The 3blue1brown channel is a treasure trove of great teaching. I stumbled upon his playlist on linear algebra a few months ago while dabbling in graphics. It is hard to overstate how helpful it was. In a few videos he is able to effectively answer the question "why linear algebra?" and give the watcher a really good intuition for thinking about the subject. I'm considering going back into a computer graphics stint, so I'm re-watching the series and taking notes this time.

Table of contents:

- Vectors, what even are they?
- Linear combinations, span and basis vectors
- Linear transformations and matrices

"The introduction of numbers as coordinates is an act of violence" -Hermann Weyl

The mathematics, computer science and physics students usually think of vectors in slightly different ways.

The physics student thinks of a vector as an arrow in space with a direction and magnitude.

The computer science student thinks of a vector as an ordered list of numbers.

The mathematician thinks of vectors in a more abstract way; as entities where there is a sensible notion of adding two vectors and multiplying a vector by a number.

For now though, we will think of a vector as a combination of the physics and computer science perspective.

"A vector is an arrow inside a coordinate system with its tail sitting at the origin. This arrow is also a list of numbers! In a two dimensional coordinate system for example, this list contains the instructions for how to get from the tail of the arrow to its tip."

Back to mathematician's view for a second. Interestingly, every topic in linear algebra will center around vector addition and multiplication by numbers. How do we think about these two operations from our perspective?

**Vector addition:** Each vector represents a certain "movement"; a step in a certain direction and distance in space. If you "move" along the first vector, then "move" along the second, you will end up in a location that can also be represented by one "movement". This "movement" is the sum of both vectors.
Notice that this parallels adding numbers on a number line. 2 + 5 means moving two steps to the right, then 5 steps to the right. The result is also moving 7 steps to the right.
Numerically, in terms of a list of numbers, we are simply adding all the rightward motions, then all the vertical motions.

**Multiplication by a number:** If you multiply a vector by the number 3, you are stretching out that vector so that it is three times longer than when you started. When you multiply the vector by a number between 0 and 1, you are squishing it down.
If you multiply it by a negative number, the same applies except you also switch its direction first before squishing/stretching. This process is called "scaling", which is why the numbers that act like this are called "scalars".
Numerically, you multiply each of the instructions in the list by that scalar.

Even though we will stick with the combination of the computer science and physics student perspectives for now, we will come back to the mathematics student's view later. The usefulness of linear algebra has less to do with which view you take, and more with the ability to switch between those views. It gives someone who is number crunching a way to visualize what he is doing visually, and it gives the visual creator a way to represent what she is trying to do mathematically. Win - Win.

"Mathematics requires a small dose, not of genius, but of an imaginative freedom which, in larger a larger dose, would be insanity." - Angus K. Rodgers

In the standard x-y coordinate system, there are two important vectors, the basis vectors:

`i`

the vector that points in x direction with length of 1`j`

the vector that points in the y direction with a length of 1

So we can think of the vectors as a list of scalars that scale the basis vectors. Or as the sum of the scaled basis vectors. Or as a *linear combination* of the basis vectors.

- This means that a vector will "look" different depending on what basis vectors it is scaling.
- This also means the choice of basis vectors is arbitrary and we could chose other ones, which would give us a different coordinate system.

The set of all possible vectors you can reach with a linear combination of a pair of vectors is called the *span* of those two vectors.

We say that the *span* of the basis vectors is all of two dimensional space, because we can reach all of two dimensional space by applying different linear combinations to the basis vectors.

The span of almost every pair of vectors will also be all of two dimensional space unless the vectors are on the same line in which case the span is that line. Remember, a linear combination is a sum of two scaled vectors and scaling stretches or squishes vectors.

We say a set of vectors is *linearly dependent* when the span does not change if you remove one of them. Said in another way, one of those vectors is not adding any new reachable vectors to the the span. Said yet another way, one of the vectors can be expressed as a linear combination of the others, since it is already in the span of the others.

A set of vectors is *linearly independent* when you can't remove any of them without also reducing the span. Each vector introduces some new reachable vectors to the span.

"Whos the monster that taught me matrices but forgot to tell me about their origin and beauty" - Alex Sere, Youtube Commenter.

Linear transformations are a way to move around space in a way that keeps the grid lines parallel and equally spaced, keeping the origin fixed. These transformations can be described using only the coordinates of where the basis vectors end up after the transformation.

Matrices give us a way to represent these transformations by having the matrix columns represent the coordinates where the basis vectors end up.

This means:

- A linear transformation is completely described by the final position of the basis vectors.
- Given a matrix with the final position of the basis vectors,
*you can find where any other vector ended up*because like we said earlier,vectors can be thought of as the sum of scaled basis vectors. The rows in the vectors represent the scalars by which the basis vectors are scaled.