In other words, we see that the
$\mathfrak{g}\mathfrak{l}(1,1)$
corresponding to the pair
$(\mathfrak{g},\mathfrak{l})$
sits diagonally in the direct product of the two copies of
$\mathfrak{g}\mathfrak{l}(1,1)$
described above. Let us summarize the above discussion:
Corollary 6.7.
Assume
$(\mathfrak{g},\mathfrak{k})$
is a hermitian symmetric pair, let
$\mathfrak{q}=\mathfrak{l}\oplus \mathfrak{u}$
be a
$\theta $
stable parabolic subalgebra of
$\mathfrak{g}$
, and assume that
$\mathfrak{l}\subset \mathfrak{k}$
and
$\mathfrak{u}\supset {\mathfrak{p}}^{+}$
. Then there are two supercommuting copies of
$\mathfrak{g}\mathfrak{l}(1,1)$
inside
$U(\mathfrak{g})\otimes C(\mathfrak{s})$
. One is spanned by
${C}_{\Delta}(\mathfrak{k},\mathfrak{l})$
,
${C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})$
,
$\Delta E(\mathfrak{k},\mathfrak{l})$
and
$\Delta D(\mathfrak{k},\mathfrak{l}{)}^{2}$
and the other is spanned by
$C(\mathfrak{g},\mathfrak{k})$
,
${C}^{}(\mathfrak{g},\mathfrak{k})$
,
$E(\mathfrak{g},\mathfrak{k})$
, and
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
. The diagonal of the product of these two super subalgebras is the copy of
$\mathfrak{g}\mathfrak{l}(1,1)$
spanned by
$C$
,
${C}^{}$
,
$E$
and
${D}^{2}$
from the end of Section 2. □
7 Hodge decomposition for
$\overline{\mathfrak{u}}$
cohomology in hermitian case
In this section,
$(\mathfrak{g},\mathfrak{k})$
is a hermitian symmetric pair,
$\mathfrak{q}=\mathfrak{l}\oplus \mathfrak{u}$
is a
$\theta $
stable parabolic subalgebra of
$\mathfrak{g}$
, and we assume
$\mathfrak{l}\subset \mathfrak{k}$
and
$\mathfrak{u}\supset {\mathfrak{p}}^{+}$
.
Let
$V$
be a unitary
$(\mathfrak{g},K)$
module, and consider the form
$\langle ,\rangle $
on
$V\otimes S$
introduced in Section 4. To apply the results of Section 4, we decompose
$$V\otimes S=V\otimes {S}_{\mathfrak{p}}\otimes {S}_{\mathfrak{s}\cap \mathfrak{k}}=V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}\otimes {\wedge}^{\cdot}\mathfrak{u}\cap \mathfrak{k},$$
and embed
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
as
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}\otimes 1$
. The form
$\langle ,\rangle $
restricts to the analogous definite form on
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
.
Denote as before by
$D=D(\mathfrak{g},\mathfrak{l})$
the Dirac operator for the pair
$(\mathfrak{g},\mathfrak{l})$
and by
$C=C(\mathfrak{g},\mathfrak{l})$
and
${C}^{}={C}^{}(\mathfrak{g},\mathfrak{l})$
its parts coresponding to
$\mathfrak{u}$
and
$\overline{\mathfrak{u}}$
. By Corollary 6.7,
$C={C}_{\Delta}(\mathfrak{k},\mathfrak{l})+C(\mathfrak{g},\mathfrak{k})$
, and similarly for
${C}^{}$
. Moreover, the copy of
$\mathfrak{g}\mathfrak{l}(1,1)$
corresponding to the pair
$(\mathfrak{g},\mathfrak{k})$
supercommutes with the copy of
$\mathfrak{g}\mathfrak{l}(1,1)$
corresponding to the pair
$(\mathfrak{k},\mathfrak{l})$
.
By Corollary 4.3, the adjoints of
${C}_{\Delta}(\mathfrak{k},\mathfrak{l})$
and
$C(\mathfrak{g},\mathfrak{k})$
are respectively
${C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})$
and
${C}^{}(\mathfrak{g},\mathfrak{k})$
. So the adjoint of
$C$
is
${C}^{adj}={C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l}){C}^{}(\mathfrak{g},\mathfrak{k})$
.
We consider the positive semidefinite operator
$\Delta =C{C}^{adj}+{C}^{adj}C=[C,{C}^{adj}]$
. By the above remarks we have
$$\Delta =\left[{C}_{\Delta}\right(\mathfrak{k},\mathfrak{l})+C(\mathfrak{g},\mathfrak{k}),{C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l}){C}^{}(\mathfrak{g},\mathfrak{k}\left)\right]=\left[{C}_{\Delta}\right(\mathfrak{k},\mathfrak{l}),{C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l}\left)\right]\left[C\right(\mathfrak{g},\mathfrak{k}),{C}^{}(\mathfrak{g},\mathfrak{k}\left)\right]={D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}D(\mathfrak{g},\mathfrak{k}{)}^{2}.$$
We know from Section 4 that
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
decomposes into eigenspaces of
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
for eigenvalues
$\lambda \le 0$
. Each eigenspace is
$\stackrel{~}{K}$
invariant, and each
$\stackrel{~}{K}$
isotypic component of
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
is contained in an eigenspace. We assume
$V$
is admissible, so the eigenspaces are finitedimensional.
Passing from
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
to
$V\otimes S$
is tensoring with the finitedimensional
$\mathfrak{l}$
module
${\wedge}^{\cdot}\mathfrak{u}\cap \mathfrak{k}$
. On this last space, there is no action of
$U(\mathfrak{g})$
or
$U\left({\mathfrak{k}}_{\Delta}\right)$
. So every eigenspace of
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
on
$V\otimes {\wedge}^{\cdot}{\mathfrak{p}}^{+}$
just gets tensored with
${\wedge}^{\cdot}\mathfrak{u}\cap \mathfrak{k}$
, and this gives the eigenspace on
$V\otimes S$
for the same eigenvalue.
Since
${D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}$
commutes with
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
, it preserves these eigenspaces. Moreover, the Levi subgroup
$L\subset K$
corresponding to
$\mathfrak{l}$
is compact. So is then the double cover
$\stackrel{~}{L}$
, which acts on
$V\otimes S$
. Since
$\stackrel{~}{L}$
commutes with
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
, it also preserves its eigenspaces and hence these eigenspaces decompose into
$\stackrel{~}{L}$
irreducibles. Since
${D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}$
is up to a constant equal to the Casimir element of
${\mathfrak{l}}_{\Delta}$
, it follows that
${D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}$
diagonalizes on each eigenspace of
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
. To conclude:
Lemma 7.1.
$V\otimes S$
is a direct sum of eigenspaces for
$\Delta $
. In particular,
$V\otimes S=Ker\Delta \oplus Im\Delta $
. □
This is an analogue of Corollary 4.5 for
$\Delta $
in place of
${D}^{2}$
. Now the arguments proving Lemma 4.6 and Theorem 4.7 work without change, and we obtain
Theorem 7.2.
(a)
$Ker\Delta =KerC\cap Ker{C}^{adj}$
; (b)
$V\otimes S=Ker\Delta \oplus ImC\oplus Im{C}^{adj}$
; (c)
$KerC=Ker\Delta \oplus ImC$
; (d)
$Ker{C}^{adj}=Ker\Delta \oplus Im{C}^{adj}$
.
In other words, we have obtained a Hodge theorem for
$\overline{\mathfrak{u}}$
cohomology.
To obtain it also for
$\mathfrak{u}$
homology, we note that
$({C}^{}{)}^{adj}=({C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})+{C}^{}(\mathfrak{g},\mathfrak{k}){)}^{adj}={C}_{\Delta}(\mathfrak{k},\mathfrak{l})C(\mathfrak{g},\mathfrak{k})$
, and so
$$[{C}^{},({C}^{}{)}^{adj}]=[{C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})+{C}^{}(\mathfrak{g},\mathfrak{k}),{C}_{\Delta}(\mathfrak{k},\mathfrak{l})C(\mathfrak{g},\mathfrak{k})]=[{C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l}),{C}_{\Delta}(\mathfrak{k},\mathfrak{l})][{C}^{}(\mathfrak{g},\mathfrak{k}),C(\mathfrak{g},\mathfrak{k})]=\Delta .$$
So the situation for
${C}^{}$
is exactly the same as for
$C$
and we conclude
Theorem 7.3.
(a)
$Ker\Delta =Ker{C}^{}\cap Ker({C}^{}{)}^{adj}$
; (b)
$V\otimes S=Ker\Delta \oplus Im{C}^{}\oplus Im({C}^{}{)}^{adj}$
; (c)
$Ker{C}^{}=Ker\Delta \oplus Im{C}^{}$
; (d)
$Ker({C}^{}{)}^{adj}=Ker\Delta \oplus Im({C}^{}{)}^{adj}$
.
In other words, Hodge decomposition also holds for
$\mathfrak{u}$
homology. Moreover, we see that
$\overline{\mathfrak{u}}$
cohomology and
$\mathfrak{u}$
homology have the same set of harmonic representatives,
$Ker\Delta $
. In particular they are isomorphic.
We now want to relate
$\overline{\mathfrak{u}}$
cohomology and
$\mathfrak{u}$
homology to Dirac cohomology with respect to
$D(\mathfrak{g},\mathfrak{l})$
. The main observation here is
Lemma 7.4.
$Ker\Delta =Ker{D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}\cap KerD(\mathfrak{g},\mathfrak{k}{)}^{2}=Ker{D}_{\Delta}(\mathfrak{k},\mathfrak{l})\cap KerD(\mathfrak{g},\mathfrak{k})$
.

Proof.
The operators
${D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}$
and
$D(\mathfrak{g},\mathfrak{k}{)}^{2}$
are both positive semidefinite and their sum is
$\Delta $
. This immediately implies the first equality. The second follows from
$Ker{D}_{\Delta}(\mathfrak{k},\mathfrak{l}{)}^{2}=Ker{D}_{\Delta}(\mathfrak{k},\mathfrak{l})$
(since
${D}_{\Delta}(\mathfrak{k},\mathfrak{l})$
is selfadjoint) and
$KerD(\mathfrak{g},\mathfrak{k}{)}^{2}=KerD(\mathfrak{g},\mathfrak{k})$
(since
$D(\mathfrak{g},\mathfrak{k})$
is antiselfadjoint). □
We can now combine Theorems 7.2 and 7.3 with Lemmas 7.4 and 4.6 to conclude
Corollary 7.5.
$Ker\Delta =Ker{C}_{\Delta}(\mathfrak{k},\mathfrak{l})\cap Ker{C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})\cap KerC(\mathfrak{g},\mathfrak{k})\cap Ker{C}^{}(\mathfrak{g},\mathfrak{k}).$

Proof.
It is obvious that the left hand side contains the right hand side. Conversely, if
$x\in Ker\Delta $
, then
$Cx=0$
by Theorem 7.2,
$D(\mathfrak{g},\mathfrak{k})x=0$
by Lemma 7.4, so
$C(\mathfrak{g},\mathfrak{k})x=0$
by Lemma 4.6 and so also
$CxC(\mathfrak{g},\mathfrak{k})x={C}_{\Delta}(\mathfrak{k},\mathfrak{l})x=0$
. Analogously,
${C}^{}(\mathfrak{g},\mathfrak{k})x=0$
and
${C}_{\Delta}^{}(\mathfrak{k},\mathfrak{l})x=0$
. □
Since
$Ker{C}_{\Delta}(\mathfrak{k},\mathfrak{l})\cap KerC(\mathfrak{g},\mathfrak{k})$
can be thought of as the kernel of
${C}_{\Delta}(\mathfrak{k},\mathfrak{l})$
acting on the kernel of
$C(\mathfrak{g},\mathfrak{k})$
, and similarly for the
${C}^{}$
operators, in view of Theorem 4.7 and Remark 4.8 we can reinterprete Corollary 7.5 as follows:
Corollary 7.6.
To calculate the
$\overline{\mathfrak{u}}$
cohomology of
$V$
, one can first calculate the
${\mathfrak{p}}^{}$
cohomology of
$V$
to obtain a
$\stackrel{~}{K}$
module, and then calculate the
$\overline{\mathfrak{u}}\cap \mathfrak{k}$
cohomology of this module. Analogously, to calculate the
$\mathfrak{u}$
homology of
$V$
, one can first calculate the
${\mathfrak{p}}^{+}$
homology of
$V$
, and then the
$\mathfrak{u}\cap \mathfrak{k}$
homology of the resulting
$\stackrel{~}{K}$
module. □
Remark 7.7.
Note that this is in fact the HochschildSerre spectral sequence for the ideal
${\mathfrak{p}}^{}$
of
$\overline{\mathfrak{u}}$
respectively the ideal
${\mathfrak{p}}^{+}$
of
$\mathfrak{u}$
. What we have obtained is that these HochschildSerre spectral sequences are always degenerate for a unitary
$(\mathfrak{g},K)$
module
$V$
. □
We now turn our attention to the Dirac cohomology of
$D=D(\mathfrak{g},\mathfrak{l})$
. In addition to the above considerations, we bring in Corollary 5.6, and note that for both
${D}_{\Delta}(\mathfrak{k},\mathfrak{l})$
and
$D(\mathfrak{g},\mathfrak{k})$
the cohomology is the same as the kernel or the kernel of the square. Thus we obtain:
Theorem 7.8.
The Dirac cohomology
${H}_{D}(\mathfrak{g},\mathfrak{l};V)$
of a unitary
$(\mathfrak{g},K)$
module
$V$
is isomorphic to the
$\overline{\mathfrak{u}}$
cohomology of
$V$
and the
$\mathfrak{u}$
homology of
$V$
up to appropriate modular twists. Moreover, all three cohomologies have the same set of harmonic representatives,
$Ker\Delta $
. □
8 Homological properties of Dirac cohomology
Let us start by showing that although we proved that in some cases Dirac cohomology of a unitary
$(\mathfrak{g},K)$
module with respect to
$D(\mathfrak{g},\mathfrak{l})$
can be identified with
$\overline{\mathfrak{u}}$
cohomology or
$\mathfrak{u}$
homology, one should by no means expect that these notions agree for general
$(\mathfrak{g},K)$
modules. Let us see that this is not the case even for
$(\mathfrak{s}\mathfrak{l}(2,\mathbb{C}),SO(2\left)\right)$
modules.
Consider the module
$V$
which is a nontrivial extension of the discrete series representation
$W$
of highest weight
$2$
by the trivial module
$\mathbb{C}$
:
$$0\to \mathbb{C}\to V\to W\to 0.$$
(In other words,
$V$
is a dual Verma module.) The
$\mathfrak{k}$
weights of
$V$
(for the basis element
$\left[\begin{array}{cc}0& i\\ i& 0\end{array}\right]$
of
$\mathfrak{k}$
) are
$0,2,4,...$
. We are considering the case
$\mathfrak{l}=\mathfrak{k}$
,
$\mathfrak{u}$
is spanned by
$u=\frac{1}{2}\left[\begin{array}{cccc}1& i& i& 1\end{array}\right]$
and
$\overline{\mathfrak{u}}$
is spanned by
${u}^{*}=\frac{1}{2}\left[\begin{array}{cccc}1& i& i& 1\end{array}\right]$
.
For any
$(\mathfrak{s}\mathfrak{l}(2,\mathbb{C}),SO(2\left)\right)$
module
$X$
we have
$$X\otimes S=X\otimes 1\oplus X\otimes u,$$
with
$d:X\otimes 1\to X\otimes u$
given by
$d(v\otimes 1)={u}^{*}\cdot v\otimes u$
,
$\partial :X\otimes u\to X\otimes 1$
given by
$\partial (v\otimes u)=u\cdot v\otimes 1$
, and
$D=d2\partial $
. By an easy direct calculation, we see that
$${H}_{0}(\mathfrak{u};V)=0;{H}_{1}(\mathfrak{u};V)=\mathbb{C}{v}_{0}\otimes u;{H}^{0}(\overline{\mathfrak{u}};V)=\mathbb{C}{v}_{0}\otimes 1;{H}^{1}(\overline{\mathfrak{u}};V)=\mathbb{C}{v}_{0}\otimes u\oplus \mathbb{C}{v}_{2}\otimes u;{H}_{D}\left(V\right)=KerD=\mathbb{C}{v}_{0}\otimes u.$$
as vector spaces. Here
${v}_{i}$
denotes a vector in
$V$
of
$\mathfrak{k}$
weight
$i$
. So we see
$${H}_{D}\left(V\right)={H}_{\cdot}(\mathfrak{u};V)\ne {H}^{\cdot}(\overline{\mathfrak{u}};V)$$
On the other hand, for
$\mathbb{C}$
and
$W$
another easy calculation (which can be shortened by using Theorem 4.7, as both
$\mathbb{C}$
and
$W$
are unitary) implies
$${H}^{0}(\overline{\mathfrak{u}};\mathbb{C})={H}_{0}(\mathfrak{u};\mathbb{C})={H}_{D}^{0}(\mathbb{C})=\mathbb{C}1\otimes 1;{H}^{1}(\overline{\mathfrak{u}};\mathbb{C})={H}_{1}(\mathfrak{u};\mathbb{C})={H}_{D}^{1}(\mathbb{C})=\mathbb{C}1\otimes u;{H}^{0}(\overline{\mathfrak{u}};W)={H}_{0}(\mathfrak{u};W)={H}_{D}^{0}\left(W\right)=0;{H}^{1}(\overline{\mathfrak{u}};W)={H}_{1}(\mathfrak{u};W)={H}_{D}^{1}\left(W\right)=\mathbb{C}{v}_{2}\otimes u.$$
To explain why Dirac cohomology of
$V$
differs from the
$\overline{\mathfrak{u}}$
cohomology of
$V$
, we will examine their behavior with respect to extensions. Recall the well known long exact sequences for Lie algebra homology and cohomology corresponding to our short exact sequence
$$0\to \mathbb{C}\to V\to W\to 0.$$
They are
$$0\to \mathbb{C}1\otimes u\to \mathbb{C}{v}_{0}\otimes u\to 0\mathbb{C}{v}_{2}\otimes u\to \mathbb{C}1\otimes 1\to 0\to 0\to 0$$
for
$\mathfrak{u}$
homology and
$$0\to \mathbb{C}1\otimes 1\to \mathbb{C}{v}_{0}\otimes 1\to 0\to \mathbb{C}1\otimes u\to \mathbb{C}{v}_{2}\otimes u\oplus \mathbb{C}{v}_{0}\otimes u\to \mathbb{C}{v}_{2}\otimes u\to 0$$
for
$\overline{\mathfrak{u}}$
cohomology. Here all arrows are the obvious ones except for the one labelled by
$0$
.
For Dirac cohomology, instead of a long exact sequence (which clearly does not makes sense in general, as Dirac cohomology is not
$\mathbb{Z}$
graded), there is a sixterm exact sequence.
In the above example, this sequence is
$$\mathbb{C}1\otimes 1\to 0\to 0\uparrow \downarrow \mathbb{C}{v}_{2}\otimes u\leftarrow 0\mathbb{C}{v}_{0}\otimes u\leftarrow \mathbb{C}1\otimes u$$
We see that in this example the sixterm sequence agrees with the
$\mathfrak{u}$
homology long exact sequence only because of the presence of zeros; in general, all three sequences are different.
To define the sixterm sequence in a more general situation, let us assume that the Dirac cohomology is
${\mathbb{Z}}_{2}$
graded. This happens whenever we are starting from
$\mathfrak{g}=\mathfrak{r}\oplus \mathfrak{s}$
with
$\mathfrak{s}$
evendimensional; this is automatic when
$\mathfrak{r}=\mathfrak{l}$
is a Levi subalgebra. Let
$$0\to X\to iY\to pZ\to 0$$
be a short exact sequence of
$(\mathfrak{g},K)$
modules. Tensor this sequence by
$S$
, and denote the arrows still by
$i$
and
$p$
(they get tensored by the identity on
$S$
). Assuming that
${D}^{2}$
is a semisimple operator for each of the three modules, we can construct a sixterm exact sequence
$${H}_{D}^{0}\left(X\right)\to {H}_{D}^{0}\left(Y\right)\to {H}_{D}^{0}\left(Z\right)\uparrow \downarrow {H}_{D}^{1}\left(Z\right)\leftarrow {H}_{D}^{1}\left(Y\right)\leftarrow {H}_{D}^{1}\left(X\right)$$
The horizontal arrows are induced by
$i$
and
$p$
. The vertical arrows are the connecting homomorphisms, defined as follows. Let
$z\in Z\otimes S$
represent a Dirac cohomology class, so
$Dz=0$
. Choose
$y\in Y\otimes S$
such that
$py=z$
. Since
${D}^{2}$
is semisimple, we can assume
${D}^{2}y=0$
. Since
$pDy=Dpy=Dz=0$
, we see that
$Dy=ix$
for some
$x\in X$
. Since
${D}^{2}y=0$
, we see that
$Dx=0$
, so
$x$
defines a cohomology class. This class is by definition the image of the class of
$z$
under the connecting homomorphism. Clearly, we changed parity when we applied
$D$
, and this defines both vertical arrows at once. It is easy to see that this map is well defined, and that the resulting sixterm sequence is exact. To conclude:
Theorem 8.1.
Let
$\mathfrak{g}=\mathfrak{r}\oplus \mathfrak{s}$
be an orthogonal decomposition, with
$\mathfrak{r}$
a reductive subalgebra and
$\mathfrak{s}$
evendimensional. Let
$0\to X\to Y\to Z\to 0$
be a short exact sequence of
$(\mathfrak{g},K)$
modules and assume that the square of the Dirac operator
$D(\mathfrak{g},\mathfrak{r})$
is a semisimple operator for
$X$
,
$Y$
and
$Z$
. Then there is a sixterm exact sequence corresponding to this short exact sequence, as described above. □
Finally, let us comment on what can be done when
$\mathfrak{s}$
is odd dimensional, say
$dim\mathfrak{s}=2n+1$
. The usual spin modules
${S}_{1}$
and
${S}_{2}$
are not
${\mathbb{Z}}_{2}$
graded and thus it seems the above construction does not make sense. Recall that
${S}_{1}$
and
${S}_{2}$
are defined by writing
$\mathfrak{s}={\mathbb{C}}^{2n}\oplus \mathbb{C}$
, considering the spin module for
$C\left({\mathbb{C}}^{2n}\right)$
, and letting the last basis element of
$\mathfrak{s}$
act in two different ways (preserving the even and odd subspace instead of exchanging them).
We can instead consider the unique irreducible graded module
$\stackrel{~}{S}$
of
$C(\mathfrak{s})$
. It can be constructed as the restriction of the (unique) spin module for
$C\left({\mathbb{C}}^{2n+2}\right)$
to
$C(\mathfrak{s})\subset C\left({\mathbb{C}}^{2n+2}\right)$
. As a nongraded module,
$\stackrel{~}{S}$
decomposes as
${S}_{1}\oplus {S}_{2}$
. If we define Dirac cohomology using
$\stackrel{~}{S}$
in place of
${S}_{1}$
or
${S}_{2}$
, we double it, but we do get a
${\mathbb{Z}}_{2}$
grading.
Then the above construction works also in the odd case. Thus, this is probably a more natural definition of Dirac cohomology in the odd case.
It remains to see what can be done if
${D}^{2}$
is not a semisimple operator. One possibility might be to consider a more general definition of Dirac cohomology in that setting.
References

J. Adams, Nilpotent cohomology of the oscillator representation, J. Reine Angew. Math. 449 (1994), 1–7.

A. Alekseev, E. Meinrenken, Lie theory and the ChernWeil homomorphism, math.RT/0308135.

M. Atiyah, W. Schmid, A geometric construction of the discrete series for semisimple Lie groups, Invent. Math. 42 (1977), 1–62.

A. Borel, N. Wallach, Continuous cohomology, discrete subgroups, and representations of reductive groups, Second edition. Mathematical Surveys and Monographs, 67. American Mathematical Society, Providence, RI, 2000.

W. Casselman, M. S. Osborne, The
$\mathfrak{n}$
cohomology of representations with an infinitesimal character, Comp. Math. 31 (1975), 219–227.

C. Chevalley, The algebraic theory of spinors, Columbia University Press, 1954.

D. H. Collingwood, The
$\mathfrak{n}$
homology of HarishChandra modules: generalizing a theorem of Kostant, Math. Ann. 272 (1985), no. 2, 161–187.

A. Connes, H. Moscovici, The
${L}^{2}$
index theorem for homogeneous spaces of Lie groups, Ann. of Math. 115 (1982), no. 2, 291–330.

T. J. Enright, Analogues of Kostant's
$\mathfrak{u}$
cohomology formulas for unitary highest weight modules, J. Reine Angew. Math. 392 (1988), 27–36.

R. Hotta, R. Parthasarathy, A geometric meaning of the multiplicities of integrable discrete classes in
${L}^{2}(\Gamma \backslash G)$
, Osaka J. Math. 10 (1973), 211–234.

J.S. Huang, P. Pandžić, Dirac cohomology, unitary representations and a proof of a conjecture of Vogan, J. Amer. Math. Soc. 15 (2002), 185–202.

J.S. Huang, P. Pandžić, Dirac operators in representation theory, Representations of Real and Padic Groups, Lecture Notes Series, Vol. 2, Institute for Mathematical Sciences, National University of Singapore, Singapore University Press and World Scientific, 2004, pp. 163219.

V. Kac, Lie superalgebras, Adv. in Math. 26 (1977), 8–96.

B. Kostant, Lie algebra cohomology and the generalized BorelWeil theorem, Ann. of Math. 74 (1961), 329–387.

B. Kostant, A cubic Dirac operator and the emergence of Euler number multiplets of representations for equal rank subgroups, Duke Math. Jour. 100 (1999), 447–501.

B. Kostant, A generalization of the BottBorelWeil theorem and Euler number multiplets of representations, Lett. Math. Phys. 52 (2000), 61–78.

B. Kostant, Dirac cohomology for the cubic Dirac operator, Studies in memory of I. Schur, in Progress of Math. vol. 210 (2003), 69–93.

S. Kumar, Induction functor in noncommutative equivariant cohomology and Dirac cohomology, preprint, University of North Carolina, 2003.

R. Langlands, The dimension of spaces of automorphic forms, Amer. J. of Math. 85 (1963), 99–125.

R. Parthasarathy Dirac operator and the discrete series, Ann. of Math. 96 (1972), 1–30.

W. Schmid,
${L}^{2}$
cohomology and the discrete series, Ann. of Math. 103 (1976), 375–394.

D.A. Vogan, Representations of real reductive Lie groups, Birkhäuser, Boston, 1981.

D.A. Vogan, Dirac operators and unitary representations, 3 talks at MIT Lie groups seminar, Fall 1997.

D.A. Vogan,
$\mathfrak{n}$
cohomology in representation theory, a talk at “Functional Analysis VII”, Dubrovnik, Croatia, September 2001.

N. R. Wallach, Real Reductive Groups, Volume I, Academic Press, 1988.
Department of Mathematics, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong SAR, China Email address : mahuang@ust.hk Department of Mathematics, University of Zagreb, Bijenicka 30, 10000 Zagreb, Croatia Email address : pandzic@math.hr Centre de mathematiques Laurent Schwartz, Ecole Polytechnique, 91128 Palaiseau Cedex, France Email address : renard@math.polytechnique.fr