Skip to content

Cheatsheet: What vectors actually are

The three views (one object, three angles)

Section titled “The three views (one object, three angles)”
ViewA vector is…Strength
PhysicsAn arrow with length and directionIntuition; “how far and which way”
Computer scienceAn ordered list of numbers, e.g. [3, 4]Computation; scales to any dimension
MathAnything you can add and scale coherentlyGenerality; covers functions, polynomials, model states

The arrow view and the list view are connected by a coordinate system: lay down axes, and every arrow gets a unique list (and back). Numbers for the computer, arrows for your head, the same object underneath.

OperationNumeric ruleGeometric meaning
AdditionAdd component by componentTip-to-tail: walk the first arrow, then the second from its tip
Scalar multiplicationMultiply every component by one numberStretch (>1), squish (<1), or flip (negative)

Everything else in linear algebra (linear combinations, span, matrix transformations) is built from exactly these two moves.

Addition:

[ 1 ] [ 3 ] [ 4 ]
[ 2 ] + [ 1 ] = [ 3 ]

Scaling:

[ 3 ] [ 6 ] [ 4 ] [ 2 ] [ 2 ] [ -2 ]
2 · [ 1 ] = [ 2 ] 0.5 · [ 2 ] = [ 1 ] (-1) · [ 1 ] = [ -1 ]
(stretch) (squish) (flip)
SymbolMeaning
[3, 4] or column of 3 over 4A vector with components 3 and 4
Column vector as matrixAn N by 1 matrix; in NumPy a nested array like [[3], [4]]
Component / coordinateOne entry in the list; says how far along one axis
DimensionHow many components (length of the list)
ScalarA plain number used to scale a vector

Coordinates describe the vector in a chosen coordinate system. Different axes give the same arrow different coordinates. The vector is the object; the coordinates are one description of it.

  • Arrow must start somewhere specific. No. A vector is a displacement; same length and direction means same vector, wherever drawn. Rooted at the origin by convention only.
  • Vector equals its coordinates. No. Coordinates are one frame’s description; change the frame and the numbers change while the vector does not.
  • Addition is concatenation. No. [1, 2] + [3, 4] = [4, 6], not [1, 2, 3, 4]. Add matching components; same dimension in and out.
  • High dimensions are mystical. No. A 300-D vector is a list of 300 numbers. Undrawable, but every 2D rule applies unchanged. In ML, dimensions usually stand for features or learned properties, not physical directions.
  • A vector must be an arrow or a list. No. The general definition is “anything you can add and scale coherently.” Arrows and lists are the two common examples.
  • Vector: an object you can add and scale; commonly pictured as an arrow or stored as a list of numbers.
  • Component / coordinate: a single entry in the vector’s list, measured along one axis.
  • Dimension: the number of components.
  • Scalar: a number whose role is to stretch, squish, or flip a vector without rotating it.
  • Tip-to-tail: the geometric picture of addition, placing the second arrow’s tail at the first arrow’s tip.

A vector is an arrow you can compute with, a list of numbers you can picture, and anything you can add and scale coherently. Three faces of one object, the smallest unit everything else in this series builds on.