Isometries: Translation, Rotation, Reflection

,

Pre-requisites

You should know trigonometry and basic highschool algebra.

Review of the Rectangular Coordinate System

We use a pair of numbers {x,y} to identify a point in the plane. The first number x in {x,y} indicates the horizontal position of the point, and the second number indicates height of the point in the plane. For example, {3,4} would be a point positioned at 3 units to the right and 4 units above. {-2,-5} would be a point 2 units to the left, and 5 units down.

The point {0,0} is very useful, we call it the Origin, and we denote the point O.

Direction of a Point

A point in the plane has a direction with respect to the origin. For example, going from {0,0} to {1,0} is heading east. Going from {0,0} to {0,-1} is heading west. Going from {0,0} to {1,1} is north east.

A point's direction is specified precisely by the angle it makes with the positive x-axis. For example, the point {1,0} would have angle 0, and the point {1,1} would have angle π/4, and point {-1,0} would have angle π.

Length of a Point

We can define a length for a point as the distance from the origin to the point.

From trigonometry, we know that the length of a point with coordinate {x,y} can be expressed as Sqrt[x^2+y^2].

Translation of a Point

Sometimes we like to move a whole image in the plane to a new location.

By Translation, we mean moving a point to a new position. For example, if we have a point A:={3,4}, and we want to move it 2 units to the right and 6 units up, then the new point would have a coordinates of {3+2, 4+6}.

We think of translation as addition of two points. For example, if A:={3,4} and B:={2,6}, then A+B=={5,10}.

Dilation of a Point

Sometimes we like to magnify or shrink a image in the plane. We call these processes together as dilation.

In the concept of magnification or shrink, there must be a center. So, when we do dilation, we choose the origin as the center of dilation. A dilation can be specified by by a scaling factor.

If we have a point A:={a,b}, and we want a new point B, such that B is in the same direction of vector A, but is longer by a factor of r, then the new point has the coordinate {r*a,r*b}.

Proof: This is so, because we know the length of {a,b} is Sqrt[a^2+b^2]. Now, the length and direction of the new point {r*a,r*b} is Sqrt[(r*a)^2+(r*b)^2], which simplifies to Sqrt[r^2*(a^2+b^2)] and r*Sqrt[a^2+b^2], which is just r times the length of A.

The direction of {a,b} is the direction of the new point {r*a,r*b} is ArcTan[(r*b)/(r*a)], which simplifies to ArcTan[b/a]. So we know that the direction is unchanged.

Suppose a point A has α angle with the positive x-axis, and the distance from A to origin is r. From trigonometry, we know that its coordinate would be {r*Cos[α],r*Sin[α]}.

Rotation

Sometimes we like to rotate a whole image of the plane around the origin. Given a point {a,b}, what's the coordinate of the point rotated by α around the origin?

Let's say point A has angle α. From trigonometry, we can write point A as {r*Cos[α],r*Sin[α]} where r is the length of A. The point A rotated around origin by β can be written as {r*Cos[α+β],r*Sin[α+β]}.

From trig identities, we can rewrite {r*Cos[α+β],r*Sin[α+β]} as {r*(Cos[α]*Cos[β]-Sin[α]*Sin[β]),r*(Sin[α]*Cos[β]+Cos[α]*Sin[β])}

Now, consider a point B, that has length 1 and angle β. B is then {Cos[β],Sin[β]}. Now if we consider the expression for point A rotated by β, we can see that it can be expressed simply as the coordinates of A and B. Namely, replace

Cos[α] by a
Sin[α] by b
Cos[β] by c
Sin[β] by d

and we see that a point {a1,a1} having α angle, rotated by β angle, can be expressed as {a*c - b*d, b*c + a*d}, where {c,d} is a point with β angle.

Definition of Complex Numbers

A Complex Number is defined as a pair {a,b}, where a and b are real numbers.

We will sometimes use capital letters such as A, B, C to denote complex numbers. For us, the complex numbers is just a point in the plane.

Definitions of Plus and Times functions

We define two functions that can be applied to complex numbers. Plus[A,B] and Times[A,B]. Each is a function that takes two complex numbers as arguments, and returns a complex number.

As a notation convenience, we write “complexPlus[A,B]” as “A⊕B”, and we write “complexTimes[A,B]” as “A⊗B”.

complexPlus[{a,b},{c,d}] := {a + c, b + d}

complexTimes[{a,b},{c,d}] := {a*c - b*d, b*c + a*d}

We also define a function complexInversion, which gives us the multiplicative inverse of a given argument. That is, complexInversion[A] has a value such that A⊗complexInversion[A]=={1,0}. We use a circle slash ⊘ as a shorthand operator notation for complexInversion. That is, complexInversion[Z] can be written as {1,0}⊘Z:

complexInversion[{a,b}] = 1/(a^2+b^2) ⊙ {a,-b}

and we also define a negative element function, with shorthand operator syntax of ⊖Z. So that,

addictiveInverse[{a,b}] = {-a,-b}

This definition of multiplication has important meaning when interpreted geometrically. Complex numbers can be thought as vectors, with a multiplication rule.

Here's a explanation: Suppose the distance from {c,d} to {0,0} is r, and suppose it makes an angle θ with the positive x-axis. Then, a complex number {a,b} multiplied by {c,d} by the definition {a*c - b*d, b*c + a*d}, is geometrically equivalent of rotating the point {a,b} by θ, then scale it by r.

Note: above we used the concept of distance and angle. These can be defined formally. Distance can be defined using the Pythagoras Theorem, and angle can be defined using trigonometry.

complex multiplication
Complex Multiplication. _z_complex_times.html

A rotation of θ around origin is expressed in complex number as multiplication by {Cos[θ],Sin[θ]}. In other words, a point {a,b} rotated by θ around the origin can be written as {a,b}⊗{Cos[θ],Sin[θ]}.

A translation by {a,b} is expressed as adding a complex number {a,b}.


Here's how to find products of rotation, using complex numbers.

A rotation of θ centered on A can be done by a sequence of:

 1. translation of -A,
 2. rotation of θ on origin,
 3. translate it back by A.

In formulas, we have:

 r[A,θ] == t[A]^-1 * r[0,θ] * t[A].
 
Expressed as complex numbers, the right hand side is

 R (Z - A) + A
  
where Z is the complex number to be acted on, and R is {Cos[θ],Sin[θ]}.

Expand and collect we have

 R Z + (-R+1) A

The R Z term is a rotation of Z, while the other term is a
translation. So, we have shown that a rotation on arbitrary center is
equivalent to a rotation on origin followed by a translation.

Similarly, a translation followed by a rotation on origin:

 t[A] * r[0,θ]

written in complex numbers

 (Z + A) R

is equal to

 Z R + A R

meaning that it is equivalent to a rotation on origin followed by a
translation.

A rotation θ on a point B followed by a translation is also
equivalent to a rotation on origin followed by a translation. Witness:

r[B,θ] * t[A]

Since rotation on an arbitrary point B is equivalent to rotation on
origin followed by a translation, as show above, so we can rewrite the
r[B,θ] to be

r[{0,0},α] * t[C]

for some α and C. Thus

r[B,θ] * t[A] == r[{0,0},α] * t[C] * t[A]

This shows that a rotation on a point followed by a translation is
just a rotation on origin followed by a translation.

The exact coordinate any product of symmetry can be easily calculated
by complex numbers.

The sequence of rotations
 r[{a1,a2},α] * r[{b1,b2},β]
applied to the point
 {x,y}

in complex numbers is then:

 (
    (  ({x,y} - {a1,a2}) * {Cos[α],Sin[α]} + {a1,a2}  )
    - {a1,a2}
 )
 * {Cos[β],Sin[β]}
 + {b1,b2}

In Mathematica, if we define the complex multiplication as cTimes, and
addition as cPlus:

 cPlus[{a_, b_}, {c_, d_}] := {a + c, b + d}
 cTimes[{a_, b_}, {c_, d_}] := {a*c - b*d, b*c + a*d}

then our sequence of rotations:

cPlus[cTimes[cPlus[cPlus[cTimes[cPlus[{x, y}, -{a1, a2}], {Cos[α],
Sin[α]}], {a1, a2}], -{b1, b2}], {Cos[β], Sin[β]}], {b1,
b2}]

Expands to:

{b1 + Cos[β]*(a1 - b1 + (-a1 + x)*Cos[α] - (-a2 +
y)*Sin[α]) - (a2 - b2 + (-a2 + y)*Cos[α] + (-a1 +
x)*Sin[α])*Sin[β], b2 + Cos[β]*(a2 - b2 + (-a2 +
y)*Cos[α] + (-a1 + x)*Sin[α]) + (a1 - b1 + (-a1 +
x)*Cos[α] - (-a2 + y)*Sin[α])*Sin[β]}

blog comments powered by Disqus
2006-06