Vector Magnitude: Norm, Distance & Unit Vectors

Magnitude in Two and Three Dimensions

Every arrow has a length — but a vector is given by its components, not by a length.

\mathbf{v} = (4, 3)

says "4 across, 3 up". How long is the arrow from its tail to its tip? The components do not say so directly.

For v = (4, 3) the components are the legs of a right triangle, and the arrow is its hypotenuse: length 5.

Going 4 across and then 3 up traces two sides of a right triangle; the arrow cuts straight across. With legs 4 and 3 the long side is

\sqrt{16 + 9} = 5

. The same works for any vector in the plane.

vector	legs	legs squared and added	length
(4, 3)	4 and 3	16 + 9 = 25	5
(1, 1)	1 and 1	1 + 1 = 2	√2 ≈ 1.414
(0, −2)	0 and 2	0 + 4 = 4	2
(5, 12)	5 and 12	25 + 144 = 169	13

The length is forced by the Pythagorean theorem, because the two components are always perpendicular.

This is a direct application of the Pythagorean theorem — the horizontal component

v_1

and the vertical component

v_2

form the two legs, and the vector itself is the hypotenuse.

In

\mathbb{R}^3

, the same reasoning applies in two stages. A vector

\mathbf{v} = (v_1, v_2, v_3)

can be decomposed into a component in the

xy

-plane with length

\sqrt{v_1^2 + v_2^2}

and a vertical component

v_3

. Applying the Pythagorean theorem a second time gives:

Magnitude in Two and Three Dimensions

\|\mathbf{v}\| = \sqrt{v_1^2 + v_2^2 + v_3^2}

The definition

The magnitude (or norm) of a vector

\mathbf{v}

, written

\|\mathbf{v}\|

, is its length: the square root of the sum of the squares of its components.

In

\mathbb{R}^2

, the magnitude of a vector

\mathbf{v} = (v_1, v_2)

is the length of the hypotenuse of the right triangle formed by its components:

Magnitude in Two and Three Dimensions

\|\mathbf{v}\| = \sqrt{v_1^2 + v_2^2}

So

\|(4, 3)\| = 5

.

The magnitude is a single number, never negative, that ignores direction.

For a concrete case in three dimensions, the vector

\mathbf{v} = (2, -1, 2)

has magnitude

\sqrt{2^2 + (-1)^2 + 2^2} = \sqrt{4 + 1 + 4} = \sqrt{9} = 3

. Every component enters the sum as a square, so the sign of the middle entry plays no part in the length.

The geometric intuition is the same in both cases: the magnitude is the straight-line distance from the tail of the vector to its tip. The only difference is the number of perpendicular components being combined.

Squares summed, root about to be taken

The squares of the components have been accumulated into a single number and the square root is the last step left. Every component contributes its square and nothing else, which is why the sign of a component cannot affect the length, and why the formula reads the same in two dimensions as in twenty. Run it on your own vector on the magnitude visualizer.

The same sum-of-squares pattern is what the next section generalises to any number of dimensions.

The square root goes over the whole sum, not over each term.

Pitfall: Taking the root term by term

Taking the root of each square separately,

\sqrt{3^2} + \sqrt{(-4)^2} = 3 + 4 = 7

, gives the wrong length for

(3, -4)

. The root applies to the whole sum:

\|(3, -4)\| = \sqrt{9 + 16} = \sqrt{25} = 5

.

Taking roots term by term just adds up the absolute values of the components, which is a different measure of size. The general norm carries the correct formula to

\mathbb{R}^n

.

To compute the magnitude of a specific vector, use the vector calculator.

The rest of the page covers norm notation, the general norm in

n

dimensions, its properties, distance between vectors, unit vectors and normalization.

Norm Notation

Notation

Norm Notation

Why the bars come in pairs, the subscripts that name a whole family of lengths, and the hat that promises length one.

Bold marking and

\mathbb{R}^n

— vector notation; single bars on scalars — absolute value.

\|\mathbf{v}\|

The norm of v; the magnitude of v

Double bars for vector length — the bar count is doing real work: one pair for a scalar's absolute value, two for a vector's norm, and the scalar-vector split is spelled out in The General Norm below. The mathematical keyboard carries the double pair as its own key, so the count is never a typing accident.

CasesThe scaling law mixes both in one line:

\|c\mathbf{v}\| = |c|\,\|\mathbf{v}\|

— single bars on the scalar, double on the vector, each pair applying its own operation.

Also written

|\mathbf{v}|

— single bars for magnitude, standard in physics and engineering texts; unambiguous there because bold already marks the vector, but it collides with the determinant's bars once matrices arrive.

Do not confuseParallel lines or “evaluated at” bars. The paired-delimiter reading is the only one — and nothing here is a fraction:

\|\mathbf{v}\|

wraps, it never divides.

\|\mathbf{v}\|_1 \; \|\mathbf{v}\|_2 \; \|\mathbf{v}\|_{\infty}

The one-norm, two-norm, and infinity-norm of v

The subscript selects a member of the norm family that The General Norm below names in words:

1

sums absolute values,

2

is the Euclidean default,

\infty

takes the largest component. An unsubscripted

\|\mathbf{v}\|

means the

2

-norm.

CasesThe same family carries the script-ell names

\ell^1, \ell^2, \ell^{\infty}

— the letter honours Lebesgue, and the script form exists because a plain

l

drowns among

1

and

I

in print.

Also written

\|\mathbf{v}\|_p

— the general member, exponent

p

inside the sum,

p

-th root outside.

Do not confuseA component index.

\|\mathbf{v}\|_2

is not “the norm of the second component” — on a norm, the subscript picks the measuring rule, not a slot.

\hat{\mathbf{u}}

u hat

The hat is a promise: whatever wears it has norm exactly

1

. Writing

\hat{\mathbf{u}}

asserts

\|\hat{\mathbf{u}}\| = 1

with no further words — the convention behind Unit Vectors below.

CasesThe coordinate directions carry two naming traditions: mathematics writes

\mathbf{e}_1, \mathbf{e}_2, \mathbf{e}_3

; physics writes

\hat{\imath}, \hat{\jmath}, \hat{k}

— dotless letters under the hats, same three vectors.

Also written

\mathbf{u}/\|\mathbf{u}\|

— the normalization recipe spelled out, as in Normalization below; the hat names its result.

Do not confuseAn estimator. In statistics the same hat means “estimate of” —

\hat{p}

has nothing to do with unit length; the glyph changes fields and changes jobs.

All linear algebra symbols →↑ Vectors

DownloadPNG2×PDF

Notation · Cheat sheet

Vector Notation at a Glance

The marks from the notation cards across the vector pages: how each is read and what it means. Every group links to the full card on its page.

Vectors

\mathbf{v}, \; \overset{\to}{v}

vector v

Bold in print, an arrow by hand: either marks a vector, not a number.

\overset{\longrightarrow}{AB}

vector A B

The vector from point

A

to point

B

.

\mathbf{i}, \mathbf{j}, \mathbf{k}

the standard basis

Unit vectors along the axes; also

\mathbf{e}_1, \ldots, \mathbf{e}_n

.

Basic operations

(v_1, \ldots, v_n)

v with components v one to v n

The subscript names a slot:

v_2

is the second component, a number.

\mathbb{R}^n

R n, real n-space

All vectors with

n

real components.

MagnitudeThis page

\|\mathbf{v}\|

the norm of v

Length; two bars for a vector, one for a number.

\|\mathbf{v}\|_1, \; \|\mathbf{v}\|_\infty

one-norm, infinity-norm

The subscript picks the norm;

\|\mathbf{v}\|_2

is the usual length.

\hat{\mathbf{u}}

u hat

A unit vector:

\|\hat{\mathbf{u}}\| = 1

.

Dot product

\mathbf{a} \cdot \mathbf{b}

a dot b

A number from two vectors;

\mathbf{a}\mathbf{b}

with no dot means nothing.

\mathbf{a}^{T}\mathbf{b}

a transpose b

The same number in matrix form: row times column.

Cross product

\mathbf{a} \times \mathbf{b}

a cross b

A vector from two vectors, in

\mathbb{R}^3

only.

\mathbf{a} \cdot (\mathbf{b} \times \mathbf{c})

a dot, b cross c

The scalar triple product: a signed volume.

Vector properties

\mathbf{a} \parallel \mathbf{b}

a is parallel to b

One is a scalar multiple of the other:

\mathbf{a} = c\mathbf{b}

.

\mathbf{a} \perp \mathbf{b}

a is orthogonal to b

Their dot product is

0

.

\mathbf{0}

the zero vector

Parallel and orthogonal to every vector.

Same marks, different job

|x|

vs

\|\mathbf{v}\|

One pair of bars for a number, two for the length of a vector.

\cdot

and

\times

Between numbers, plain multiplication; between vectors, the dot and cross products.

v_2

vs

\mathbf{v}_2

Plain: the second component of

\mathbf{v}

; bold: the second vector in a list.

Vector symbols table Vectors

Vector Notation at a Glance·/linear-algebra/vectorsLearn Math Class

The General Norm

The pattern from

\mathbb{R}^2

and

\mathbb{R}^3

extends without modification to any

\mathbb{R}^n

. For a vector

\mathbf{v} = (v_1, v_2, \ldots, v_n)

:

Vector Norm

\|\mathbf{v}\| = \sqrt{v_1^2 + v_2^2 + \cdots + v_n^2} = \sqrt{\sum_{i=1}^{n} v_i^2}

Learn more about this formula: Vector Norm →

Browse all linear algebra formulas →

This quantity is called the Euclidean norm, or equivalently the

\ell^2

norm. The superscript distinguishes it from other ways of measuring vector size — the

\ell^1

norm sums absolute values, the

\ell^\infty

norm takes the largest absolute component — but when "norm" appears without qualification in this section, it always refers to the Euclidean norm.

The double-bar notation

\|\mathbf{v}\|

is standard for the norm. Single bars

|x|

are reserved for the absolute value of a scalar. The distinction matters: magnitude is a property of vectors, absolute value is a property of numbers, and the notational separation keeps the two from being confused.

In dimensions beyond three, the formula cannot be visualized as a geometric length, but it retains the same algebraic role. It measures how far

\mathbf{v}

sits from the origin in the coordinate system of

\mathbb{R}^n

, generalizing the concept of length purely through algebra.

Fact: A unit cube with a diagonal of 10

The far corner

(1, 1, \ldots, 1)

lies

\sqrt{1 + 1 + \cdots + 1} = \sqrt{n}

from the origin, so the diagonal keeps growing with the dimension. In

\mathbb{R}^{100}

a cube whose edges all have length

1

has a diagonal of length

10

.

The vector whose components all equal

1

makes a convenient test case. In

\mathbb{R}^2

it runs from the origin to the far corner of the unit square, and in

\mathbb{R}^3

to the far corner of the unit cube. In

\mathbb{R}^n

it joins two opposite corners of the

n

-dimensional unit cube, every edge of which still has length

1

, and its norm comes from the same formula: one squared

1

for each of the

n

components.

Properties of the Norm

The Euclidean norm satisfies three properties that any reasonable notion of length should obey. Together, they are what make the norm a reliable measure of vector size.

Non-Negativity

Properties of the Norm

\|\mathbf{v}\| \geq 0, \quad \text{with equality if and only if } \mathbf{v} = \mathbf{0}

The sum of squared components is never negative, and its square root is zero only when every component is zero. Length is always non-negative, and the only vector with zero length is the zero vector itself.

Scaling

Norm Scaling Property

\|c\mathbf{v}\| = |c|\,\|\mathbf{v}\|

Learn more about this formula: Norm Scaling Property →

Browse all linear algebra formulas →

Multiplying a vector by a scalar

c

multiplies its length by

|c|

. The absolute value is necessary because a negative scalar reverses the direction without producing a negative length. This property connects the norm directly to scalar multiplication: scaling a vector by

c

does exactly what the geometric picture suggests — it stretches or compresses by

|c|

and possibly flips.

Triangle Inequality

\|\mathbf{a} + \mathbf{b}\| \leq \|\mathbf{a}\| + \|\mathbf{b}\|

Learn more about this formula: Triangle Inequality →

Browse all linear algebra formulas →

Pitfall: Adding the lengths

Taking

\|\mathbf{a} + \mathbf{b}\| = \|\mathbf{a}\| + \|\mathbf{b}\|

for

\mathbf{a} = (3, 4)

and

\mathbf{b} = (-3, 4)

gives

5 + 5 = 10

. The sum is

(0, 8)

, so

\|\mathbf{a} + \mathbf{b}\| = 8

.

The length of a sum never exceeds the sum of the individual lengths. Geometrically, in the tip-to-tail picture of vector addition, the direct path from start to finish (the sum) is never longer than the path that follows both vectors end to end. Equality holds only when one of

\mathbf{a}

,

\mathbf{b}

is

\mathbf{0}

or they point in the same direction — when the path is already straight. The triangle inequality is proved using the Cauchy–Schwarz inequality, which is developed on the dot product page.

DownloadPNG2×PDF

Vectors · magnitude

Properties of the norm

Three axioms define a norm; everything below them is derived. Watch the equality cases — each one identifies a geometric configuration, not just an edge case in the algebra.

06properties

Unconditional4

ii

Absolute homogeneity§ 3

\|c\mathbf{v}\| = |c| \, \|\mathbf{v}\|

Scaling by c multiplies the length by |c|, not by c.

Unconditional

Requires: any scalar $c$
Common error: dropping the absolute value — $\|{-2}\mathbf{v}\| = 2\|\mathbf{v}\|$ , not $-2\|\mathbf{v}\|$

Scaling by $c$ multiplies the length by $|c|$ , not by $c$ . The absolute value is doing real work: for $c > 0$ the direction is preserved, for $c < 0$ it reverses, and in both cases the length grows by the same factor. A negative length would violate the axiom above.

iii

Triangle inequality§ 3

\|\mathbf{u} + \mathbf{v}\| \leq \|\mathbf{u}\| + \|\mathbf{v}\|

The direct path is never longer than the tip-to-tail path.

Unconditional

Requires: any $\mathbf{u}, \mathbf{v}$ ; equality iff one is $\mathbf{0}$ or $\mathbf{u}$ and $\mathbf{v}$ point the same way

The direct path is never longer than the tip-to-tail path. Equality holds exactly when one vector is a non-negative multiple of the other — the triangle degenerates to a line. This is the axiom that makes distance a metric, and it follows from the Cauchy–Schwarz inequality.

Both extremesu = (3, 0), v = (4, 0): ‖7‖ = 7 = 3 + 4 equality
u = (3, 0), v = (0, 4): ‖5‖ = 5 < 3 + 4 strict

iv

Reverse triangle inequalitypitfalls

\bigl|\, \|\mathbf{u}\| - \|\mathbf{v}\| \,\bigr| \leq \|\mathbf{u} - \mathbf{v}\|

A lower bound to go with the upper one.

Unconditional

Requires: derived, not assumed

A lower bound to go with the upper one. It says two vectors of very different lengths cannot be close together — which is what makes the norm a continuous function, and is used constantly in analysis.

v

Norm and the dot product§ 2

\|\mathbf{v}\|^2 = \mathbf{v} \cdot \mathbf{v}

Length is the square root of the self-dot product.

Unconditional

Requires: the Euclidean norm specifically

Length is the square root of the self-dot product. This is the bridge between the two: positive definiteness of the dot product is the same statement as non-negativity of the norm, seen from the other side.

Conditional2

i

Non-negativity§ 3

\|\mathbf{v}\| \geq 0

Length is never negative, and only the zero vector has zero length.

Conditional

Holds when: equality iff $\mathbf{v} = \mathbf{0}$

Length is never negative, and only the zero vector has zero length. The second half is the load-bearing part — without it a nonzero vector could have length zero and normalization would break, since dividing by the magnitude assumes it is nonzero.

vi

Unit vector§ 5

\|\hat{\mathbf{v}}\| = 1

where

\hat{\mathbf{v}} = \mathbf{v} / \|\mathbf{v}\|

Immediate from absolute homogeneity with c = 1/\|v\|.

Conditional

Holds when: $\mathbf{v} \neq \mathbf{0}$
Fails when: $\mathbf{v} = \mathbf{0}$ — division by zero

Immediate from absolute homogeneity with $c = 1/\|\mathbf{v}\|$ . The zero vector cannot be normalised, which is the same fact as it having no direction.

Unconditional

Conditional

Non-negativity§ 3Conditional

\|\mathbf{v}\| \geq 0

Holds when: equality iff $\mathbf{v} = \mathbf{0}$

Length is never negative, and only the zero vector has zero length. The second half is the load-bearing part — without it a nonzero vector could have length zero and normalization would break, since dividing by the magnitude assumes it is nonzero.

Read the full section

Absolute homogeneity§ 3Unconditional

\|c\mathbf{v}\| = |c| \, \|\mathbf{v}\|

Requires: any scalar $c$
Common error: dropping the absolute value — $\|{-2}\mathbf{v}\| = 2\|\mathbf{v}\|$ , not $-2\|\mathbf{v}\|$

Scaling by

c

multiplies the length by

|c|

, not by

c

. The absolute value is doing real work: for

c > 0

the direction is preserved, for

c < 0

it reverses, and in both cases the length grows by the same factor. A negative length would violate the axiom above.

Read the full section

Triangle inequality§ 3Unconditional

\|\mathbf{u} + \mathbf{v}\| \leq \|\mathbf{u}\| + \|\mathbf{v}\|

Requires: any $\mathbf{u}, \mathbf{v}$ ; equality iff one is $\mathbf{0}$ or $\mathbf{u}$ and $\mathbf{v}$ point the same way

The direct path is never longer than the tip-to-tail path. Equality holds exactly when one vector is a non-negative multiple of the other — the triangle degenerates to a line. This is the axiom that makes distance a metric, and it follows from the Cauchy–Schwarz inequality.

Both extremesu = (3, 0), v = (4, 0): ‖7‖ = 7 = 3 + 4 equality
u = (3, 0), v = (0, 4): ‖5‖ = 5 < 3 + 4 strict

Read the full section

Reverse triangle inequalitypitfallsUnconditional

\bigl|\, \|\mathbf{u}\| - \|\mathbf{v}\| \,\bigr| \leq \|\mathbf{u} - \mathbf{v}\|

Requires: derived, not assumed

A lower bound to go with the upper one. It says two vectors of very different lengths cannot be close together — which is what makes the norm a continuous function, and is used constantly in analysis.

Read the full section

Norm and the dot product§ 2Unconditional

\|\mathbf{v}\|^2 = \mathbf{v} \cdot \mathbf{v}

Requires: the Euclidean norm specifically

Length is the square root of the self-dot product. This is the bridge between the two: positive definiteness of the dot product is the same statement as non-negativity of the norm, seen from the other side.

Read the full section

Unit vector§ 5Conditional

\|\hat{\mathbf{v}}\| = 1

where

\hat{\mathbf{v}} = \mathbf{v} / \|\mathbf{v}\|

Holds when: $\mathbf{v} \neq \mathbf{0}$
Fails when: $\mathbf{v} = \mathbf{0}$ — division by zero

Immediate from absolute homogeneity with

c = 1/\|\mathbf{v}\|

. The zero vector cannot be normalised, which is the same fact as it having no direction.

Read the full section

Any function satisfying the first three is a norm, and the Euclidean length is only one of them. The taxicab and maximum norms satisfy all three while measuring distance differently — which is why the axioms are stated separately from the square-root formula that happens to satisfy them.

Properties of the norm·/linear-algebra/vectors/magnitudeLearn Math Class

Three rules make a measure a length: zero only at the zero vector, scaling by

|c|

, and no straight path longer than a detour.

Distance Between Vectors

The norm provides a natural way to measure how far apart two vectors are. The distance between

\mathbf{a}

and

\mathbf{b}

is defined as the norm of their difference:

Distance Between Vectors

d(\mathbf{a}, \mathbf{b}) = \|\mathbf{a} - \mathbf{b}\| = \sqrt{(a_1 - b_1)^2 + (a_2 - b_2)^2 + \cdots + (a_n - b_n)^2}

Learn more about this formula: Distance Between Vectors →

Browse all linear algebra formulas →

When

\mathbf{a}

and

\mathbf{b}

are interpreted as position vectors — arrows from the origin to points in space — this formula gives the straight-line distance between those points. In

\mathbb{R}^2

, it reduces to the familiar distance formula from coordinate geometry. In

\mathbb{R}^3

, it extends to three-dimensional Euclidean distance. In higher dimensions, the algebraic expression remains identical even though the geometry can no longer be drawn.

Pitfall: Subtracting the lengths

Computing

\|\mathbf{a}\| - \|\mathbf{b}\|

for

\mathbf{a} = (3, 4)

and

\mathbf{b} = (4, 3)

gives

5 - 5 = 0

, as if the two points coincided. The distance is the length of the difference:

\|(-1, 1)\| = \sqrt{2}

.

Distance inherits its behavior from the norm. It is always non-negative, equals zero only when the two vectors are identical, is symmetric (

d(\mathbf{a}, \mathbf{b}) = d(\mathbf{b}, \mathbf{a})

, since

\|\mathbf{a} - \mathbf{b}\| = \|\mathbf{b} - \mathbf{a}\|

), and satisfies its own triangle inequality:

d(\mathbf{a}, \mathbf{c}) \leq d(\mathbf{a}, \mathbf{b}) + d(\mathbf{b}, \mathbf{c})

. These properties make the Euclidean distance a metric — a formal measure of separation on

\mathbb{R}^n

.

Unit Vectors

A unit vector is a vector whose norm equals exactly

1

:

Unit Vectors

\|\hat{\mathbf{u}}\| = 1

A unit vector retains all the directional information of the original vector while discarding any notion of scale. It answers the question "which way?" without saying "how far."

The standard basis vectors are the most immediate examples. In

\mathbb{R}^3

:

Unit Vectors

\mathbf{e}_1 = (1, 0, 0), \quad \mathbf{e}_2 = (0, 1, 0), \quad \mathbf{e}_3 = (0, 0, 1)

Fact: In the taxicab norm,

\pi

is

4

Measured by

\|\mathbf{v}\|_1 = |v_1| + |v_2|

, the unit circle is a square with corners

(\pm 1, 0)

and

(0, \pm 1)

. Each side has taxicab length

1 + 1 = 2

, so circumference over diameter is

\frac{8}{2} = 4

.

Each has exactly one component equal to

1

and the rest equal to

0

, giving a norm of

1

. These unit vectors point along the coordinate axes, and any vector can be decomposed as a linear combination of them:

\mathbf{v} = v_1\mathbf{e}_1 + v_2\mathbf{e}_2 + v_3\mathbf{e}_3

.

But unit vectors are not confined to the coordinate directions. Any direction in

\mathbb{R}^n

has a corresponding unit vector. The vector

\left(\frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)

is a unit vector in

\mathbb{R}^2

pointing at

45°

to both axes. The set of all unit vectors in

\mathbb{R}^2

traces the unit circle; in

\mathbb{R}^3

, the unit sphere.

Unit vectors in every direction

Five unit vectors, each drawn from the origin. Two lie along the axes, one at 45°, two in less obvious directions — for (−0.6, 0.8), the squares add as 0.36 + 0.64 = 1. Whatever the direction, a length of exactly 1 puts the tip on the circle of radius 1, so the unit vectors of the plane and the points of the unit circle are the same thing.

Choosing a unit vector is choosing a direction and nothing else, which is exactly what normalization below extracts.

Normalization

Normalization is the process of converting any nonzero vector into the unit vector that shares its direction. Given

\mathbf{v} \neq \mathbf{0}

, the normalized form is:

Normalization

\hat{\mathbf{v}} = \frac{\mathbf{v}}{\|\mathbf{v}\|} = \frac{1}{\|\mathbf{v}\|}(v_1, v_2, \ldots, v_n)

Learn more about this formula: Normalization →

Browse all linear algebra formulas →

This divides every component by the same positive scalar

\|\mathbf{v}\|

, which is a special case of scalar multiplication by

\frac{1}{\|\mathbf{v}\|}

. The scaling property of the norm guarantees that the result has magnitude

1

:

Normalization

\left\|\frac{\mathbf{v}}{\|\mathbf{v}\|}\right\| = \frac{\|\mathbf{v}\|}{\|\mathbf{v}\|} = 1

Pitfall: Dividing by the squared length

Dividing

\mathbf{v} = (1, 2, 2)

by

1 + 4 + 4 = 9

gives

\left(\frac{1}{9}, \frac{2}{9}, \frac{2}{9}\right)

, whose length is

\frac{1}{3}

, not

1

. Dividing by

\|\mathbf{v}\| = \sqrt{9} = 3

gives

\hat{\mathbf{v}} = \left(\frac{1}{3}, \frac{2}{3}, \frac{2}{3}\right)

.

In a computation, normalization comes down to two steps. The length is found first, from the components of

\mathbf{v}

, and then every component is divided by that one number. The identity above doubles as a check on the result: the squares of the new components must add up to exactly

1

.

Because the scalar

\frac{1}{\|\mathbf{v}\|}

is positive, the direction of

\mathbf{v}

is preserved — the normalized vector

\hat{\mathbf{v}}

points exactly the same way as

\mathbf{v}

, just with unit length.

RULE

The zero vector cannot be normalized.

It has

\|\mathbf{0}\| = 0

, and dividing by zero is undefined. This is consistent with the geometric fact that the zero vector carries no directional information — there is no "way" it points, so extracting a direction from it is impossible.

Normalization appears throughout the rest of this section. The angle formula on the dot product page involves the quantity

\frac{\mathbf{a} \cdot \mathbf{b}}{\|\mathbf{a}\|\|\mathbf{b}\|}

, which is the dot product of the two normalized vectors

\hat{\mathbf{a}} \cdot \hat{\mathbf{b}}

. Orthogonal projection decomposes a vector relative to a direction, and that direction is specified most cleanly through a unit vector. Whenever a formula isolates the directional component of a vector, normalization is doing the work underneath.

Dividing through by the norm

Each component is being divided by the length just computed, so the direction survives untouched while the size collapses to exactly one. This is why normalisation fails for the zero vector alone — it is the single case with no length to divide by and no direction to preserve. Normalise a vector of your choosing on the magnitude visualizer.

A normalised vector is therefore a pure direction, which is what makes it the natural building block for bases and projections later on.

Summary: Norm and the Concepts Built from It

The norm is the foundation of three related concepts this page has developed: distance (the norm of a difference), unit vector (a vector with norm exactly 1), and normalization (rescaling any nonzero vector to unit length). The table below collects each alongside its definition, formula, and direct relationship to the norm ‖v‖ — showing the family tree of ideas that grow out of a single length measurement.

Concept	Definition	Formula	Built from the norm by
Norm ‖v‖	length of v	√(v₁² + v₂² + ⋯ + vₙ²)	(the base concept — everything below is derived from this)
Distance d(a, b)	straight-line distance between two vectors viewed as points	‖a − b‖ = √(Σ(aᵢ − bᵢ)²)	applying the norm to the difference a − b
Unit vector û	a vector that carries pure direction with no scale	‖û‖ = 1	requiring the norm to equal exactly 1
Normalization	turn any nonzero vector into the unit vector pointing the same way	v̂ = v / ‖v‖	dividing v by its own norm (undefined for v = 0)

Common Pitfalls

Four mistakes account for most wrong answers in computations with the norm. Each one is marked where it happens in the sections above; here each is worked through in full, with the wrong and the right working side by side.

Taking the root term by term

The magnitude is computed by taking the square root of each squared component on its own and adding the results, as if the root could be split across the sum.

Wrong working:

\|(3, -4)\| = \sqrt{3^2} + \sqrt{(-4)^2} = 3 + 4 = 7

Right working: add the squares first, then take one square root of the total.

\|(3, -4)\| = \sqrt{3^2 + (-4)^2} = \sqrt{9 + 16} = \sqrt{25} = 5

Why it happens: for non-negative numbers the root does split over a product,

\sqrt{ab} = \sqrt{a}\,\sqrt{b}

, and the same move is carried over to a sum, where it fails.

How to check: square the answer; it must equal the sum of the squared components. The wrong answer gives

7^2 = 49

, but

3^2 + (-4)^2 = 25

. The right answer gives

5^2 = 25

.

Adding the lengths

The norm is treated as if it passed through addition: the length of a sum is taken to be the sum of the two lengths.

Wrong working: for

\mathbf{a} = (3, 4)

and

\mathbf{b} = (-3, 4)

,

\|\mathbf{a} + \mathbf{b}\| = \|\mathbf{a}\| + \|\mathbf{b}\| = 5 + 5 = 10

Right working: add the vectors first, then measure the sum.

\mathbf{a} + \mathbf{b} = (0, 8), \qquad \|\mathbf{a} + \mathbf{b}\| = \sqrt{0^2 + 8^2} = 8

As the triangle inequality requires,

8 \leq 10

.

Why it happens: a scalar does come out of the norm,

\|c\mathbf{v}\| = |c|\,\|\mathbf{v}\|

, and addition is expected to behave the same way. It does only in the equality case of the triangle inequality.

How to check: the sum of lengths is exact only when one vector is a non-negative multiple of the other. Here

(-3, 4)

is not a multiple of

(3, 4)

, so

10

can only be an upper bound, and the length of the sum must be computed directly.

Subtracting the lengths

The distance between two vectors is computed as the difference of their lengths,

\|\mathbf{a}\| - \|\mathbf{b}\|

, instead of the length of their difference,

\|\mathbf{a} - \mathbf{b}\|

.

Wrong working: for

\mathbf{a} = (3, 4)

and

\mathbf{b} = (4, 3)

,

d(\mathbf{a}, \mathbf{b}) = \|\mathbf{a}\| - \|\mathbf{b}\| = 5 - 5 = 0

Right working: subtract the vectors first, then take the norm.

\mathbf{a} - \mathbf{b} = (-1, 1), \qquad d(\mathbf{a}, \mathbf{b}) = \sqrt{(-1)^2 + 1^2} = \sqrt{2}

Why it happens: the two expressions differ only in where the bars close, and the word "difference" suggests subtracting the two lengths that are already known.

How to check: a distance of

0

means the two vectors are identical, and

(3, 4) \neq (4, 3)

, so

0

cannot be right. In general the difference of the lengths is only a lower bound for the distance, by the reverse triangle inequality.

Dividing by the squared length

When a vector is normalized, its components are divided by the sum of their squares,

\|\mathbf{v}\|^2

, because the square root was never taken.

Wrong working: for

\mathbf{v} = (1, 2, 2)

, the sum of squares is

1 + 4 + 4 = 9

, and

\frac{\mathbf{v}}{9} = \left(\frac{1}{9}, \frac{2}{9}, \frac{2}{9}\right)

Right working: the length is

\|\mathbf{v}\| = \sqrt{9} = 3

, so

\hat{\mathbf{v}} = \frac{\mathbf{v}}{3} = \left(\frac{1}{3}, \frac{2}{3}, \frac{2}{3}\right)

Why it happens: the sum of squares is the first number the computation produces, and it is easy to stop there. The squared norm also appears on its own in formulas such as

\|\mathbf{v}\|^2 = \mathbf{v} \cdot \mathbf{v}

, so it looks like a finished quantity.

How to check: square and add the components of the result. The wrong answer gives

\frac{1}{81} + \frac{4}{81} + \frac{4}{81} = \frac{1}{9}

, a length of

\frac{1}{3}

. The right answer gives

\frac{1}{9} + \frac{4}{9} + \frac{4}{9} = 1

.

Vector Magnitude FAQ

Why does the norm use double bars instead of single ones?

+

The bar count marks what is being measured. One pair takes the absolute value of a scalar, two pairs take the length of a vector. The scaling law puts both in a single line:

\|c\mathbf{v}\| = |c|\,\|\mathbf{v}\|

. Physics texts often write single bars for magnitude, which is fine until determinants arrive and claim the same glyph.Read more →

What does the subscript in $\|\mathbf{v}\|_2$ refer to?

+

The measuring rule, not a component. The

1

-norm sums absolute values, the

2

-norm is the familiar Euclidean length, and the

\infty

-norm takes the largest component. An unsubscripted

\|\mathbf{v}\|

means the

2

-norm by default. Reading it as the norm of the second component is a common misparse.Read more →

What does the hat in $\hat{\mathbf{u}}$ promise?

+

That the vector has length exactly

1

. Writing

\hat{\mathbf{u}}

asserts

\|\hat{\mathbf{u}}\| = 1

without further comment, which is why normalizing a vector is described as putting a hat on it. Watch the field, though: in statistics the same hat means "estimate of" and says nothing about length.Read more →

Is the distance between two vectors the difference of their lengths?

+

No, it is the length of their difference,

\|\mathbf{a} - \mathbf{b}\|

. For

\mathbf{a} = (3, 4)

and

\mathbf{b} = (4, 3)

both lengths are

5

, so

\|\mathbf{a}\| - \|\mathbf{b}\| = 0

, as if the points coincided, yet

\mathbf{a} - \mathbf{b} = (-1, 1)

gives a distance of

\sqrt{2}

. The difference of the lengths is only a lower bound,

\bigl|\, \|\mathbf{a}\| - \|\mathbf{b}\| \,\bigr| \leq \|\mathbf{a} - \mathbf{b}\|

, the reverse triangle inequality.Read more →

When does the triangle inequality become an equality?

+

Exactly when one vector is a non-negative multiple of the other, which includes the case where either is

\mathbf{0}

. Then the tip-to-tail path is already straight:

(3, 0)

and

(4, 0)

give a sum of length

7 = 3 + 4

. Any bend makes the sum shorter:

(3, 0)

and

(0, 4)

give

5 < 7

, and

(3, 4)

with

(-3, 4)

gives

\|(0, 8)\| = 8 < 10

.Read more →

How long is the diagonal of a cube in $n$ dimensions?

+

\sqrt{n}

for a unit cube. The diagonal runs from the origin to the corner

(1, 1, \ldots, 1)

, and the norm adds one squared

1

per component, giving

\sqrt{1 + 1 + \cdots + 1} = \sqrt{n}

. That is

\sqrt{2}

across the unit square and

\sqrt{3}

through the unit cube, and it keeps growing: in

\mathbb{R}^{100}

a cube whose edges all have length

1

has a diagonal of length

10

.Read more →

What does a circle look like in the taxicab norm?

+

A square standing on a corner. Measured by

\|\mathbf{v}\|_1 = |v_1| + |v_2|

, the points at distance

1

from the origin form the square with corners

(\pm 1, 0)

and

(0, \pm 1)

. In the same norm each side has length

1 + 1 = 2

, so the circumference is

8

against a diameter of

2

: in the taxicab plane the ratio that plays the part of

\pi

is

4

.Read more →

Magnitude of a Vector

Measuring Length and Extracting Direction

Magnitude in Two and Three Dimensions

Norm Notation

Norm Notation

The General Norm

Properties of the Norm

Non-Negativity

Scaling

Triangle Inequality

Properties of the norm

Distance Between Vectors

Unit Vectors

Normalization

Summary: Norm and the Concepts Built from It

Common Pitfalls

Vector Magnitude FAQ

Why does the norm use double bars instead of single ones?

What does the subscript in $\|\mathbf{v}\|_2$ refer to?

What does the hat in $\hat{\mathbf{u}}$ promise?

Is the distance between two vectors the difference of their lengths?

When does the triangle inequality become an equality?

How long is the diagonal of a cube in $n$ dimensions?

What does a circle look like in the taxicab norm?

Magnitude of a Vector

Measuring Length and Extracting Direction

Magnitude in Two and Three Dimensions

Norm Notation

Norm Notation

The General Norm

Properties of the Norm

Non-Negativity

Scaling

Triangle Inequality

Properties of the norm

Distance Between Vectors

Unit Vectors

Normalization

Summary: Norm and the Concepts Built from It

Common Pitfalls

Vector Magnitude FAQ

Why does the norm use double bars instead of single ones?

What does the subscript in ∥v∥2\|\mathbf{v}\|_2∥v∥2​ refer to?

What does the hat in u^\hat{\mathbf{u}}u^ promise?

Is the distance between two vectors the difference of their lengths?

When does the triangle inequality become an equality?

How long is the diagonal of a cube in nnn dimensions?

What does a circle look like in the taxicab norm?

What does the subscript in $\|\mathbf{v}\|_2$ refer to?

What does the hat in $\hat{\mathbf{u}}$ promise?

How long is the diagonal of a cube in $n$ dimensions?