Quasi-isometric embeddings of symmetric spaces (Q1637319)

Gromov proposed the program of classification of finitely generated groups up to quasi-isometry. Due to the work of several mathematicians, rigidity results are known. For instance, a finitely generated group that is quasi-isometric to an irreducible lattice in a semisimple Lie group \(G\) is `almost' a lattice (it is an extension of a lattice in \(G\) by a finite group). The authors of this paper who have contributed substantially to the study of quasi-isometry have begun in the present paper a study of quasi-isometric embeddings of symmetric spaces. The following version of Mostow rigidity is known: Two symmetric spaces (without compact or Euclidean factors) are quasi-isometric if and only if they are isometric (after a possible rescaling). In his survey article [Math. Res. Lett. 4, No. 5, 705--717 (1997; Zbl 0889.22010)], \textit{B. Farb} raised three questions for further study. One of them asks whether quasi-isometric embeddings of one symmetric space in another must be at a bounded distance from the inclusion of a totally geodesic symmetric subspace. In the paper under review, the authors make the first significant progress on the above question for symmetric spaces without compact or Euclidean factors. They observe that exotic embeddings exist even if the domain and the range have equal rank. For example, for any \(r>1\), there are quasi-isometric embeddings of the rank \(r\) space \(\mathrm{SL}_{r+1}(\mathbb{R})/O(r+1)\) into the rank \(r\) space \(\mathrm{Sp}_{2r}(\mathbb{R})/U(2r)\) - note that there is no isometric embedding in this case. The authors make a general construction based on the following result they prove: Let \(G_1\) and \(G_2\) be semisimple Lie groups of equal real rank. Let \(G_i = K_iA_iN_i\) (\(i=1,2\)) be their Iwasawa decompositions. Then, every injective homomorphism of the solvable Lie group \(A_1N_1\) as a subgroup of \(A_2N_2\) is a quasi-isometric embedding. The above result is important as a tool since every symmetric space \(K \backslash G\) is isometric to the subgroup \(AN\) of \(G\). The authors prove a variety of results with this tool. In particular, they prove: There is a quasi-isometric embedding of \((\mathbb{H}^2)^n\) into \(\mathrm{SL}_{n+1}(\mathbb{R})/O(n+1)\). Moreover, there are quasi-isometric embeddings of \(\prod_{i=1}^k \mathrm{SL}_{n_i+1}(\mathbb{R})\) into \(\mathrm{SL}_{n+1}(\mathbb{R})\) where \(\sum_{i=1}^k n_i=n\). The main rigidity result proved in the paper is as follows: Let \(X_1,X_2\) be irreducible symmetric spaces or Euclidean buildings, both of rank \(r>1\). Let \(C_1,C_2\) be their Weyl patterns on \(\mathbb{R}^r\). Then: (i) If there are no elements of \(GL_r(\mathbb{R})\) embedding \(C_1\) into \(C_2\), then there are no quasi-isometric embeddings of \(X_1\) into \(X_2\). (ii) For a fixed \(K>1\), if all \(K\)-quasiconformal embeddings of \(C_1\) into \(C_2\) are conformal, then all \((K,C)\)-quasi-isometric embeddings of \(X_1\) into \(X_2\) are at a bounded distance from a totally geodesic embedding. In particular, if all pattern embeddings are conformal, then all quasi-isometric embeddings are at a bounded distance from a totally geodesic subspace. Finally, the authors raise interesting, pertinent questions as a guide towards further research in this direction.