4.5.33 · D4Linear Algebra (Full)

Exercises — Inner product spaces — dot product generalization

2,686 words12 min readBack to topic

This page is a self-test ladder. Every problem is stated cleanly, then a collapsible solution hides the full worked answer so you can try first. The rungs climb from recognising the axioms to building new inner products from scratch.

Parent: Inner product spaces — dot product generalization. Along the way we lean on Dot Product, Norms and Distance, Cauchy-Schwarz Inequality, Orthogonality and Gram-Schmidt, Orthogonal Projections, and Fourier Series.


Level 1 — Recognition

(Can you tell whether something even IS an inner product? Pure axiom-checking.)

L1.1

Is on an inner product? (This is the ordinary Dot Product — confirm it obeys all three axioms.)

Recall Solution L1.1

Check the three axioms in order.

  • Symmetry: and . Numbers multiply in any order, so these are equal. ✓
  • Linearity in slot 1: . ✓
  • Positive-definite: , and it equals only when , i.e. . ✓

All three hold, so yes, it is an inner product.

L1.2

Is on an inner product?

Recall Solution L1.2

Symmetry ✓ and linearity ✓ (same algebra as L1.1, just a minus sign). But test positivity with : A "length-squared" came out negative. Positive-definiteness fails, so this is not an inner product.

L1.3

Is on an inner product?

Recall Solution L1.3

Symmetry: swapping gives — same thing, ✓. Linearity ✓. Positivity: . Take : Not an inner product — positivity fails again.


Level 2 — Application

(Plug into the derived-geometry formulas: norm, distance, angle.)

L2.1

Using the weighted inner product on , compute .

Recall Solution L2.1

Norm from an inner product: (see Norms and Distance).

L2.2

Same weighted inner product. Find the distance between and .

Recall Solution L2.2

Distance is the norm of the difference: .

L2.3

On with , compute the norm of .

Recall Solution L2.3


Level 3 — Analysis

(Find angles and orthogonality; verify Cauchy–Schwarz is not violated.)

L3.1

With the ordinary dot product on , find the angle between and .

Recall Solution L3.1

Angle formula: . Figure s01 shows this: is drawn flat along the x-axis (lavender), points up-right at the diagonal (coral), and the mint arc between them is the opening. The arrow rises one unit for every one unit right, so it splits the corner exactly in half — that half-of-a-right-angle is the , and is the horizontal shadow of the unit-scaled diagonal.

Figure — Inner product spaces — dot product generalization

L3.2

With the weighted inner product , find the angle between the same and . Why is it different from L3.1?

Recall Solution L3.2

Why different? The weights stretch the axes, so "angle" is measured with a different ruler. Same arrows on paper, different geometry. Figure s02 shows this: the identical arrows (lavender) and (coral) are drawn once, but two arcs sit in the corner — the mint arc labelled "dot: " (the ordinary ruler from L3.1) and, tucked inside it, the smaller butter-coloured arc labelled "weighted: ". Because the weight on the -part () is larger than on the -part (), the weighted ruler counts horizontal agreement more heavily, so the two arrows look "more aligned" and the angle shrinks from to about .

Figure — Inner product spaces — dot product generalization

L3.3

On the polynomials of degree with , are and orthogonal? Check that obeys Cauchy-Schwarz Inequality.

Recall Solution L3.3

First the cross term: Since , they are orthogonal under this inner product (see Orthogonality and Gram-Schmidt). Now the norms, so the division in is legal. Evaluate each polynomial at and : Both norms are nonzero, so which sits safely inside , so Cauchy–Schwarz holds ( in this inner product).


Level 4 — Synthesis

(Combine derivations: prove inequalities and identities from the axioms.)

L4.1

Prove the parallelogram law in any inner product space:

Recall Solution L4.1

Start from and expand by linearity in both slots (FOIL) — showing every cross term explicitly: Now apply symmetry (): the two middle terms are equal, so they combine into : Identically, for the difference (the two cross terms are and , each ): Add them: the cross terms cancel, leaving WHAT this means: the two diagonals of a parallelogram (sum and difference ) together carry exactly the energy of its four sides.

L4.2

Prove the polarization identity: in a real inner product space,

Recall Solution L4.2

Use the two FOIL expansions established in L4.1, where each "" came from the two equal cross terms collapsing by symmetry. Subtract them: The and terms cancel, and : Divide by . WHY it's remarkable: the inner product (angles!) is completely reconstructible from the norm (lengths!) alone. Length secretly knows angle.

L4.3

Prove that if (i.e. ) then the Pythagorean theorem holds: .

Recall Solution L4.3

From the FOIL expansion , the middle term is , so This is the abstract heart of every right-triangle picture.


Level 5 — Mastery

(Build inner products / apply the machinery to functions — the Fourier Series and Orthogonal Projections frontier.)

L5.1

Find all real values of for which on is a valid inner product.

Recall Solution L5.1

Symmetry and linearity hold for every . The only obstacle is positive-definiteness: For this to be for all and zero only at , take : we need .

  • If : , zero forces and . ✓
  • If : gives but — positivity fails.
  • If : gives a negative value — fails.

Answer: .

L5.2

On with , verify that and are orthogonal — the fact that launches Fourier Series.

Recall Solution L5.2

Orthogonal ✓. Integrating a full-period cosine over washes it out to zero — the constant and the wave don't overlap.

L5.3

Using the function inner product on , find the orthogonal projection of onto the constant function . (This is the "best constant approximation" — its average. See Orthogonal Projections.)

Recall Solution L5.3

The projection of onto is . The best constant approximation to on is the constant — literally its average value. Figure s03 shows this: the coral line is climbing from to ; the flat lavender line at height is the projection. The short mint segments are the vertical gaps (the residual) — they are positive on the right half and negative on the left half, and they cancel out over the interval, which is exactly why is the balanced (average) height.

Figure — Inner product spaces — dot product generalization

L5.4

Check the Cauchy-Schwarz Inequality directly, with numbers, for and on with .

Recall Solution L5.4

Compute the three pieces as integrals. Cross term: Norms: Compare: Cauchy–Schwarz holds, and the inequality is strict () because and are not scalar multiples of one another — equality in Cauchy–Schwarz happens only when the two vectors are parallel.


Recall One-line summary of the ladder

L1 test the axioms → L2 use the norm and distance formulas → L3 use angle and orthogonality → L4 prove identities from the axioms → L5 build new inner products and project functions. Each rung reuses only what the rung below earned.