A proof of The Edwards-Walsh Resolution Theorem without Edwards-Walsh CW-complexes

In the paper titled “Bockstein basis and resolution theorems in extension theory” (arXiv:0907.0491v2), we stated a theorem that we claimed to be a generalization of the Edwards-Walsh resolution theorem. The goal of this note is to show that the main theorem from (arXiv:0907.0491v2) is in fact equivalent to the Edwards-Walsh resolution theorem, and also that it can be proven without using Edwards-Walsh complexes. We conclude that the Edwards-Walsh resolution theorem can be proven without using Edwards-Walsh complexes.

💡 Research Summary

The paper revisits a result originally proved by Edwards and Walsh (the Edwards‑Walsh resolution theorem) and shows that the theorem can be established without invoking the sophisticated Edwards‑Walsh CW‑complexes that have traditionally been an essential part of the proof. The author begins by recalling a theorem from his earlier work (arXiv:0907.0491v2), which states that for an abelian group G and a connected CW‑complex K with πₙ(K)≅G and trivial lower homotopy groups, any compact metrizable space X that admits K as an absolute extensor (X τ K) admits a compact metrizable space Z of dimension ≤ n together with a cell‑like surjection π: Z→X. This theorem is claimed to be equivalent to the classical Edwards‑Walsh resolution theorem, which asserts that for any compact metrizable space X with cohomological dimension ≤ n there exists a compact metrizable space Z of dimension ≤ n and a cell‑like surjection onto X.

To establish the equivalence, the author invokes Dranishnikov’s extension theory (Theorem 2.1 and its corollary Theorem 2.2). These results say that for a simple CW‑complex M, the condition X τ M is equivalent to X τ SP^∞ M and to a family of cohomological dimension inequalities dim H_i(M) X ≤ i. Applying this to the complex K yields dim G X ≤ n, and using properties of the Bockstein basis (Lemma 2.4 in the earlier paper) one deduces that dim G X coincides with the cohomological dimension dim_Z X. Consequently, the hypothesis “dim_Z X ≤ n” in the Edwards‑Walsh theorem can be satisfied, and the existence of Z and a cell‑like map follows from the original theorem. Thus the two statements are shown to be logically equivalent.

The core contribution of the present note is a streamlined proof of a variant of Edwards’ theorem (Theorem 3.1 in the paper). The original proof relied on constructing Edwards‑Walsh complexes of dimension n+1 and then modifying maps to obtain a cell‑like resolution. The author replaces this machinery with two classical tools: (1) the resolution property (R1) for inverse sequences of compact polyhedra, as established by Mardešić and Segal, which guarantees that any map from the inverse limit can be approximated arbitrarily closely by a map that factors through a finite stage; and (2) the theory of unstable values for maps into the (n+1)-cube, originally due to Hurewicz and Wallman and later refined in shape theory. By iteratively removing the interior of each (n+1)-simplex from the image of the bonding maps, using Lemma 3.2 (which provides an open cover V and a retraction r with controlled behavior on simplices), the author constructs a sequence of approximations g₁,σ, each avoiding a particular simplex interior. After finitely many steps (since the first polyhedron L₁ has only finitely many (n+1)-simplices) one obtains a map g₁ that coincides with the original bonding map on the n‑skeleton and sends the pre‑image of each (n+1)-simplex into its boundary.

Next, using the resolution property (R1), the author selects a later stage s of the inverse system and a map b g_{s}^{1}: |L_s|→|L^{(n)}1| that is V‑close to g₁. Composing with the retraction r from Lemma 3.2 yields a map g{s}^{1} that is an L₁‑modification of the bonding map f_{s}^{1}. By checking the behavior on simplices, the author verifies that g_{s}^{1} satisfies the required modification property, thereby producing the desired cell‑like approximation without ever constructing an Edwards‑Walsh complex.

The final section compares this approach with the original proof of the Edwards‑Walsh theorem as presented in Walsh’s monograph. There, Theorem 4.1 (Edwards) and Theorem 4.2 (Walsh) are used, and the construction of Edwards‑Walsh complexes is essential to control dimensions of intermediate polyhedra. The author shows that Theorem 3.1 is essentially a weakened version of Theorem 4.1 with an additional hypothesis dim |L_i| ≤ n+1, which together with the assumption dim_Z Y ≤ n forces dim Y ≤ n. Hence the use of Edwards‑Walsh complexes can be completely avoided.

In summary, the paper demonstrates that the Edwards‑Walsh resolution theorem is equivalent to a more general statement proved in the author’s earlier work, and that both can be proved using only standard tools from shape theory, inverse limits, and unstable value theory, without the need for the intricate Edwards‑Walsh CW‑complexes. This simplification not only clarifies the logical structure of the resolution theorem but also makes the result more accessible to researchers who may be unfamiliar with the technicalities of Edwards‑Walsh complexes.