The homotopy theory of operad subcategories (Q1630573)

Let \(Op\) be the category of all topological operads, and let \(Op_*\) be the full subcategory of topological operads \(\mathsf{P}\) satisfying \(\mathsf{P}(0) = *\), also known as unitary operads. (Note that this latter category is isomorphic to the category of \(\Lambda\)-operads.) Both of these categories admit a model structure, with weak equivalences and fibration defined arity-wise [\textit{C. Berger} and \textit{I. Moerdijk}, Comment. Math. Helv. 78, No. 4, 805--831 (2003; Zbl 1041.18011)]. The authors prove that if \(\mathsf{P}\) and \(\mathsf{Q}\) are reduced operads, then there is a weak-equivalence of mapping spaces: \[ \text{Map}_{Op_*}^h(\mathsf{P}, \mathsf{Q}) \sim \text{Map}_{Op}^h(\mathsf{P}, \mathsf{Q}). \] They also extend this result to truncated operads, i.e. operads defined up to a certain arity \(k \geq 0\). Both kinds of mapping spaces in practice: the authors have used the first kind to compute mapping spaces of \(E_n\) operads [\textit{B. Fresse}, \textit{V. Turchin} and \textit{T. Willwacher}, ``The rational homotopy of mapping spaces of \(E_n\) operads'', Preprint, \url{arXiv:1703.06123}], whereas the second kind is usually used in Goodwillie-Weiss manifold calculus, and the truncated kind is used to compute polynomial approximations in this calculus. The proof roughly goes as follows. Let \(\iota : Op_* \to Op\) be the inclusion. The authors prove that this function is a right Quillen adjoint. Its left Quillen adjoint is the ``unitarization'' function \(\tau : Op \to Op_*\). Given an operad \(\mathsf{P}\), the unitary operad \(\tau \mathsf{P}\) is defined by collapsing \(\mathsf{P}(0)\) to a point, and quotienting by an equivalence relation on higher arities to retain the operad structure. The authors then prove that the left derived function \(\mathbb{L}\tau\) is left inverse to \(\iota\) on the level of homotopy categories (from which the main theorem follows). Concretely, this result is obtained by proving that if \(\mathsf{P}\) is a unitary operad and \(\mathsf{E}\) is an \(E_\infty\)-operad, then the unitarization of the Boardman-Vogt resolution of \(\mathsf{E} \times \mathsf{P}\) -- which represents \(\mathbb{L}\tau(\iota\mathsf{P})\) -- is weakly equivalent to \(\mathsf{P}\).