We suggest the possibility that the two-dimensional
SU(2)k Wess-Zumino-Witten (WZW) theory, which has global
SO(4) symmetry, can be continued to $2+\ensuremath{\epsilon}$ dimensions by enlarging the symmetry to $\mathrm{SO}(4+\ensuremath{\epsilon})$. This is motivated by the three-dimensional sigma model with
SO(5) symmetry and a WZW term, which is relevant to deconfined criticality. If such a continuation exists, the structure of the renormalization group flows at small $\ensuremath{\epsilon}$ may be fixed by assuming analyticity in $\ensuremath{\epsilon}$. This leads to the conjecture that the WZW fixed point annihilates with a new, unstable fixed point at a critical dimensionality
dc>2. We suggest that
dc<3 for all
k, and we compute
dc in the limit of large
k. The flows support the conjecture that the deconfined phase transition in
SU(2) magnets is a ``pseudocritical'' point with approximate
SO(5), controlled by a fixed point slightly outside the physical parameter space.
I study the renormalization of D-dimensional level-k Wess-Zumino-Witten theory with Stiefel-manifold target space
StN,N−D−1≅SO(N)/SO(D+1), with a particular focus on
D=3. I investigate in particular whether such a theory admits IR-stable fixed points of the renormalization group flow. Such fixed points have been suggested to describe conformal phases of matter that do not have a known dual (super-)renormalizable Lagrangian for
N≥7 in
D=3. They are hence of interest both from the point of view of quantum phases of matter as well as pure field theory. The
D-dimensional expressions enable the computation, by analytic computation, of beta functions in
D=2+ε, at least to first non-trivial order. In
D=2, a stable fixed point is found, serving a generalization of the famed
SU(2)k Wess-Zumino-Witten conformal field theory; it annihilates in
D=2+ε with an unstable fixed point which splits off from the Gaussian one for
ε>0. Although the story is thus qualitatively similar to that of SO(5) deconfined (pseudo-)criticality, for
N⩾6, the annihilation appears to occur only for
ε>1, suggesting the existence of a stable phase in
D=3. Comparisons of the scaling dimension of the lowest singlet operator are made with known results for
N=6, which is dual to QED
3 with
Nf=4 fermion flavors. The predictions for the
N=7 Stiefel liquid represent to my knowledge the first computation of this kind for a Wess-Zumino-Witten theory without a known gauge theory dual.