "minimization" result is true. The easiest way to get that is to minimize ##F(t) \equiv \| c \|^2##, which is a quadratic function of ##t##.

Very well noticed Ray Vickson!
Ok I can reformulate the question as follow:

Homework Statement


Given ##a\neq b## vectors of ##\mathbb{R}^n##. Determine ##c## which lies in the line ##r## determined by in the line segment ##[a,b]=\{a+t(b-a) ; t \in [0,1]\}##, such that ##c \perp r##. Conclude that for all ##x \in r##, with ##x\neq c## it is true that ##|c|<|x|##.

The solution I wrote before can be used in the same way, i.e:
$$c = a + t_0 (b-a) \quad where \quad t_0 = - \frac{<a,(b-a)>}{|b-a|^2}$$

But regarding the last part, i.e., to prove that ##|x|<|c|## for all ##x \in r##, I have no clue how can I minimize the function ##F(t) \equiv \| x \|^2 \equiv \| a+t(b-a) \|^2##.
 
  • #5
vela said:
You can show that ##x = c + k(b-a)## for some ##k \ne 0##. Then calculate ##\langle x,x \rangle##.

In fact vela, we can write any ##x## in the line determined by the line segment ##[a,b]## as ##x = c + k(b-a)##.
But follow your tip I stucked at the following red question mark:

upload_2018-7-12_19-58-37.png
 

Attachments

  • upload_2018-7-12_19-58-37.png
    upload_2018-7-12_19-58-37.png
    15.4 KB · Views: 363
  • #6
Don’t substitute for ##c##.
 
  • #7
Onezimo Cardoso said:
Very well noticed Ray Vickson!
Ok I can reformulate the question as follow:

Homework Statement


Given ##a\neq b## vectors of ##\mathbb{R}^n##. Determine ##c## which lies in the line ##r## determined by in the line segment ##[a,b]=\{a+t(b-a) ; t \in [0,1]\}##, such that ##c \perp r##. Conclude that for all ##x \in r##, with ##x\neq c## it is true that ##|c|<|x|##.

The solution I wrote before can be used in the same way, i.e:
$$c = a + t_0 (b-a) \quad where \quad t_0 = - \frac{<a,(b-a)>}{|b-a|^2}$$

But regarding the last part, i.e., to prove that ##|x|<|c|## for all ##x \in r##, I have no clue how can I minimize the function ##F(t) \equiv \| x \|^2 \equiv \| a+t(b-a) \|^2##.

As I said already, that is a quadratic function of ##t##. Just expand out ##\| a + (b-a)t \|^2##, using the definition of ##\| \cdot \|^2.##
 

1. What are orthogonal vectors in Rn?

Orthogonal vectors in Rn are two vectors that are perpendicular to each other, meaning they form a 90 degree angle at their intersection. They are commonly used in linear algebra and geometry to represent and solve problems involving direction and magnitude.

2. How can I determine if two vectors are orthogonal in Rn?

To determine if two vectors are orthogonal in Rn, you can use the dot product (also known as the scalar product). If the dot product of two vectors is equal to 0, then they are orthogonal. This is because the dot product measures the amount of overlap between the two vectors, and if they are perpendicular, there will be no overlap.

3. Can orthogonal vectors in Rn be of any dimension?

Yes, orthogonal vectors can exist in any dimension in Rn. In fact, the concept of orthogonality can be extended to spaces with any number of dimensions, not just three-dimensional space that we are familiar with.

4. How are orthogonal vectors used in real-world applications?

Orthogonal vectors have a wide range of applications in various fields such as physics, engineering, and computer graphics. For example, in physics, orthogonal vectors can be used to represent forces acting in different directions, while in computer graphics, they are used to create 3D models and animations.

5. Is it possible for three or more vectors to be orthogonal in Rn?

Yes, it is possible for three or more vectors to be orthogonal in Rn. In fact, in three-dimensional space, three mutually orthogonal vectors form the basis of a three-dimensional coordinate system. This concept can be extended to higher dimensions as well.

Similar threads

  • Differential Equations
Replies
5
Views
662
  • Differential Equations
Replies
1
Views
775
  • Special and General Relativity
Replies
11
Views
202
Replies
11
Views
1K
  • Calculus and Beyond Homework Help
Replies
5
Views
884
  • Calculus and Beyond Homework Help
Replies
9
Views
2K
  • Differential Equations
Replies
1
Views
757
  • Calculus and Beyond Homework Help
2
Replies
43
Views
3K
  • Calculus and Beyond Homework Help
Replies
12
Views
2K
  • Linear and Abstract Algebra
Replies
9
Views
210
Back
Top