Hitchin–Thorpe inequality

Let M be a closed, oriented, four-dimensional smooth manifold. If there exists a Riemannian metric on M which is an Einstein metric, then

${\displaystyle \chi (M)\geq {\frac {3}{2}}|\tau (M)|,}$

where χ(M) is the Euler characteristic of M and τ(M) is the signature of M.

This inequality was first stated by John Thorpe in a footnote to a 1969 paper focusing on manifolds of higher dimension.[1] Nigel Hitchin then rediscovered the inequality, and gave a complete characterization of the equality case in 1974;[2] he found that if (M, g) is an Einstein manifold for which equality in the Hitchin-Thorpe inequality is obtained, then the Ricci curvature of g is zero; if the sectional curvature is not identically equal to zero, then (M, g) is a Calabi–Yau manifold whose universal cover is a K3 surface.

Already in 1961, Marcel Berger showed that the Euler characteristic is always non-negative.[3][4]

Let (M, g) be a four-dimensional smooth Riemannian manifold which is Einstein. Given any point p of M, there exists a g_p-orthonormal basis e₁, e₂, e₃, e₄ of the tangent space T_pM such that the curvature operator Rm_p, which is a symmetric linear map of ∧²T_pM into itself, has matrix

${\displaystyle {\begin{pmatrix}\lambda _{1}&0&0&\mu _{1}&0&0\\0&\lambda _{2}&0&0&\mu _{2}&0\\0&0&\lambda _{3}&0&0&\mu _{3}\\\mu _{1}&0&0&\lambda _{1}&0&0\\0&\mu _{2}&0&0&\lambda _{2}&0\\0&0&\mu _{3}&0&0&\lambda _{3}\end{pmatrix}}}$

relative to the basis e₁ ∧ e₂, e₁ ∧ e₃, e₁ ∧ e₄, e₃ ∧ e₄, e₄ ∧ e₂, e₂ ∧ e₃. One has that μ₁ + μ₂ + μ₃ is zero and that λ₁ + λ₂ + λ₃ is one-fourth of the scalar curvature of g at p. Furthermore, under the conditions λ₁ ≤ λ₂ ≤ λ₃ and μ₁ ≤ μ₂ ≤ μ₃, each of these six functions is uniquely determined and defines a continuous real-valued function on M.

According to Chern-Weil theory, if M is oriented then the Euler characteristic and signature of M can be computed by

${\displaystyle {\begin{aligned}\chi (M)&={\frac {1}{4\pi ^{2}}}\int _{M}{\big (}\lambda _{1}^{2}+\lambda _{2}^{2}+\lambda _{3}^{2}+\mu _{1}^{2}+\mu _{2}^{2}+\mu _{3}^{2}{\big )}\,d\mu _{g}\\\tau (M)&={\frac {1}{3\pi ^{2}}}\int _{M}{\big (}\lambda _{1}\mu _{1}+\lambda _{2}\mu _{2}+\lambda _{3}\mu _{3}{\big )}\,d\mu _{g}.\end{aligned}}}$

Equipped with these tools, the Hitchin-Thorpe inequality amounts to the elementary observation

${\displaystyle \lambda _{1}^{2}+\lambda _{2}^{2}+\lambda _{3}^{2}+\mu _{1}^{2}+\mu _{2}^{2}+\mu _{3}^{2}=\underbrace {(\lambda _{1}-\mu _{1})^{2}+(\lambda _{2}-\mu _{2})^{2}+(\lambda _{3}-\mu _{3})^{2}} _{\geq 0}+2{\big (}\lambda _{1}\mu _{1}+\lambda _{2}\mu _{2}+\lambda _{3}\mu _{3}{\big )}.}$

A natural question to ask is whether the Hitchin–Thorpe inequality provides a sufficient condition for the existence of Einstein metrics. In 1995, Claude LeBrun and Andrea Sambusetti independently showed that the answer is no: there exist infinitely many non-homeomorphic compact, smooth, oriented 4-manifolds M that carry no Einstein metrics but nevertheless satisfy

${\displaystyle \chi (M)>{\frac {3}{2}}|\tau (M)|.}$

LeBrun's examples are actually simply connected, and the relevant obstruction depends on the smooth structure of the manifold.[5] By contrast, Sambusetti's obstruction only applies to 4-manifolds with infinite fundamental group, but the volume-entropy estimate he uses to prove non-existence only depends on the homotopy type of the manifold.[6]

• Besse, Arthur L. (1987). Einstein Manifolds. Classics in Mathematics. Berlin: Springer. ISBN 3-540-74120-8.