Theorem Parameter interchange and inverse bar transforms
✓ theorem proved

Parameter interchange and inverse bar transforms

For normalized Hecke operators over \(C\), \[ B_aL_s(-a)=L_s(a)^{-1}B_a,\qquad B_aB_{-a}=I. \] For the unnormalized operators and a unit \(q\), \[ B_qL_s(q^{-1})=L_s(q)^{-1}B_q,\qquad B_qB_{q^{-1}}=I. \]

Proof

Step 1. Compute an ascent of an inverse-basis vector

Suppose \(\ell(sw)=\ell(w)+1\). Inversion preserves length, so appending \(s\) to \(w^{-1}\) is a right ascent. The standard-action identity gives \[ U_{(sw)^{-1}}=U_{w^{-1}s}=U_{w^{-1}}L_s. \] Take inverses and apply to \(e_1\): \[ b_{sw}=L_s^{-1}U_{w^{-1}}^{-1}e_1=L_s^{-1}b_w. \] This applies in either convention.

Step 2. Compute the normalized descent case

Now suppose \(sw<w\). Step 1 at \(sw\) gives \(b_w(a)=L_s(a)^{-1}b_{sw}(a)\), hence \(L_s(a)b_w(a)=b_{sw}(a)\). The inverse formula therefore yields \[ L_s(a)^{-1}b_w(a)=b_{sw}(a)-ab_w(a). \] Combining the ascent and descent cases, \[ L_s(a)^{-1}b_w(a) =b_{sw}(a)-a\mathbf1_{sw<w}b_w(a). \] This is exactly \(B_a(L_s(-a)e_w)\), by the normalized basis formula and linearity of \(B_a\). Equality on basis vectors proves the first normalized operator identity.

Step 3. Compute the unnormalized descent case

For \(sw<w\), the same ascent argument gives \(L_s(q)b_w(q)=b_{sw}(q)\). Thus \[ \begin{aligned} L_s(q)^{-1}b_w(q) &=q^{-1}\bigl(b_{sw}(q)-(q-1)b_w(q)\bigr)\\ &=q^{-1}b_{sw}(q)+(q^{-1}-1)b_w(q). \end{aligned} \] This is \(B_q(L_s(q^{-1})e_w)\) in the descent branch. In the ascent branch both expressions equal \(b_{sw}(q)\), by Step 1. This proves the unnormalized intertwining identity.

Step 4. Intertwine the inverse generators

The normalized identity \(B_aL_s(-a)=L_s(a)^{-1}B_a\) can be multiplied on the left by \(L_s(a)\) and on the right by \(L_s(-a)^{-1}\). It becomes \[ B_aL_s(-a)^{-1}=L_s(a)B_a. \] The same calculation with the unnormalized identity gives \[ B_qL_s(q^{-1})^{-1}=L_s(q)B_q. \] Both multiplications use the two-sided inverses proved by the quadratic identity.

Step 5. Prove inverse composition by reduced-word induction

For normalized parameters, prove \(B_ab_w(-a)=e_w\) by induction on a reduced word for \(w\). At the empty word, \(b_1=e_1\) and \(B_ae_1=e_1\).

Write a nonempty reduced word as \(s\omega\), with tail product \(u\). Then \(su\) is an ascent. By Steps 1 and 4 and the induction hypothesis, \[ B_ab_{su}(-a) =B_aL_s(-a)^{-1}b_u(-a) =L_s(a)B_ab_u(-a) =L_s(a)e_u=e_{su}. \] The last equality is the ascent formula.

For the unnormalized case, replace \(-a\) by \(q^{-1}\) and use the second identity in Step 4. The identical induction gives \(B_qb_w(q^{-1})=e_w\).

Step 6. Conclude both-sided invertibility

Since \(B_{-a}e_w=b_w(-a)\), Step 5 proves \(B_aB_{-a}=I\). Replacing \(a\) by \(-a\) proves the reverse composite.

Likewise \(B_qB_{q^{-1}}=I\), and replacing \(q\) by \(q^{-1}\) proves \(B_{q^{-1}}B_q=I\). Thus the two parameter transforms are mutually inverse.

On \(T_s\), the calculation is \[ B_{q^{-1}}T_s=qT_s+(q-1)T_1, \] \[ B_q\bigl(qT_s+(q-1)T_1\bigr) =T_s+(1-q)T_1+(q-1)T_1=T_s. \]

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.Hecke.barBasis_simple_mul_ascent: Hecke.lean:309.

OAI.KLInvariance.Hecke.leftInv_barBasis: Hecke.lean:322.

OAI.KLInvariance.Hecke.barLinear_left: Hecke.lean:348.

OAI.KLInvariance.Hecke.barLinear_leftInv: Hecke.lean:358.

OAI.KLInvariance.Hecke.barLinear_barBasis: Hecke.lean:367.

OAI.KLInvariance.Hecke.barLinear_inverse: Hecke.lean:381.

OAI.KLInvariance.HeckeQ.barBasis_simple_mul_ascent: Hecke.lean:796.

OAI.KLInvariance.HeckeQ.leftInv_barBasis: Hecke.lean:810.

OAI.KLInvariance.HeckeQ.barLinear_left: Hecke.lean:841.

OAI.KLInvariance.HeckeQ.barLinear_leftInv: Hecke.lean:848.

OAI.KLInvariance.HeckeQ.barLinear_barBasis: Hecke.lean:862.

OAI.KLInvariance.HeckeQ.barLinear_inverse: Hecke.lean:876.

Used by