ELIAS EBNER

  • Home
  • Blog
  • Courses
  • About

Linear Algebra

  1. Vectors
    1. Scalars
    2. Vectors vs Sets
    3. Addition and Subtraction
    4. Scalar Multiplication
    5. Zero Vectors
    6. Linear Combinations
    7. Real Dot Product
    8. Length of a Vector
    9. Orthogonal Vectors
    10. Parallel Vectors
    11. Angle Between Vectors
    12. Unit Vectors
  2. Matrices
    1. Notation
    2. Indexing
    3. Submatrices
    4. Matrix-by-Vector Product
    5. Addition and Subtraction
    6. Scalar Multiplication
    7. Transpose
    8. Symmetries
    9. Matrix Multiplication
    10. Identity Matrix
    11. Non-Negative Integer Powers
    12. Reverse Order Law of Transposition
  3. Linear Systems
    1. Inverse Matrices
    2. Singular Matrices
    3. Linear Dependence
    4. Solutions
  4. Planes
    1. Vector Cross Product
  5. Gaussian Elimination
Linear Algebra β€ΊPlanes

Planes

In the context of linear algebra it’s important to have some knowledge about planes.

The general formula for a plane in 3D space is

ax+by+cz=d.ax + by + cz = d.ax+by+cz=d.

Derivation

We will derive this formula in a few simple steps using some of the knowledge acquired so far. We start with some assumptions:

We suppose that we have…

  • …a known point P0=(x0,y0,z0)P_0 = (x_0,y_0,z_0)P0​=(x0​,y0​,z0​) which lies in the plane.

  • …a point P=(x,y,z)P = (x,y,z)P=(x,y,z) which represents any possible point in the plane.

  • …a known vector that is orthogonal to the plane denoted by nβƒ—=(a,b,c)\vec{n} = (a,b,c)n=(a,b,c). We refer to this vector as the normal vector.

We then define the vectors rβƒ—0\vec{r}_0r0​ and rβƒ—\vec{r}r to be the position vectors pointing to the points P0P_0P0​ and PPP respectively.

A position vector of a point PPP is the vector from the origin to PPP.

Also, we know that the vector rβƒ—βˆ’rβƒ—0\vec{r} - \vec{r}_0rβˆ’r0​ lies inside of the plane, since PPP and P0P_0P0​ - which are the points pointed to by the vectors rβƒ—\vec{r}r and rβƒ—0\vec{r}_0r0​ respectively - both lie in the plane as well. The vector rβƒ—βˆ’rβƒ—0\vec{r} - \vec{r}_0rβˆ’r0​ is the vector which starts at P0P_0P0​ and points to PPP.

Since nβƒ—\vec{n}n is orthogonal to the plane, and rβƒ—βˆ’rβƒ—0\vec{r} - \vec{r}_0rβˆ’r0​ lies in the plane, nβƒ—\vec{n}n must also be orthogonal to rβƒ—βˆ’rβƒ—0\vec{r} - \vec{r}_0rβˆ’r0​.

If you recall from the lesson about orthogonal vectors, we know that the dot product of two orthogonal vectors is 000.

Therefore

nβƒ—β‹…(rβƒ—βˆ’rβƒ—0)=0.\vec{n} \cdot (\vec{r} - \vec{r}_0) = 0.nβ‹…(rβˆ’r0​)=0.

We can rearrange this as

nβƒ—β‹…(rβƒ—βˆ’rβƒ—0)=0nβƒ—β‹…rβƒ—βˆ’nβƒ—β‹…rβƒ—0=0nβƒ—β‹…rβƒ—=nβƒ—β‹…rβƒ—0.\begin{align*} \vec{n} \cdot (\vec{r} - \vec{r}_0) &= 0 \\ \vec{n} \cdot \vec{r} - \vec{n} \cdot \vec{r}_0 &= 0 \\ \vec{n} \cdot \vec{r} &= \vec{n} \cdot \vec{r}_0. \end{align*}nβ‹…(rβˆ’r0​)nβ‹…rβˆ’nβ‹…r0​nβ‹…r​=0=0=nβ‹…r0​.​

We now write the dot products in coordinate form:

nβƒ—β‹…rβƒ—=nβƒ—β‹…rβƒ—0[abc]β‹…[xyz]=[abc]β‹…[x0y0z0]ax+by+cz=ax0+by0+cz0.\begin{align*} \vec{n} \cdot \vec{r} &= \vec{n} \cdot \vec{r}_0 \\ \begin{bmatrix} a \\ b \\ c \end{bmatrix} \cdot \begin{bmatrix} x \\ y \\ z \end{bmatrix} &= \begin{bmatrix} a \\ b \\ c \end{bmatrix} \cdot \begin{bmatrix} x_0 \\ y_0 \\ z_0 \end{bmatrix} \\ ax + by + cz = ax_0 + by_0 + cz_0. \end{align*}nβ‹…r​abc​​⋅​xyz​​ax+by+cz=ax0​+by0​+cz0​.​=nβ‹…r0​=​abc​​⋅​x0​y0​z0​​​​

Note that ax0ax_0ax0​, by0by_0by0​, and cz0cz_0cz0​ are all known values, so usually it is said that

d=ax0+by0+cz0,d = ax_0 + by_0 + cz_0,d=ax0​+by0​+cz0​,

and then the above equation can be rewritten as

ax+by+cz=d.ax + by + cz = d.ax+by+cz=d.

And in fact, this is just the coordinate form of the equation, and our starting point

nβƒ—β‹…(rβƒ—βˆ’rβƒ—0)=0,\vec{n} \cdot (\vec{r} - \vec{r}_0) = 0,nβ‹…(rβˆ’r0​)=0,

is the vector form of the same idea.

The vector form has a straightforward geometric interpretation. It’s basically stating that we are interested in all points PPP such that the vector rβƒ—βˆ’rβƒ—0\vec{r} - \vec{r}_0rβˆ’r0​ - the vector from P0P_0P0​ to PPP - is orthogonal to the normal vector nβƒ—\vec{n}n.

Now if you notice, aaa, bbb, and ccc are the components of nβƒ—\vec{n}n. In fact, if you’re asked to find the normal (orthogonal) vector to the plane, and the plane is given to you in the form ax+by+cz=dax + by + cz = dax+by+cz=d, all you have to do is collect the components of nβƒ—\vec{n}n into a vector, and you’re done.

For example, take the plane

3x+2yβˆ’1z=8.3x + 2y - 1z = 8.3x+2yβˆ’1z=8.

Just from looking at this, I know that the vector orthogonal to the plane is

nβƒ—=[32βˆ’1]\vec{n} = \begin{bmatrix} 3 \\ 2 \\ -1 \end{bmatrix}n=​32βˆ’1​​

Finding the Equation of the Plane From Three Points

Before we can get into the explanation itself, we must first go on a small digression.

Vectors From Points

When we want to express the idea of a vector which points from some point AAA to some other point BBB, we can use the notation AB→\overrightarrow{AB}AB.

The vector AB→\overrightarrow{AB}AB can be computed as

ABβ†’=rβƒ—Bβˆ’rβƒ—A,\overrightarrow{AB} = \vec{r}_B - \vec{r}_A,AB=rBβ€‹βˆ’rA​,

where rβƒ—B\vec{r}_BrB​ and rβƒ—A\vec{r}_ArA​ are the location vectors of the points BBB and AAA respectively.

The location vectors are simply vectors which point to some point. For example, the location vector of the point C=(1,2,3)C = (1,2,3)C=(1,2,3) is

rβƒ—C=[123].\vec{r}_C = \begin{bmatrix} 1 \\ 2 \\ 3 \end{bmatrix}.rC​=​123​​.

Back to Our Main Endeavor

It’s easy to find the equation of a plane when given one point and the normal vector, since we know that

d=ax0+by0+cz0,d = ax_0 + by_0 + cz_0,d=ax0​+by0​+cz0​,

and the left-hand side only needs aaa, bbb, and ccc. These are all values which come from the information we have directly.

But if we’re only given three points, there are some extra steps involved.

We can use one point for our x0x_0x0​, y0y_0y0​, and z0z_0z0​ values. We then combine the three points to find two vectors in the plane.

If we’re given three points

P1=(1,2,0),P2=(4,1,3),P3=(2,5,1).P_1 = (1,2,0), \quad P_2 = (4,1,3), \quad P_3 = (2,5,1).P1​=(1,2,0),P2​=(4,1,3),P3​=(2,5,1).

We now want to find two vectors lying in the plane. To do this, we compute the vectors P2P1βƒ—\vec{P_2 P_1}P2​P1​​ and P2P3βƒ—\vec{P_2 P_3}P2​P3​​ as described above:

P1P2β†’=[413]βˆ’[120]=[3βˆ’13],P3P2β†’=[413]βˆ’[251]=[2βˆ’42].\overrightarrow{P_1 P_2} = \begin{bmatrix} 4 \\ 1 \\ 3 \end{bmatrix} - \begin{bmatrix} 1 \\ 2 \\ 0 \end{bmatrix} = \begin{bmatrix} 3 \\ -1 \\ 3 \end{bmatrix}, \\ \overrightarrow{P_3 P_2} = \begin{bmatrix} 4 \\ 1 \\ 3 \end{bmatrix} - \begin{bmatrix} 2 \\ 5 \\ 1 \end{bmatrix} = \begin{bmatrix} 2 \\ -4 \\ 2 \end{bmatrix}.P1​P2​​=​413β€‹β€‹βˆ’β€‹120​​=​3βˆ’13​​,P3​P2​​=​413β€‹β€‹βˆ’β€‹251​​=​2βˆ’42​​.

We now know that our normal vector must be perpendicular to both of these vectors, thus

P1P2β†’β‹…nβƒ—=0\overrightarrow{P_1 P_2} \cdot \vec{n} = 0P1​P2​​⋅n=0

and

P3P2β†’β‹…nβƒ—=0.\overrightarrow{P_3 P_2} \cdot \vec{n} = 0.P3​P2​​⋅n=0.

We have a linear system of equations:

{P1P2β†’β‹…nβƒ—=0P3P2β†’β‹…nβƒ—=0{[3βˆ’13]β‹…[abc]=0[2βˆ’42]β‹…[abc]=0{3aβˆ’b+3c=02aβˆ’4b+2c=0.\begin{align*} &\begin{cases} \overrightarrow{P_1 P_2} \cdot \vec{n} = 0 \\ \overrightarrow{P_3 P_2} \cdot \vec{n} = 0 \end{cases} \\ &\begin{cases} \begin{bmatrix} 3 \\ -1 \\ 3 \end{bmatrix} \cdot \begin{bmatrix} a \\ b \\ c \end{bmatrix} = 0 \\[2em] \begin{bmatrix} 2 \\ -4 \\ 2 \end{bmatrix} \cdot \begin{bmatrix} a \\ b \\ c \end{bmatrix}= 0 \\ \end{cases} \\ &\begin{cases} 3a - b + 3c = 0 \\ 2a - 4b + 2c = 0 \end{cases}. \end{align*}​{P1​P2​​⋅n=0P3​P2​​⋅n=0β€‹βŽ©βŽ¨βŽ§β€‹β€‹3βˆ’13​​⋅​abc​​=0​2βˆ’42​​⋅​abc​​=0​{3aβˆ’b+3c=02aβˆ’4b+2c=0​.​

To solve this in terms of ccc, we can use any method we want. I will employ substitution:

{3aβˆ’b+3c=02aβˆ’4b+2c=0{b=3a+3c2aβˆ’4(3a+3c)+2c=0{b=3a+3c2aβˆ’12aβˆ’12c+2c=0{b=3a+3cβˆ’10aβˆ’10c=0{b=0a=βˆ’c\begin{align*} &\begin{cases} 3a - b + 3c = 0 \\ 2a - 4b + 2c = 0 \end{cases} \\ &\begin{cases} b = 3a + 3c \\ 2a - 4(3a + 3c) + 2c = 0 \end{cases} \\ &\begin{cases} b = 3a + 3c \\ 2a - 12a - 12c + 2c = 0 \end{cases} \\ &\begin{cases} b = 3a + 3c \\ -10a - 10c = 0 \end{cases} \\ &\begin{cases} b = 0 \\ a = -c \end{cases} \\ \end{align*}​{3aβˆ’b+3c=02aβˆ’4b+2c=0​{b=3a+3c2aβˆ’4(3a+3c)+2c=0​{b=3a+3c2aβˆ’12aβˆ’12c+2c=0​{b=3a+3cβˆ’10aβˆ’10c=0​{b=0a=βˆ’c​​

We can choose any value for ccc, say c=1c = 1c=1, then

a=βˆ’1,b=0. a = -1, \quad b = 0.a=βˆ’1,b=0.

So now,

nβƒ—=[βˆ’101].\vec{n} = \begin{bmatrix} -1 \\ 0 \\ 1 \end{bmatrix}.n=β€‹βˆ’101​​.

For x0x_0x0​, y0y_0y0​, and z0z_0z0​, we can pick any one of the given points, say P1P_1P1​.

Our equation now becomes:

ax+by+cz=ax0+by0+cz0βˆ’1x+0y+1z=βˆ’1β‹…1βˆ’0β‹…1+1β‹…0βˆ’x+z=βˆ’1,\begin{align*} ax + by + cz &= ax_0 + by_0 + cz_0 \\ -1x + 0y + 1z &= -1 \cdot 1 - 0 \cdot 1 + 1 \cdot 0 \\ -x + z &= -1, \end{align*}ax+by+czβˆ’1x+0y+1zβˆ’x+z​=ax0​+by0​+cz0​=βˆ’1β‹…1βˆ’0β‹…1+1β‹…0=βˆ’1,​

which is the equation for our plane.

Checking the Solution

We can even by using the points given as our xxx, yyy, and zzz:

For P1P_1P1​:

x=1,y=2,z=0x = 1, \quad y = 2, \quad z = 0x=1,y=2,z=0 βˆ’x+z=βˆ’1βˆ’1+0=βˆ’1βˆ’1=βˆ’1.\begin{align*} -x + z &= -1 \\ -1 + 0 &= -1 \\ -1 &= -1. \\ \end{align*}βˆ’x+zβˆ’1+0βˆ’1​=βˆ’1=βˆ’1=βˆ’1.​

For P2P_2P2​:

x=4,y=1,z=3x = 4, \quad y = 1, \quad z = 3x=4,y=1,z=3 βˆ’x+z=βˆ’1βˆ’4+3=βˆ’1βˆ’1=βˆ’1.\begin{align*} -x + z &= -1 \\ -4 + 3 &= -1 \\ -1 &= -1. \\ \end{align*}βˆ’x+zβˆ’4+3βˆ’1​=βˆ’1=βˆ’1=βˆ’1.​

For P3P_3P3​:

x=2,y=5,z=1x = 2, \quad y = 5, \quad z = 1x=2,y=5,z=1 βˆ’x+z=βˆ’1βˆ’2+1=βˆ’1βˆ’1=βˆ’1.\begin{align*} -x + z &= -1 \\ -2 + 1 &= -1 \\ -1 &= -1. \\ \end{align*}βˆ’x+zβˆ’2+1βˆ’1​=βˆ’1=βˆ’1=βˆ’1.​

A Final Remark on the Normal Vector

It’s important to note that the normal vector cannot be all zeroes, so

n⃗≠0⃗.\vec{n} \ne \vec{0}.n=0.

This is because - as outlined in the lesson about orthogonal vectors - by definition, the zero vector 0βƒ—\vec{0}0 is orthogonal to all vectors, and it not uniquely determine a plane.

Previous
Solutions to a Linear System
Next
Vector Cross Product
This website does not collect personal data, does not use cookies, and does not perform any tracking.