Inner product space

Intuition [Beginner]

An inner product is a dot product. It measures length, angle, and projection all at once.

In ordinary plane geometry, the dot product tells whether two arrows are perpendicular. It also tells how much of one arrow points in the direction of another.

Inner product spaces keep this geometry in broader settings: functions, signals, and infinite lists can have dot-product-like measurements. Hilbert spaces add completeness to this picture.

Visual [Beginner]

An inner product lets one vector cast a projection onto another direction.

Orthogonality means the projection measurement is zero.

Worked example [Beginner]

In the plane, take $u=(1,0)$ and $v=(3,4)$. The dot product with $u$ reads off the horizontal part of $v$.

The result is $3$. That means the shadow of $v$ in the $u$ direction has length $3$.

If $w=(0,5)$, then the dot product of $u$ and $w$ is $0$. The two arrows are perpendicular.

What this tells us: inner products turn algebraic vectors into metric geometry.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $V$ be a real vector space. An inner product is a function $⟨-,-⟩:V\times V\to \mathbb{R}$ that is bilinear, symmetric, and positive definite:

$$⟨v,v⟩>0\text{for }v\ne 0.$$

For a complex vector space, the standard analytic convention is sesquilinear: linear in one variable, conjugate-linear in the other, conjugate symmetric, and positive definite ^{[Conway §I]}.

An inner product induces a norm

$$\Vert v\Vert =\sqrt{⟨v,v⟩}.$$

Thus every inner product space is a normed vector space 02.11.06.

Key theorem with proof [Intermediate+]

Theorem (Cauchy-Schwarz inequality). In a real inner product space,

$$|⟨v,w⟩|\le \Vert v\Vert \Vert w\Vert .$$

Proof. If $w=0$, both sides are zero. Assume $w\ne 0$. For every real $t$,

$$0\le \Vert v-tw{\Vert }^2=⟨v-tw,v-tw⟩=\Vert v{\Vert }^2-2t⟨v,w⟩+t^2\Vert w{\Vert }^2.$$

This quadratic polynomial in $t$ is nonnegative for all real $t$, so its discriminant is at most zero:

$$4⟨v,w{⟩}^2-4\Vert v{\Vert }^2\Vert w{\Vert }^2\le 0.$$

Therefore $⟨v,w{⟩}^2\le \Vert v{\Vert }^2\Vert w{\Vert }^2$, and taking square roots gives the result. $\square$

Bridge. The construction here builds toward 02.11.08 (hilbert space), where the same data is developed in the next layer of the strand. The defining pattern appears again in those units in a sharpened form, where the local data is glued or quotiented. Putting these together, the foundational insight is that the data of this unit gives the structural signature that the rest of the strand reads off.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

lean_status: none is recorded because this unit is a curriculum bridge between bilinear-form language and analytic inner-product-space notation. Mathlib itself has strong support for inner product spaces.

Advanced results [Master]

The induced norm of an inner product satisfies the parallelogram identity. Conversely, over real vector spaces, a norm satisfying the parallelogram identity comes from an inner product by polarization ^{[Conway §I]}.

Orthogonal projection is the geometric operation that sends a vector to its closest point in a subspace when the required closest point exists. In complete spaces, this becomes the Hilbert projection theorem 02.11.08.

Synthesis. Inner product spaces introduce geometry into linear algebra: the inner product provides lengths, angles, orthogonality, and projections, making available the Pythagorean theorem, the Cauchy-Schwarz inequality, and the Gram-Schmidt process. The completion of an inner-product space to a Hilbert space is the setting for the spectral theorem (self-adjoint operators are diagonalisable in an orthonormal basis), the Riesz representation theorem (every continuous linear functional is an inner product with a unique vector), and the Parseval identity (Fourier coefficients preserve the L^2 norm). The pattern of orthogonal decomposition — V = W + W-perp for any closed subspace W — is the analytic counterpart of Maschke's theorem in representation theory and underpins the method of least squares, the theory of orthogonal polynomials, and the construction of wavelet bases.

Full proof set [Master]

The triangle inequality for the induced norm follows from Cauchy-Schwarz as in Exercise 6. Thus every inner product space is a normed vector space.

For the parallelogram identity, expand both terms using bilinearity and symmetry:

$$\Vert v+w{\Vert }^2+\Vert v-w{\Vert }^2=⟨v+w,v+w⟩+⟨v-w,v-w⟩=2\Vert v{\Vert }^2+2\Vert w{\Vert }^2.$$

Connections [Master]

• Inner product spaces depend on normed vector spaces 02.11.06 and bilinear forms 01.01.15. Hilbert spaces 02.11.08 are complete inner product spaces. Orthogonal groups 03.03.03 preserve inner products, and the orthogonal frame bundle 03.05.03 is built from orthonormal bases.

• Unbounded self-adjoint operators 02.11.03 and CFT state spaces 03.10.02 use Hilbert-space geometry built from this unit.

Historical & philosophical context [Master]

Inner products generalize Euclidean dot products to function spaces and sequence spaces. Reed and Simon use this structure as the geometric foundation for quantum mechanical Hilbert spaces ^{[Reed-Simon §II]}.

Conway develops inner product spaces before Hilbert spaces because completeness changes the projection theory and operator theory ^{[Conway §I]}.

Bibliography [Master]

• Michael Reed and Barry Simon, Methods of Modern Mathematical Physics, Vol. I, §II. ^{[Reed-Simon §II]}
• John Conway, A Course in Functional Analysis, §I. ^{[Conway §I]}