A Schubert basis in equivariant elliptic cohomology (Q2360799)

Let \(G\) be a connected complex semisimple Lie group and \(B\) a Borel subgroup, and consider the corresponding generalized flag manifold \(G/B\) (with maximal torus \(T\).) If \(E^*(\cdot)\) is a complex oriented cohomology theory, one can obtain in certain cases topological Schubert classes, which are elements in \(E^*(G/B)\). This is true for singular cohomology and \(K\)-theory, and has also been done for their torus-equivariant versions \(E_T^*(\cdot)\), but not much is known for other theories. This work is concerned with defining Schubert classes for the (\(T\)-equivariant) algebraic oriented cohomology theory corresponding to the formal group law of a singular cubic curve in Weierstrass form (here, the Levine-Morel context of algebraic oriented cohomology theories is implicit, which expands on the concept of complex oriented cohomology theories.) The authors consider this, which will be denoted by \(\mathcal{SE}^*(\cdot)\) and is called \textit{elliptic cohomology}, the first interesting case, in terms of complexity, after \(K\)-theory, and had already dealt with the corresponding Schubert classes in a previous paper. The new results here have to do with a different definition of these Schubert classes that is independent of certain choices, as seen below. Let \(W\) be the Weyl group of \(G\) with respect to \(T\). For each \(w \in W\), define the Schubert variety \(X(w)\) by \(\overline{BwB/B}\). If we consider a reduced word \(\Gamma_w\) for \(w\) and a Bott-Samelson resolution for the Schubert variety, namely \(\Gamma_{I_w} \rightarrow X(w) \rightarrow G/B\), we get a so-called Bott-Samelson class \(\zeta_{I_w}\) in \(E^*_T (G/B)\), for any theory as before, via push-forwarding \(1 \in \Gamma_{I_w}\). If the theory is singular cohomology or \(K\)-theory, this class is independent of the chosen reduced word (and all the \(\zeta_{I_w}\), for \(w \in W\), form a basis for \(E^*_T (G/B)\) over the corresponding formal group algebra \(S\).) This is not the case for other cohomology theories, including \(\mathcal{SE}^*(\cdot)\). To circumvent this difficulty, the authors propose a new way of defining the Schubert classes. To calculate Bott-Samelson classes, one can consider the recursive formula that appears here on page 717, giving \(\zeta_{I_w}\) as \(Y_{I_1} \cdots Y_{I_l} \zeta_{\emptyset}\), the successive action of elements \(Y_{I_j}\) on the identity element of \(E^*_T (G/B)\) (we consider here that \(I_w = (i_1, \cdots , i_l)\) is a reduced word for \(w\).) These elements belong to the formal Demazure algebra for \(W\), and the action is on the Borel model for \(E^*_T (G/B)\) (details in section 2.3.) The approach of the authors is to modify each operator \(Y_{I_w} = Y_{I_1} \cdots Y_{I_l}\) so that it will not depend on the reduced word chosen for \(w\). This is done by working with the Kazhdan-Lusztig basis for the Hecke algebra corresponding to \(W\), and in definition 3.5. elements \(\mathfrak{G}_w \in \mathcal{SE}^*(G/B)\) are proposed and called \textit{Kazhdan-Lusztig Schubert classes}. Varying \(w \in W\), we obtain a basis for \(\mathcal{SE}^*(G/B)\) (Corollary 3.6.) and the previous Bott-Samelson classes \(\zeta_{I_w}\) are seen to be limits of the \(\mathfrak{G}_w\) (Corollary 3.6.) With this new definition, two conjectures are proposed. First, a positivity property generalizing the one in Graham's formula for \(K\)-theory (conjecture 3.9.); secondly, the agreement of \(\mathfrak{G}_w\) with the topologically defined Schubert classes \([X(w)]\) (from section 3) whenever the Schubert variety \(X(w)\) is smooth (conjecture 3.12.) This last conjecture is proved, in theorem 3.14., for some cases, including some \(W\) with root systems of type \(A_n\) or \(C_n\).