Abstract
Let \(M = G/H\) be a connected simply connected homogeneous manifold of a compact, not necessarily connected Lie group \(G\). We will assume that the isotropy \(H\)-module \(\mathfrak{g/h }\) has a simple spectrum, i.e. irreducible submodules are mutually non-equivalent. There exists a convex Newton polytope \(N=N(G,H)\), which was used for the estimation of the number of isolated complex solutions of the algebraic Einstein equation for invariant metrics on \(G/H\) (up to scaling). Using the moment map, we identify the space \(\mathcal{M }_1\) of invariant Riemannian metrics of volume 1 on \(G/H\) with the interior of this polytope \(N\). We associate with a point \({x \in \partial N}\) of the boundary a homogeneous Riemannian space (in general, only local) and we extend the Einstein equation to \(\partial N\). As an application of the Alekseevsksky–Kimel’fel’d theorem, we prove that all solutions of the Einstein equation associated with points of the boundary are locally Euclidean. We describe explicitly the set \(T\subset \partial N\) of solutions at the boundary together with its natural triangulation. Investigating the compactification \({\overline{\mathcal{M }}}_{1} = N\) of \(\mathcal{M }_1\), we get an algebraic proof of the deep result by Böhm, Wang and Ziller about the compactness of the set \(\mathcal{E }_1 \subset \mathcal{M }_1\) of Einstein metrics. The original proof by Böhm, Wang and Ziller was based on a different approach and did not use the simplicity of the spectrum. In Appendix, we consider the non-symmetric flag manifolds \(G/H\) with the second Betti number \(b_2=1\). We calculate the normalized volumes \(2,6,20,82,344\) of the corresponding Newton polytopes and discuss the number of complex solutions of the algebraic Einstein equation and the finiteness problem.
Similar content being viewed by others
Notes
Moreover, \( \Delta _{\max } \) allows to deal only with global homogeneous geometries instead of local ones.
The compactification \(Z\) of \( \mathcal{M }_1 (G_1, H_1) \) is non-admissible, since \(T\) contains the face \(\tau \) of \(Z\).
The Minkowski sum \(\frac{z+1}{2}S+\frac{z-1}{2}S^{\prime }\) of opposite simplices \(S\) and \(S^{\prime }=-S\).
This polytope has \(2^d-2=2,6,14,\ldots \) facets, just as the classical (non-degenerate) permutohedra.
Let \(G/H\) be a Kähler homogeneous space of a simple Lie group \(G\), and let \(d>2\). Then the set of vertices of \(\Delta \) is \(\{e_i+e_j-e_k : 1\le i,j,k\le d, i\ne k,j\ne k, [i,j,k]\ne 0\}\). In this case, conditions of Tests 1 and 2 for a \(k\)-dimensional face \(\gamma \), \(0<k<d-1\) can be simplified as follows:
-
(1)
\(\gamma \) is a pyramid with apex \(a\) and base \(B\) such that if \(e_i\in \gamma \), then either \(e_i=a\) or \(e_i\in B\);
-
(2)
\(\gamma \) is a ‘\(k\)-dimensional octahedron’ with vertices \(e_{i_0} + v_p, e_{i_0}-v_p\), \(p=1,\ldots ,k\), for some linearly independent vectors \(v_p\in \mathbb{R }^d\); the face \(\gamma \) contains no points \(e_i\) with \(i\ne i_0\) (then \(\gamma \) is the intersection of all faces \(\beta \ni e_{i_0}\) of \(\Delta \)). For \(b_2(G/H)=1\), \(d>2\) there are \([d/2]\) faces \(\gamma \) satisfying (2).
-
(1)
References
Alekseevsky, D.V., Kimel’fel’d, B.N.: Structure of homogeneous Riemann spaces with zero Ricci curvature. Funct. Anal. Appl. 9, 97–102 (1975)
Besse, A.L.: Einstein Manifolds. Springer, Berlin (1987)
Bernshtein, D.N.: The number of roots of a system of equations. Funktsional. Anal. i Prilozhen. 9(3), 1975, pp. 1–4. English transl. Funct. Anal. Appl. 9(3), 183–185 (1975)
Boehm, M., Wang, C., Ziller, W.: A variational approach for compact homogeneous Einstein manifolds. GAFA 14, 681–733 (2004)
Boyer, C.P., Galicki, K.: Sasakian geometry. In: Oxford Mathematical Monographs. Oxford University Press, Oxford (2008)
Chrysikos, I., Sakane, Y.: On the classification of homogeneous Einstein metrics on generalized flag manifolds with \(b_2(M)=1\). arXiv:1206.1306v1
Fulton, W.: Introduction to Toric Varieties. Princeton University Press, Princeton (1993)
Graev, M.M.: On the number of invariant Einstein metrics on a compact homogeneous space, Newton polytopes and contractions of Lie algebras. Int. J. G. M. Mod. Phys. 3(5 & 6), 1047–1075 (2006)
Graev, M.M.: The number of invariant Einstein metrics on a homogeneous space, Newton polytopes and contractions of Lie algebras. Izvestiya RAN: Ser. Mat. 71(2), 29–88 (2007). English transl., Izvestiya: Mathematics 71(2), 247–306 (2007)
Heber, J.: Noncompact homogeneous Einstein spaces. Invent. Math. 133, 279–352 (1998)
Jensen, G.: The scalar curvature of left-invariant Riemannian metrics. Indiana Univ. Math. J. 20(12), 1125–1144 (1971)
Lauret, J.: Convergence of homogeneous manifolds. arXiv:1105.2082
Postnikov, A.: Permutohedra, associahedra, and beyond, arXiv:math.CO/0507163
Stanley, R.P.: Enumerative Combinatorics, vol. 2. Cambridge University Press, Cambridge (1999)
Acknowledgments
I am grateful to all participants of the Postnikov seminar of Moscow State University, listening to the presentation of this work May 20, 2009 and March 23, 2011.
Note added in proof
In the second version of the preprint [6] (2012-10-10) the positive definite Einstein metricson \(E_8/T^1\cdot A_4\cdot A_2\cdot A_1\) are calculated (five metrics up to scale). Unfortunately, in this version authors remove a part of the text after Eq. (25) about Einstein metrics on \(E_8/T^1\cdot A_3\cdot A_4\) which I had used above.
Author information
Authors and Affiliations
Corresponding author
Additional information
M. M. Graev: Deceased.
Supported by RFBR, grant 10-01-00041a.
Appendix: Case of Kähler homogeneous space with \(b_2=1\)
Appendix: Case of Kähler homogeneous space with \(b_2=1\)
Consider now a Kähler homogeneous space \(G/H\) with the second Betti number \(b_2=1\). Assume that the isotropy \(H\)-module \(\mathfrak{g }/\mathfrak{h }\) is split into \( d>1 \) irreducible submodules.
Lemma 8
Given a Kähler homogeneous space \(G/H\) with \(d>1\) of a simple Lie group \(G\), then \(2^{-b_2(G/H)}\nu (G/H) \in \mathbb{Z }\).
Idea of proof
\(2^{b_2(G/H)} = [\mathbb{Z }^d:L]\), where \(L \subset \mathbb{Z }^d\) is the subgroup generated by vertices of the polytope \(\Delta \). (Remark that \(\Delta = \Delta _\mathrm{min}=\Delta _{\max }\)). \(\square \)
Let, moreover, \(b_2(G/H)=1\). Then \(2\le d \le 6\). For \(d=2\) the polytope \(\Delta \) is the segment with ends \(e_2\) and \(2e_1-e_2\). For \(3\le d \le 6\) the vertices of the polytope \(\Delta \) are the points \(e_i+e_j-e_k \in \mathbb{R }^d\) with \(1\le i,j,k \le d\), \(i\ne k\), \(j\ne k\), \(i\pm j\pm k =0\). Here \(e_1=(1,0,\ldots ,0),\ldots ,e_d=(0,\ldots ,0,1)\). Using MAPLE, one can triangulate these polytopes and find their normalized volumes \(\nu =\nu (G/H)\). Thus, we obtain:
Claim
\( 2^{-1}\nu \in \{1,3,10,41,172\}. \)
The following table gives some information about polytopes \(\Delta \) corresponding to Kähler homogeneous spaces \(G/H\) with the second Betti number \(b_2=1\) and \(d>1\).
For completeness, we find the volume of a similar \((d-1)\)-dimensional polytope with \(d=7\). Here \(f\) is the number of facets of \(\Delta \), and \(m\) the number of all faces \(\gamma \) of \(\Delta \) with \(0<\dim (\gamma )<d-1\) (which we call marked) that NOT satisfy conditions of Test 1 or Test 2 of [9, Section 7.1]. A marked face \(\gamma \) is not a vertex. Footnote 5 Moreover, one can prove that \(\gamma \) is not an edge, so \(1<\dim (\gamma )<\dim (\Delta )\).
We write also the known numbers \(\varepsilon \) and \(\delta = \nu -\varepsilon \). By [9], if \(m=0\), then \(\nu =\varepsilon \). The first non-trivial case is \(d=4\).
For \(d\le 5\) all the positive solutions of the algebraic Einstein equations are known.
In the case \(d=5\) they calculated by Chrysikos and Sakane [6].
They also prove that all the complex solution are isolated (\(d=5\)).
Remark
(the case \(d=4,\;\nu =20\)) (Fig. 4). Here \(\Delta \) is a three-dimensional polytope with three marked faces \(\gamma ,\) namely, a trapezoid \(\gamma _1\), a parallelogram \(\gamma _2\), and a pentagon \(\gamma _3\). To prove that \(\nu =\varepsilon \) one can associate with each marked face \(\gamma \) a complex hypersurface \(s_{\gamma }(x_1,\ldots ,x_4)=0\) in \((\mathbb{C }\setminus 0)^4\) and check that it is non-singular. Here \(s(x)\) is the above Laurent polynomial (scalar curvature), and \(s_{\gamma }(x)\) is the sum of all monomials of \(s(x)\) whose vector exponents belong to \(\gamma \). See [8, 9, Section 1.7.2]. This is essentially a two-dimensional problem (\(2=\dim (\gamma _i)\)), i.e., we must check that a plane curve is non-singular. It is easy to prove this for \(\gamma =\gamma _1\), but for \(\gamma _k\), \(k=2,3\) the problem reduces to \(D_k[s]\ne 0\), where \(D_k[s]\) is a homogeneous polynomial (a \(k\)-monomial) in coefficients of \(s(x)\) with \(\deg (D_k) = k\). The coefficients of \(s(x)\) depend on \(G/H\). There are four Kähler homogeneous spaces \(G/H\) with \(b_2=1\) and \(d=4\), namely, the spaces \( {E_8/T^1\cdot A_1\cdot A_6},\quad {E_8/T^1\cdot A_2\cdot D_5},\quad {E_7/T^1\cdot A_1\cdot A_2\cdot A_3},\quad {F_4/T^1\cdot \widetilde{A}_1\cdot A_2} \) (we use the Dynkin’s notation \(\widetilde{A}_1\) for the three-dimensional subgroup associated with a short simple root). The corresponding scalar curvature polynomials \(s(x)\) are computed by A.Arvanitoyeorgos and I.Chrysikos (arXiv:0904.1690). For each of them one can check that \(D_2[s]D_3[s]\ne 0\). This proves that \(\varepsilon =\nu \) for \(d=4\).
The case \(d=5\), \(\nu =82\). There is a unique Kähler homogeneous space \(G/H\) with \(b_2(G/H)=1\) and \(d=5\), namely, the space \({G/H=E_8/T^1\cdot A_3\cdot A_4}.\) By [6, Section 3, the text after eq.(25)] it implies that the algebraic Einstein equation has, up to scale, \(81\) complex solutions, corresponding to roots of some polynomial \((x_5-5)h_1(x_5)\) of degree \(81\) in one variable \(x_5\). There exist six positive solutions [6, Theorem A]; in particular, the root \(x_5=5\) corresponds to a unique, up to scale, invariant Kähler metric on \(G/H\). Using MAPLE, one can check that the polynomial \(h_1(x_5)\) has \(80\) simple roots, and \(81\) solutions of Einstein equation are distinct (moreover, it has \(30\) real roots, and Einstein equation has \(31\) real solutions). Thus
We prove independently that \(\nu > \varepsilon \). Let \(s(x_1, \dots ,x_5)\) be the scalar curvature of a invariant metric \(g_x\), as above. It is a Laurent polynomial in \(x_i^{-1}\). We claim that there exists a limit
and the homogeneous function \(s_{\infty }\) depends essentially on \(3\) variables. Indeed, it follows from Proposition 4 and the above description of the Newton polytope \(\Delta \) that
Then there is a two-dimensional face of the polytope \(\Delta \), the parallelogram \(P\) with vertices \(e_4+e_5-e_1\), \(e_1+e_4-e_3\), \(2e_1-e_4\), and \(e_3+e_5-e_2\). The face \(P\) is orthogonal to the vector
According to [6, Proposition 7] we have \( [1, 1, 2]=12, [1, 2, 3]=8, [1, 3, 4]=4, [1, 4, 5]=4/3, [2, 2, 4]=4, [2, 3, 5]=2 \). Then the product of monomials, corresponding to each pair of opposite vertices of \(P\), coincides with \(2 {\frac{{x_{2}}}{{x_{1}}\,{x_{4}}\,{x_{5}}}}\), and \(s_{\infty }\) can be represented as
where \(z_0=-\frac{{x_{2}}}{{x_{3}}\,{x_{5}}}\). Since for \(z_1=z_2=-1\) we have
the complex hypersurface \(s_{\infty }(x)=0\) has a singular point \(x\) with \(\prod x_i\ne 0\).
By [8, 9, Section 1.7.2] this implies that \(\nu - \varepsilon >0\).
Note that \(\Delta \) has \(m-1=12\) marked faces, other than \(P\); namely, \(6\) three-dimensional faces with normal vectors
and \(6\) parallelograms defined by the following normal vectors (such as \(\mathbf{f}=(2, 4, 3, 1, 1)\)):
The corresponding \(12\) complex hypersurfaces are non-singular.
Additional remarks (the case \(d=5\)). Consider \((z_0,z_1,z_2)=(1,-1,-1)\) as a point \(p\) in the four-dimensional toric variety \(\Delta ^{\mathbb{C }}\). Let \(O\subset \Delta ^{\mathbb{C }}\) be the orbit of the group \((\mathbb{C }\setminus 0)^5/\mathbb{C }^\times \) through \(p\). The closure of \(O\) is the two-dimensional toric subvariety \(P^{\mathbb{C }}\). The point \(p\in O\) is a solution at infinity (in the sence of Sect. 10) of the algebraic Einstein equation.
Our example is excellent as the following lemma shows.
Lemma 9
We claim now that \(\Delta ^{\mathbb{C }}\) is smooth at each point \(q\in O\). Moreover, assuming \(\varphi :\Delta ^{\mathbb{C }}\rightarrow \mathbb{P }^{N-1}(\mathbb C )\) be the natural map into the complex projective space \(\mathbb{P }^{N-1}(\mathbb C )\), \(N=\#(\mathbb{Z }^5\cap \Delta ) \), then \(\varphi ^{-1}(\varphi (q))=\{q\}\), and \(\varphi (\Delta ^{\mathbb{C }})\) is smooth at \(\varphi (q)\).
We will apply the localization along \(O\) to prove that the point \(p\) is an isolated solution (at infinity) with the multiplicity \(1\) of the algebraic Einstein equation.
Proof
Let \(v_0,v_1,v_{12},v_2\) are vertices of the parallelogram \(P\), so \(v_0+v_{12}=v_1+v_2\), and
The set of vectors \(\{u_i : i=1,\ldots ,5\}\), and, hence, \(\{v_0,v_1,v_2,u_4,u_5\}\) are biases in \(\mathbb{Z }^5=\bigoplus \mathbb{Z }e_i\). Let \(\pi (a):=(a_4,a_5)\) for each \(a=\sum a_i u_i \in \mathbb{Z }^5\). We prove, that \(\pi (\mathbb{Z }^5\cap \Delta )\) generates the semigroup \(\mathbb{Z }_+^2\). The face \(P\) of \(\Delta \) is the intersection of two facets with normal vectors \(\mathbf{f}_i\), \(i=1,2\), so that
For any \(a\in \mathbb{Z }^5\cap \Delta \) we have \(a_4 = \langle \mathbf{f}_1,a\rangle \ge 0\), and \(a_5 = \langle \mathbf{f}_2,a\rangle \ge 0\). Then \(\pi (a)\in \mathbb{Z }_+^2\). This prove the assertion, since \(\pi (e_4) = (1,0),\;\pi (e_5)=(0,1)\), \(e_4,e_5\in \Delta \). The lemma follows \(\square \)
Now let \((z_0,z_1,z_2,y_1,y_2)\) be coordinates on \((\mathbb{C }\setminus 0)^3\times \mathbb{C }^2\). Assume that for \(y_1y_2\ne 0\)
so
Then
\(s\) is a polynomial in \(y_1\), \(y_2\), and
where \([2]\) denotes the terms with degree \(\ge 2\). Similarly, for \(s_i = x_i\partial s/\partial x_i\), \(i=1,\ldots ,5\) we have
where \([2]\) denotes \((y_1^2, y_1y_2, y_2^2)\). Computing the matrix \(J= \frac{\partial (s_1,s_2,s_3,s_4,s_5)}{\partial (z_1,z_2,y_1,y_2)}\), setting \(z_0=1,z_1=z_2=-1\), \(y_1=y_2=0\), adding the row \((d_1,\ldots ,d_5)\) of dimensions \(d_i=\dim (\mathfrak{m }_i)\), and finding the determinant, we obtain
Then the solution \(p\in \Delta ^{\mathbb{C }}\) of the algebraic Einstein equation with local coordinates \(z_0=1,z_1=z_2=-1, y_1=y_2=0\) is isolated, and non-degenerate.
The case \(d=6\), \(\nu = 344\). There is a unique Kähler homogeneous space \(G/H\) with \(b_2(G/H)=1\) and \(d=6:\)
The corresponding \(5\)-dimensional polytope \(\Delta \) in \(\mathbb{R }^6\) has \(36\) facets, i.e., \(4\)-dimensional faces. Each of them can be defined by the orthogonal vector \(\mathbf{f}=(y_1, \dots ,y_6)\) such that \(\gcd (y_1, \dots ,y_6)=1\), and \(y_i\ge 0\); then \(\langle \mathbf{f},x \rangle \,\ge 0\) for any \(x \in \Delta \).
For example, the vector \(\mathbf{f}=(1, 2, 3, 4, 5, 6)\) is orthogonal to the facet with \(9\) vertices
where \(\mathrm{e}^k_{ij}=e_i+e_j-e_k\). We write all the facets:
(1) \(16\) four-dimensional simplices with normal vectors
(2) \(8\) four-dimensional pyramids with normal vectors
(3) \(5\) other facets with normal vectors with positive entries:
(4) \(7\) facets with normal vectors with non-negative entries:
Facets (1) and (2) are not marked faces. E.g., simplices (1) and its sub-faces satisfy [9, Test 7.1].
Facets (3) and (4) are marked faces. There are \(13\) three-dimensional and \(15\) two-dimensional marked faces.
We get \(13\) vectors, orthogonal to three-dimensional marked faces (each of them is proportional to the sum of two distinct vectors (2)–(4)):
In the two-dimensional case we obtain:
(a) \(6\) parallelograms with normal vectors
(b) \(9\) parallelograms with normal vectors
Each of parallelograms listed in (a) and (b) [with the exception of two last entries in (b)] belongs exactly to \(3\) facets.
We claim now, that \(6\) marked parallelograms (a) correspond to singular complex hypersurfaces as above (consequently \(\varepsilon <\nu \)), and \(9\) marked parallelograms (b) correspond to non-singular hypersurfaces. For the proof, one can calculate \(6+9\) determinants \(\left| \begin{array}{ll}a&{}b\\ b^{\prime }&{}a^{\prime }\end{array}\right| \), where \(a,a^{\prime },b,b^{\prime }\) are some coefficients of \(s(x_1,\ldots ,x_6)\), using equalities ([6, Prop.13]): \( [1, 1, 2] = 8, [1, 2, 3] = 6, [1, 3, 4] = 4, [1, 4, 5] = 2, [1, 5, 6] = 1, [2, 2, 4] = 6, [2, 3, 5] = 2, [2, 4, 6] = 2, [3, 3, 6] = 2 \).
Thus we may unmark \(9\) of \(40\) marked faces.
Corollary
The hypothesis that all complex solutions of the algebraic Einstein equation on \( G/H = E_8/T^1\cdot A_4\cdot A_2\cdot A_1 \) are isolated reduces to examination of \(31 = 40-9\) cases, corresponding to \(12\) four-dimensional, \(13\) three-dimensional, and \(6\) two-dimensional faces of the polytope \(\Delta \).
In each case we may unmark the \(k\)-dimensional face, if the corresponding complex hypersurface is non-singular (this is the \(k\)-dimensional problem, \(k<5\)); otherwise we must examine Einstein equation in a neighborood \(U \subset \Delta ^{\mathbb{C }}\) of the ‘solution at infinity’ defined by each singular point (cf. Sect. 10).
Rights and permissions
About this article
Cite this article
Graev, M.M. On the compactness of the set of invariant Einstein metrics. Ann Glob Anal Geom 44, 471–500 (2013). https://doi.org/10.1007/s10455-013-9377-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10455-013-9377-x