MathLabs
TheoremProved

Sylvester's law of inertia

Statement

No matter which invertible change of basis is used to diagonalize a real quadratic form QQ on Rn\mathbb{R}^n, the number n+n_+ of positive coefficients, the number n−n_- of negative coefficients, and the number n0n_0 of zero coefficients are always the same; the triple (n+,n−,n0)(n_+, n_-, n_0) is an intrinsic invariant of QQ.

Why is it true?

Switching to a new coordinate system can stretch the axes and change the individual magnitudes of the diagonal numbers did_i, but it can never turn an upward-curving direction into a downward-curving one without passing through a flat direction — so the counts of upward, downward, and flat axes are locked in forever.

Proof sketch

Suppose QQ is diagonalized in two bases {e1,…,en}\{\mathbf{e}_1,\dots,\mathbf{e}_n\} and {f1,…,fn}\{\mathbf{f}_1,\dots,\mathbf{f}_n\}, with positive, negative, and zero counts (n+,n−,n0)(n_+, n_-, n_0) in the first basis and (p+,p−,p0)(p_+, p_-, p_0) in the second. Order each basis so the positive coefficients come first, then the negative ones, then the zeros.

Suppose toward a contradiction that n+>p+n_+ > p_+. Let V+=span(e1,…,en+)V_+ = \mathrm{span}(\mathbf{e}_1,\dots,\mathbf{e}_{n_+}), a subspace of dimension n+n_+ on which Q(x)>0Q(\mathbf{x}) > 0 for every nonzero x∈V+\mathbf{x} \in V_+ (since only the positive squares in the e\mathbf{e}-expansion are active on V+V_+).

Similarly, let W−=span(fp++1,…,fn)W_- = \mathrm{span}(\mathbf{f}_{p_+ + 1},\dots,\mathbf{f}_n), a subspace of dimension n−p+n - p_+ on which Q(x)≤0Q(\mathbf{x}) \le 0 for every x∈W−\mathbf{x} \in W_- (since only the negative and zero squares in the f\mathbf{f}-expansion are active on W−W_-).

Now count dimensions inside Rn\mathbb{R}^n: since n+>p+n_+ > p_+, we have dim⁡V++dim⁡W−=n++(n−p+)>n\dim V_+ + \dim W_- = n_+ + (n - p_+) > n. By the dimension formula for subspaces, two subspaces whose dimensions add up to more than nn cannot have trivial intersection; hence there exists a nonzero vector v∈V+∩W−\mathbf{v} \in V_+ \cap W_-.

Because v∈V+\mathbf{v} \in V_+ and v≠0\mathbf{v} \neq \mathbf{0}, we must have Q(v)>0Q(\mathbf{v}) > 0; because v∈W−\mathbf{v} \in W_-, we must simultaneously have Q(v)≤0Q(\mathbf{v}) \le 0, an outright contradiction. Therefore n+≤p+n_+ \le p_+, and by symmetry of the two bases p+≤n+p_+ \le n_+, so n+=p+n_+ = p_+. Applying the exact same argument to −Q-Q gives n−=p−n_- = p_-, and finally n0=n−n+−n−=p0n_0 = n - n_+ - n_- = p_0.

Topics that use this theorem

Step-by-step proofs

No step-by-step proof yet for this theorem.

References

  1. Roger A. Horn, Charles R. Johnson (2012). Matrix Analysis (2nd ed.)
  2. Gilbert Strang (2016). Introduction to Linear Algebra