Sign in to save

Bookmark this page so you can find it later.

Sign in to save

Bookmark this page so you can find it later.

Inner Product Spaces & Orthogonality Reference cheat sheet - grade college

Click image to open full size

Math Grade college

Inner Product Spaces & Orthogonality Reference Cheat Sheet

A printable reference covering inner products, norms, orthogonality, projections, Gram-Schmidt, and orthonormal bases for college grades.

TopicsVectorsAlgebraMathematicsStudy GuideProcess
Download PNG

Related Tools

Related Labs

Related Worksheets

Related Infographics

Study as Flashcards

Inner product spaces extend dot product geometry to vectors, functions, and abstract vector spaces. This cheat sheet helps students connect algebraic rules with geometric ideas such as length, angle, perpendicularity, and projection. It is useful for linear algebra, differential equations, numerical methods, and applied mathematics courses.

The reference focuses on definitions and formulas that appear often in proofs and computations.

The core idea is that an inner product ⟨u,v⟩\langle u,v\rangle produces a scalar that measures alignment between two vectors. From it, students define the norm ∥v∥=⟨v,v⟩\|v\|=\sqrt{\langle v,v\rangle}, orthogonality by ⟨u,v⟩=0\langle u,v\rangle=0, and projection by proj⁡u(v)=⟨v,u⟩⟨u,u⟩u\operatorname{proj}_u(v)=\frac{\langle v,u\rangle}{\langle u,u\rangle}u. Orthonormal bases simplify coordinates because v=∑i⟨v,ei⟩eiv=\sum_i \langle v,e_i\rangle e_i when {ei}\{e_i\} is orthonormal.

The Gram-Schmidt process converts an independent list into an orthogonal or orthonormal basis.

Key Facts

  • An inner product satisfies positivity ⟨v,v⟩≥0\langle v,v\rangle\ge 0, definiteness ⟨v,v⟩=0\langle v,v\rangle=0 only when v=0v=0, linearity in one argument, and conjugate symmetry ⟨u,v⟩=⟨v,u⟩‾\langle u,v\rangle=\overline{\langle v,u\rangle}.
  • The norm induced by an inner product is ∥v∥=⟨v,v⟩\|v\|=\sqrt{\langle v,v\rangle}, and it is always nonnegative.
  • Two vectors are orthogonal exactly when ⟨u,v⟩=0\langle u,v\rangle=0.
  • The Cauchy-Schwarz inequality states ∣⟨u,v⟩∣≤∥u∥∥v∥|\langle u,v\rangle|\le \|u\|\|v\|.
  • The angle between nonzero real vectors satisfies cos⁡θ=⟨u,v⟩∥u∥∥v∥\cos \theta=\frac{\langle u,v\rangle}{\|u\|\|v\|}.
  • The projection of vv onto a nonzero vector uu is proj⁡u(v)=⟨v,u⟩⟨u,u⟩u\operatorname{proj}_u(v)=\frac{\langle v,u\rangle}{\langle u,u\rangle}u.
  • If {e1,…,en}\{e_1,\dots,e_n\} is an orthonormal basis, then every vector has the expansion v=∑i=1n⟨v,ei⟩eiv=\sum_{i=1}^{n}\langle v,e_i\rangle e_i.
  • In Gram-Schmidt, wk=vk−∑i=1k−1⟨vk,wi⟩⟨wi,wi⟩wiw_k=v_k-\sum_{i=1}^{k-1}\frac{\langle v_k,w_i\rangle}{\langle w_i,w_i\rangle}w_i and ek=wk∥wk∥e_k=\frac{w_k}{\|w_k\|}.

Vocabulary

Inner product
A rule ⟨u,v⟩\langle u,v\rangle that assigns scalars to pairs of vectors while preserving the algebraic properties needed to measure length and angle.
Norm
The length of a vector induced by an inner product, defined by ∥v∥=⟨v,v⟩\|v\|=\sqrt{\langle v,v\rangle}.
Orthogonal vectors
Vectors uu and vv are orthogonal when their inner product is zero, so ⟨u,v⟩=0\langle u,v\rangle=0.
Orthonormal set
A set of vectors is orthonormal when each vector has norm 11 and distinct vectors have inner product 00.
Projection
The projection of vv onto uu is the component of vv in the direction of uu, given by proj⁡u(v)=⟨v,u⟩⟨u,u⟩u\operatorname{proj}_u(v)=\frac{\langle v,u\rangle}{\langle u,u\rangle}u for u≠0u\ne 0.
Orthogonal complement
The orthogonal complement of a subspace WW is W⊥={v:⟨v,w⟩=0 for all w∈W}W^{\perp}=\{v:\langle v,w\rangle=0\text{ for all }w\in W\}.

Common Mistakes to Avoid

  • Using ⟨v,u⟩∥u∥u\frac{\langle v,u\rangle}{\|u\|}u for projection is wrong because the denominator must be ⟨u,u⟩=∥u∥2\langle u,u\rangle=\|u\|^2 unless uu is already a unit vector.
  • Assuming orthogonal means orthonormal is wrong because orthogonal vectors only require ⟨u,v⟩=0\langle u,v\rangle=0, while orthonormal vectors also require ∥u∥=1\|u\|=1.
  • Forgetting complex conjugation is wrong in complex inner product spaces because conjugate symmetry requires ⟨u,v⟩=⟨v,u⟩‾\langle u,v\rangle=\overline{\langle v,u\rangle}.
  • Applying the angle formula to a zero vector is wrong because cos⁡θ=⟨u,v⟩∥u∥∥v∥\cos \theta=\frac{\langle u,v\rangle}{\|u\|\|v\|} is undefined when either norm is 00.
  • Skipping normalization in Gram-Schmidt is wrong when an orthonormal basis is required because the vectors wkw_k are orthogonal but may not have norm 11.

Practice Questions

  1. 1 In R3\mathbb{R}^3 with the dot product, compute ⟨u,v⟩\langle u,v\rangle and decide whether u=(1,2,−1)u=(1,2,-1) and v=(3,−1,1)v=(3,-1,1) are orthogonal.
  2. 2 Find proj⁡u(v)\operatorname{proj}_u(v) for u=(2,1)u=(2,1) and v=(3,4)v=(3,4) using proj⁡u(v)=⟨v,u⟩⟨u,u⟩u\operatorname{proj}_u(v)=\frac{\langle v,u\rangle}{\langle u,u\rangle}u.
  3. 3 Use one step of Gram-Schmidt to turn v1=(1,1,0)v_1=(1,1,0) and v2=(1,0,1)v_2=(1,0,1) into an orthogonal pair w1,w2w_1,w_2.
  4. 4 Explain why an orthonormal basis makes coordinate calculations simpler than a basis that is independent but not orthogonal.

Understanding Inner Product Spaces & Orthogonality Reference

An inner product does more than assign a number to two vectors. It determines which directions count as close, long, or perpendicular in a particular setting. This matters because the usual coordinate dot product is not the only possible choice.

In a space of functions, an inner product can be built by multiplying two functions and adding their products across an interval. A weighted version gives some parts of that interval more importance than others. This appears when measurements have unequal reliability, or when a physical model gives one region more influence.

Projection is best understood through the leftover part. When a vector is projected onto a direction or subspace, the result is the part that can be explained using that direction or subspace. The residual is what remains after that explanation.

A key fact is that the residual is perpendicular to every allowed direction in the subspace. This is why projection gives the closest point in the subspace. It minimizes the length of the residual.

In data fitting, this idea produces least squares methods. A line fitted to noisy data cannot usually pass through every point, so the method chooses the line with the smallest total squared error.

Orthonormal bases are especially useful because each coordinate can be found independently. There is no need to solve a full system of equations just to determine how much of each basis direction is present. This makes calculations clearer in Fourier series, signal processing, and quantum mechanics.

A sound recording can be represented as contributions from different frequency patterns. A signal with unwanted noise can be approximated by keeping important orthogonal components and discarding small ones. The success of this method depends on choosing a basis that matches the type of signal being studied.

Gram Schmidt is a constructive process, but students should watch its assumptions carefully. The starting vectors must be linearly independent. If a new vector is already made from earlier vectors, its leftover part becomes the zero vector, and it cannot be normalized.

The order of the starting list can change the resulting orthonormal basis, even though the final span stays the same. In exact hand calculations, fractions and square roots are normal. In computer calculations, nearly dependent vectors can cause rounding errors because subtraction removes nearly equal quantities.

More stable methods are used in serious numerical work. For proofs, pay attention to the reason each new leftover vector is perpendicular to all earlier ones. That argument is the heart of the process.