JOVANA
Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Symmetric Algebra and Raising/Lowering Indices

The mirror image of the wedge is the symmetric product, which models polynomials and quadratic forms. Then an inner product lets us convert between vectors and covectors — raising and lowering indices.

The symmetric product

Where the wedge demands sign flips, the symmetric algebra demands the opposite: v . w = w . v. Quotient the tensor algebra by the relations v tensor w - w tensor v = 0 and the graded pieces Sym^k V become the space of homogeneous degree-k polynomials in the basis vectors. Sym^2 V is exactly the home of quadratic forms — the things Vol I wrote as u^T B v with symmetric B.

For n = dim V = 3:
  dim Sym^k V = C(n+k-1, k).
  Sym^2 of R^3 has dim C(4,2) = 6, basis:
    x^2, y^2, z^2, xy, xz, yz   (the symmetric monomials)
  -> exactly the 6 free entries of a symmetric 3x3 matrix.

Compare the two quotients of degree 2 over R^3:
  dim Sym^2 V = 6   (symmetric part:   T_{ij} = T_{ji})
  dim Lambda^2 V = 3 (antisymmetric:    T_{ij} = -T_{ji})
  6 + 3 = 9 = dim(V (x) V).   Every 2-tensor splits sym + antisym.
Degree-2 tensors split into a symmetric part (Sym^2) and an antisymmetric part (Lambda^2).

Vectors versus covectors

An upper-index object v^i lives in V; a lower-index object a_i lives in the dual space V* of linear functionals. Without extra structure these are different spaces — there is no canonical way to turn a vector into a functional. But an inner product supplies exactly that bridge.

Given an inner product with matrix g_{ij}, lowering an index means v_i = g_{ij} v^j (a contraction with the metric), and raising uses the inverse g^{ij}: v^i = g^{ij} v_j. In an orthonormal basis g is the identity, so raising and lowering do nothing visible — which is why Vol I could blur the distinction between a vector and its transpose.

A worked index dance

Let us push indices on a concrete example. Take V = R^2 with the non-orthonormal inner product given by g = [2, 1; 1, 1], so g^{-1} = [1, -1; -1, 2]. We will lower then raise the same vector and confirm we return to where we started.

Metric and its inverse:
  g   = [ 2  1 ;  1  1 ]        g^{-1} = [  1  -1 ;  -1   2 ]
  (check:  g * g^{-1} = I)

Start with vector  v^j = (3, 4)   (upper index, a genuine vector).

LOWER:  v_i = g_{ij} v^j
  v_1 = 2*3 + 1*4 = 10
  v_2 = 1*3 + 1*4 = 7        ->  v_i = (10, 7)   (now a covector)

RAISE back:  v^i = g^{ij} v_j
  v^1 =  1*10 + (-1)*7 = 3
  v^2 = (-1)*10 + 2*7  = 4   ->  v^i = (3, 4)   back to start.

Norm via the metric (a double contraction):
  ||v||^2 = g_{ij} v^i v^j = v_i v^i = 10*3 + 7*4 = 58.
Lower with g, raise with g-inverse; the round trip is the identity, and the norm is a contraction.