Wednesday, February 1, 2017

Matrices

Vectors

Rotation Matrix
Proportional Scaling Matrix
Non Proportional Scaling Matrix
Shear Matrix
Reflect Matrix

Determinant of a Matrix

Lorentz Transform Matrix


Vectors


Vectors are a way to describe point locations with numbers. Vectors can be used to build simple shapes like a cube or just about any shape you can imagine.

We do lots of stuff to these vectors with matrix multiplication.  We can grow, shrink, spin, stretch, squeeze, tilt and flip these guys.



First we'll look at the things you can do to vectors on a plane with computer drawing programs like Adobe Illustrator. Below are some items from the Illustrator tool box that employ matrices.



Rotation Matrix



Rotating a polygon doesn't change it's area. The area remains the same. The determinant of this matrix is 1.



Proportional Scaling Matrix



Doubling size as well as height boosts a polygon's area by a factor of four. This determinant of this matrix is 4.



Non Proportional Scaling Matrix



This matrix stretches the width to twice the original and squeezes the height to half of what it was. Overall the area is unchanged. The determinant of this matrix is 1. However the determinant of a non proportional scaling matrix can be more or less than 1.



Shear or Skew Matrix


When I was using Macromedia Freehand, the graphics program called this transformation "skew". Then Adobe ate Macromedia and I was forced to use Adobe Illustrator. Illustrator calls it "shear".


This transformation transforms a horizontally aligned rectangle to a parallelogram with same base and height. Area remains unchanged. The determinant of this matrix is 1.


Flip Matrix



Making the first term in the main diagonal negative flips polygons about the y axis. Making the lower right term negative would flip polygons about the x axis.

Determinant is -1. Not sure what that means geometrically but absolute value of the area remains the same.

Illustrator Tool Box



Determinant of a Matrix


Below is a general 3x3 matrix multiplied by each of the usual basis vectors in 3-space.
Notice the first basis vector is transformed into the first column of the matrix, the 2nd basis vector is transformed into the second column, and so on.

The 3 basis vectors form edges of a unit cube. Multiplying each of the vertices of this unit cube by our general matrix, we get a parallelepiped with edges (a, b, c), (l, m, n) and (x, y, z).

Of course the volume of the unit cube is one cubic unit. To find the volume of the transformed parallelepiped we take the determinant of the matrix

There are some matrices that don't change the size or shape of the objects they transform. Rotation matrices, for example. These have a determinant of 1. Matrices that don't change the size but flip the chirality of an object (say, change a left shoe into a right shoe), have a determinant of -1.

Here is a nice vid on the determinant of a matrix.




Lorentz Transformation


In ordinary Euclidean space, a point (x, y, x)'s distance from the origin would be
sqrt(x2 + y2 + z2 ), a metric easily arrived at with the Pythagorean theorem.

But the time space manifold we dwell in is a little strange. The metric is
sqrt(-t2 + x2 + y2 + z2 ). One of these dimensions is not like the other one.

In ordinary Euclidean space, changing Point Of View (POV) entails a translation and/or a rotation. In our space time, changing POV entails a Lorentz Transformation.

Adam Zalcman did a nice job of portraying the Lorentz transformation as a matrix. Here is a screen capture from his physics stack exchange answer:


A Lorentz matrix for a 2 dimensional Minkowski space looks like this:


Above is our two dimensional Minkowski space. As they move through time inhabitants can move either right or left. The ship leaves earth in the present. A year later it has moved half a light year to the right. It is moving .5 c.

Pluggin .5 c into our Lorentz transform matrix we get:

Transforming our Minkowski space with this matrix we get:
The ship's world line has been shoved to the left. From the ship passengers' point of view, they aren't moving. Also they perceive .866 years have elapsed, not a full year. The earth's world line has also been shoved to the left. From the ship's POV the earth is moving .5 c to the left.

Note that the diagonals remain in the same place. Earth folks as well as ship passengers both perceive light photos to be traveling at 1 c (c is the speed of light).

While the transformation stretches along one diagonal, it also squeezes along another diagonal. So the area remains the same. Determinant of this matrix is one.

When I first saw the transformed coordinate system I was thinking "Wait a minute. Earth is now more than half a light year away and only .866 years have passed on the ship. Seems like earth is going more than .5 c.  My mistake was in using the word "now". What were simultaneous events from one frame are no longer simultaneous.


Note that from the ship's P.O.V. Earth's clock is running slower. This is possible because simultaneous events along worldlines change depending on which frame the viewer's in.


Here is an animation showing different transforms where the ship's speed varies from -.9 c to .9 c.


Winchell Chung of Atomic Rockets describes a scene from a Heinlein novel where a student asks:

     “Mr. Ortega, admitting that you can’t pass the speed of light, what would happen if the Star Rover got up close to the speed of light—and then the Captain suddenly stepped the drive up to about six g and held it there?”
     “Why, it would—No, let’s put it this way—” He broke off and grinned; it made him look real young. “See here, kid, don’t ask me questions like that. I’m an engineer with hairy ears, not a mathematical physicist.” He looked thoughtful and added, “Truthfully, I don’t know what would happen, but I would sure give a pretty to find out. Maybe we would find out what the square root of minus one looks like—from the inside.”

Let's take a look at world lines where one rocket is moving .5 c to the left and the other is moving .5 c to the right. At first glance it'd seem like the rocket moving to the left would be moving the speed of light with regard to the other rocket.


Transform the scene on the left to the orange ship's point of view. From the orange ship's P. O. V., the purple ship is moving .8 c to the left. The Lorentz transformation doesn't shift the purple ship all the way to the edge of the light cone. (Click on image to get a larger version).


From the point of view of each world line, his immediate neighbors are moving either .5 c to the left or .5 c to the right. The arrowhead on each line corresponds to the passage of one year from that world line's point of view. The horizontal line indicates simultaneous events from the P.O.V of the central world line after one year. These trace out a hyperbola with the edges of the light cone as asymptotes.

If the above lines were a cone, plane of simultaneous events would cut the cone along a circle and the world lines would pierce that circle in points closer and closer the edge as the world lines approached c. This would be a Poincare disk.

M. C. Escher's Circle Limit prints are based on Poincare disks.


Each angel or demon on this disk perceives themselves to be at the center while their neighbors shrinking towards the boundary of this world as they grow more distant. So it would be with Mr. Ortega's student who would step on the gas when he's moving .999 c. He'd just shift his position to the another part of the disk and he would be no closer to the edge.