WorksMathematics
A counterexample to a stability question for unramified multiplicity one symmetric pairs
Abstract
We construct a split reductive symmetric pair over with multiplicity at most one for every irreducible unramified representation and which is not a stable symmetric pair.
Introduction
Let be a -adic local field of characteristic zero with ring of integers . For a symmetric pair , fix a hyperspecial compact subgroup . An irreducible smooth complex representation is unramified with respect to if . We say that the pair has unramified multiplicity one if
for every such . Here acts by right translation.
Aizenbud and Gourevitch call a symmetric pair satisfying Definition 3.2 a good pair [1 Definition 7.1.6]. We use Carmeli’s term stable for the same property [2]. We consider the following question.
Does unramified multiplicity one imply stability of a symmetric pair? $$ {#eq:1.1}
We take
Let , with coordinates , and put
Write for the -th coordinate projection. Define using the characters
We introduce the group as
The involution
has fixed subgroup .
Theorem 1.1. The symmetric pair defined by (1.5)–(1.6) has unramified multiplicity one (Proposition 2.3) and fails stability (Proposition 3.6).
Unramified multiplicity one
Write and let be the torus projection. For a unit in a coefficient ring set . The homomorphism
splits over and restricts to a section . The kernels of and are respectively and its blockwise diagonal copy . These descriptions give smooth group schemes with connected reductive geometric fibers. Indeed, the semidirect product descriptions give smoothness and connectedness, and in each fiber a connected normal unipotent subgroup maps trivially to the torus and hence lies in the semisimple kernel. The diagonal matrices give a split maximal torus in each group, since in each matrix factor the diagonal torus of already has diagonal centralizer. In particular is split reductive and is hyperspecial.
For context we recall the general criterion of Sakellaridis. If is a homogeneous affine spherical variety for a connected split reductive group , write for the unramified character torus of a maximal split torus of , and use , , , and for the corresponding objects of [4 §8]. The restriction map in that criterion is
Theorem 2.1 ([4 Corollary 8.0.4]). Assume the standing hypotheses of [4 §§1.1, 1.7, 8], including its integral orbit axioms and the single rational open Borel orbit hypothesis, and assume Statements 6.3.1 and 7.1.5 of loc. cit. for the spherical roots and virtual weighted colors. If the restriction map is surjective, then every irreducible unramified representation of satisfies
Proof. Let be the hyperspecial subgroup and its spherical Hecke algebra. By [4 Corollary 8.0.4(1)], is cyclic over . Pair it with by integration against an invariant measure. This pairing separates functions, since the compact open -orbits have positive volume. If is a simultaneous eigenfunction with character , its pairing with applied to a cyclic generator is its pairing with the generator multiplied by , where . Thus each eigenspace has dimension at most one. An -equivariant map out of is determined by the image of a nonzero spherical vector, which spans and must map into the eigenspace for its Hecke character. □
For the pair (1.5)–(1.6) we give a direct proof. It rests on a choice of double-coset representatives for which averaging a spherical vector determines every -invariant functional from its value on that vector.
Lemma 2.2. Every double coset in has a representative such that .
Proof. Put and let be the valuation with . For the class of modulo is determined by the two integers and . Indeed, division by an element of allows us to set equal to , and removes the remaining unit factors. Thus, using on the left and on the right, we may take the torus component to be
The last coordinate may be chosen freely because contains the last coordinate subtorus. The valuations of the six characters in (1.2)–(1.4) are
In each column at least one entry is even. For even this holds throughout the first row. For both odd only its first entry is odd, and is even. In the remaining case , the only odd entries of the first row are in columns 3 and 5, where the second row has even entries.
Write the matrix components as . Since has pure uniformizer powers as coordinates and , at least one matrix in each pair has determinant for some . If that matrix is , multiplication on the left by on both blocks is allowed in the diagonal and makes scalar. We can also make the other block diagonal. Indeed, elementary divisors over give with diagonal for the other block . Multiplying on the left and on the right by diagonal unit matrices makes both determinants one and preserves diagonality. Apply on the left of both blocks, then on the right of the scalar block and on the right of the other. These operations belong to and , respectively. Denote the resulting representative by .
The subgroup commutes with , since all the matrices are diagonal. As , it remains to move a pair past the th block . If is scalar, then
If is scalar, then
In each identity the left factor belongs to the diagonal subgroup and the right factor is integral of determinant one. Taking the product over proves the inclusion. □
Proposition 2.3. The symmetric pair defined by (1.5)–(1.6) has unramified multiplicity one with respect to .
Proof. Let be irreducible and -spherical. By the spherical representation theorem, for any nonzero spherical vector [3 §6]. Fix . For as in Lemma 2.2, the function is constant on , since it is left -invariant and right -invariant. Give volume one. Then
where the averaged vector lies in and equals . The integral is a finite sum after passage to a sufficiently small open subgroup of . If , the matrix coefficient therefore vanishes on every . The -translates of span , so . Evaluation at thus embeds in . Finally, evaluation of a -map at the identity coset identifies the Hom space in the introduction with . The inverse sends to . □
Verification of Sakellaridis’s criterion
Proposition 2.4. For and the integral model of over , all the hypotheses of Theorem 2.1 hold. In particular , and the restriction
is surjective. Consequently Theorem 2.1 also gives Proposition 2.3.
Proof. The open orbit and its integral model. Take the Borel lower triangular in the first matrix of each pair and upper triangular in the second. Write for its diagonal torus. Identify with by . Define six characters of by
Indeed, . Over the map identifies with
To see this over any -algebra, lift to , set , and take . This gives a section, and two lifts differ by a unique left multiplier in . In particular . The model is smooth and affine, since multiplying on the left by identifies it as a scheme with . The derived group of is and is simply connected.
The open orbit through is given by nonvanishing of all proper leading principal minors of the . Gaussian elimination factors each matrix there as lower unitriangular, diagonal, and upper unitriangular. The diagonal locus is : the section just constructed takes diagonal tuples to . Thus the open orbit is a product of with the two families of unitriangular groups over . It is smooth, has homogeneous special fiber, and its -points are one -orbit. The group and the stabilizer are unimodular, so has an invariant measure. The open stabilizer is a split torus, and is trivial on it because the two triangular factors have opposite roots. This gives a -invariant measure on the open orbit. Every simple minimal parabolic enlarges that orbit, so , its Levi is , and .
Bruhat strata and Axiom 2.2.1. In (2.1) the -strata are indexed by tuples : the th matrix lies in . Here are the triangular Borels of , and we take signed permutation representatives of determinant one. The torus is free to vary. Indeed, the diagonal factor used to trivialize (2.1) preserves these cells, the section moves between fibers, and the determinant-one triangular groups act transitively on each cell with fixed determinant. The integral cells are products of split tori and affine spaces by Bruhat elimination. Their frontiers are the usual unions of Bruhat cells, defined by the same northwest submatrix rank conditions over and in both fibers. In particular the boundaries, also for a union of two cells related by a minimal parabolic, acquire no extra special-fiber stratum: every boundary cell is flat over and its generic fiber is dense in it.
To identify the rank-one quotient, take the representative of a stratum and enlarge one triangular factor to . The stabilizer projected to preserves one coordinate line: the other flag in the same block is a full coordinate flag in permutation order, and it singles out one of the two lines in the affected two-dimensional flag quotient. The image is its entire Borel. It contains the corresponding elementary unipotent matrices, acting on both flags with the permutation conjugacy, and the diagonal -torus maps onto the adjoint torus as a group scheme for the faithfully flat topology. These matrix descriptions work over . Thus is a split Borel quotient of , of type U in [4 §2.2]. Together with the boundary description this gives Axiom 2.2.1 and excludes all type-N edges.
The Cartan and Iwahori axioms. Let and . Write for . An element of consists of an exponent vector and two integer vectors of length in each block with
An element of consists of and integer vectors with . The projection of lattices is
where denotes the image of in . The antidominant cone consists of the in each block. Indeed for a symmetric variety it is the image of the ambient antidominant cone over [4 §2.1]. Here is decreasing and increasing in that cone. Conversely any increasing lifts rationally by taking constant with its prescribed sum and .
The invariants of a -orbit are the two valuations of and the ordered elementary-divisor exponents of every . They determine the orbit. First normalize to pure powers of using . Smith normal form in block gives integral invertible with . At the normalized the determinants of and agree. Thus , and replacing by for a diagonal unit matrix makes both determinants one while fixing . These left and right operations on are induced by the two factors of . Conversely the invariants are unchanged under . Hence is precisely the set of antidominant integral parameters just described (Axiom 2.4.1).
For Axiom 2.4.2 take any diagonal point in , with block , so that . For an element of the Iwahori write for its blocks. They are respectively lower and upper triangular modulo . Gaussian elimination over gives with integral lower triangular invertible and upper unitriangular congruent to modulo . Then is integral and upper triangular modulo , so it has a factorization with lower unitriangular and upper triangular integral invertible. Now
The left factor is integral lower triangular invertible by the order of the valuations. Its determinant is , and . Taking its inverse as the first block, as the second block, and the original torus component in produces an element of with exactly the same action on the given point. Thus for every .
Spherical roots and virtual weighted colors. On the Bruhat indices the ambient Weyl group acts separately by left and right permutations, since each simple step exchanges adjacent positions or adjacent values. The open index is the identity tuple. Its stabilizer is the blockwise diagonal , using and [4 §2.2]. The simple reflection for positions in block is the product of the corresponding left and right simple reflections. Their length-two path passes through the same codimension-one cell reached from the open orbit on either side. Thus the normalized spherical root is the sum of the two corresponding simple roots, of type G in the notation of [4 §6.2]. Its supporting Levi has derived group . Thus [4 Theorem 6.3.2] applies. In this case the rank-one computation is exactly the two-sided group computation in [4 Example 5.4.1]: the adjacent root subgroups act by left and right multiplication on the same two-dimensional matrix block. This verifies Statement 6.3.1.
For clarity, the color in this path is the divisor of the leading -rowed principal minor in block . Its valuation on is the cocharacter , with zero torus component, where denotes a standard coordinate vector in the th block. Indeed the other leading minors and the -coordinates are units at its generic point, and vanishes there to order one. For the order of vanishing, take and the identity matrices with one two-by-two block replaced by at positions , where and all the other leading minors equal one. On the diagonal locus successive quotients of the leading minors are the diagonal coordinates. These evaluations give , and the Bruhat cells of codimension one exhaust the colors. Each color is defined over , belongs to its single type-G root and has weight , as seen by lifting to the right simple coroot. The definition in [4 §7.1] therefore gives, with multiplicity one,
The cone of color valuations is generated by the adjacent differences , so the positive members have . Each simple transposition sends exactly its own positive simple coroot to a negative one. This proves both assertions of Statement 7.1.5.
Surjectivity on invariants. The coordinate rings of and are the group algebras and . Restriction sends to by (7). Fix . Lift to with chosen as in Lemma 2.2. The two valuation rows in its proof say, for each , that either or is even. In the first case choose the constant vector and put . In the other case choose and put . In each case the constant vector has length . Since and the sum of is prescribed, both determinant equations hold. These choices give a lift of with a constant vector on at least one side of every block.
The two symmetric groups of each ambient block permute independently, whereas permutes the entries of . With one vector constant, subtraction gives a bijection of the distinct ambient and relative orbit points in that block, including when entries repeat. The Weyl groups fix the torus coordinates. Taking products yields
The sums on the right form a basis of . This proves surjectivity and completes the verification of the hypotheses, independently of the multiplicity conclusion.
Stable symmetric pairs
Symmetric pairs
Definition 3.1. A symmetric pair is a triple , where and are reductive groups, is an involution of , and . We call the pair connected if is connected as an algebraic variety.
For a symmetric pair , define the anti-involution and the symmetrization map
For , we write , since . The adjoint action of preserves this space.
Definition 3.2. The pair is stable if for every whose double coset is closed in the -adic topology.
An unstable symmetric pair
In this section, we let
Choose roots of unity characterized by
and
These roots exist uniquely by Hensel’s lemma, and . Write
For and , let
In the basis of , with the class of , this is multiplication by . Its characteristic polynomial is , and .
Define
Writing , its characteristic polynomial gives and . In particular both and equal . Thus is defined over and
The six matrices used in the construction are
We set
and put
Writing as defined in the introduction, we obtain
by (8) and the characters (1.2)–(1.4). Thus .
For the remainder of this section , where . We first determine the centralizers that enter the double-coset calculation.
Lemma 3.3. Each is cyclic of degree . For it is unramified, and for it is totally ramified. The matrix has irreducible characteristic polynomial of degree . Its centralizer in is acting by multiplication, and the determinants of its elements form .
Proof. Write , , and . For , a root of over satisfies . The element has order four in , so the four roots form one Frobenius orbit. The polynomial has irreducible separable reduction, giving an unramified extension of degree four. For or the polynomial is Eisenstein at , so is totally ramified of degree and is a uniformizer. Since has order , in all cases the extension is cyclic with generator .
In the power basis of , the matrix represents multiplication by
The unique involution of sends to . It fixes because is even. If it fixed , cross-multiplication would give , which is impossible. Every nontrivial subgroup of this cyclic -group contains the involution, so the stabilizer of is trivial and .
Multiplication by therefore has irreducible minimal and characteristic polynomials of degree . A -linear map commuting with it is -linear on the one-dimensional -space , hence multiplication by an element of . The determinant of multiplication by any such element is its field norm.
Lemma 3.4. The rational double coset is closed in .
Proof. For put . Under left multiplication by the matrices are unchanged. A right multiplier with blocks sends to . Let consist of the with such that each has the characteristic polynomial of . These conditions are closed in the -adic topology and invariant under both copies of . Hence is a closed union of double cosets containing .
We show that each double coset in is relatively open. Fix and put . By Lemma 3.3 its characteristic polynomial is irreducible, so any nonzero vector is cyclic. For near the matrix
is invertible. In these cyclic bases and have the same companion matrix whenever . Consequently satisfies and tends to the identity as . For sufficiently close to , Hensel’s lemma gives an th root of in , since . Multiplying by this scalar preserves the conjugacy and gives in every block.
Right multiplication of by now gives the required relative matrices . For the left multiplier take torus component and common blocks . Their determinants are , so the multiplier belongs to . This gives itself, because . All double cosets in are thus relatively open. Their complements in are unions of relatively open double cosets, so each is also relatively closed. Since is closed in , the assertion follows.
It remains to decide whether the anti-involution preserves this closed orbit. To record the necessary norm conditions, let and let record the square class of the residue of the unit , with for a square and for a nonsquare. Both are homomorphisms from to the additive group .
Lemma 3.5. Suppose is a norm from for one of the pairs occurring in the construction. Then
| necessary condition on | |
| 3 | |
| 17 | |
| 51 |
Table 1.
Proof. If then and the extension is unramified. The valuation of a norm is divisible by four, since all four conjugates have the same integral valuation. Thus .
In the ramified cases put , , , and with . Every element of is for a unit of , where and . If , valuation gives . Total ramification means all conjugates of have the same residue in , whence
In the cyclic group of order , with dividing , an th power is an th power only if is a square. Indeed, writing as the th power of a generator gives for some , so is even. Finally is a square modulo and is a nonsquare (). For this gives , and for it gives .
Proposition 3.6. The symmetric pair defined by (1.5)–(1.6) is not stable.
Proof using Carmeli’s second obstruction. Put and . In our coordinates,
By Lemma 3.3, has separable characteristic polynomial, so the matrix components of are semisimple. Its torus component is semisimple as well. Hence is a semisimple element of , and the double coset of is closed by Lemma 3.4.
Let . Since , the involution preserves and represents a cocycle for its action on . Carmeli identifies the second obstruction with the class represented by and proves that is stable exactly when this class is neutral in [2 Propositions 4.18–4.19]. Here neutrality means that
Indeed, in the convention one takes . In the equation already has the solution .
We determine when it has a solution in . Set and . The two blocks and have the same centralizer in , and the determinant on is the field norm (Lemma 3.3). Thus an element of has , with and .
The torus component of (3.2) requires , or equivalently with . The first matrix components require , hence . It follows that (3.2) is solvable exactly when there exists a single such that
For sufficiency, choose with that determinant and set . Since and , we get . Also , so these choices solve the full equation in .
Write . The six bases of the powers in (3.3) are
Apply Lemma 3.5 in each pair of blocks. For the quotient of the bases gives , so . For the quotient gives . For it gives . These three equalities contradict one another in . Hence (3.3) has no solution and in . By Carmeli’s Proposition 4.19 the semisimple symmetrization is unstable. Thus the closed double coset of is moved by , proving the proposition. □
Proof of Theorem 1.1. Proposition 2.3 gives the multiplicity bound for every irreducible -unramified representation of . Proposition 3.6 exhibits a closed rational double coset for this same pair that is moved by . Together they give a negative answer to (1.1). □
References
- [1]A. Aizenbud and D. Gourevitch, Generalized Harish-Chandra descent, Gelfand pairs, and an Archimedean analog of Jacquet–Rallis’s theorem, Duke Math. J. 149 (2009), 509–567.
- [2]S. Carmeli, On the stability and Gelfand property of symmetric pairs, arXiv:1511.01381.
- [3]B. H. Gross, On the Satake isomorphism, in Galois Representations in Arithmetic Algebraic Geometry, Cambridge University Press, 1998, 223–238.DOI
- [4]Y. Sakellaridis, Spherical functions on spherical varieties, Amer. J. Math. 135 (2013), 1291–1381.