Hardness of embedding simplicial complexes in Rd

Let EMBEDk -> d be the following algorithmic problem: Given a finite simplicial complex K of dimension at most k, does there exist a (piecewise linear) embedding of K into R-d? Known results easily imply the polynomiality of EMBEDk -> 2 (k = 1, 2; the case k = 1, d = 2 is graph planarity) and of EMBEDk -> 2k for all k >= 3. We show that the celebrated result of Novikov on the algorithmic unsolvability of recognizing the 5-sphere implies that EMBEDd -> d and EMBED(d-1)-> d are undecidable for each d >= 5. Our main result is the NP-hardness of EMBED2 -> 4 and, more generally, of EMBEDk -> d for all k, d with d >= 4 and d >= k >= (2d - 2)/3. These dimensions fall outside the metastable range of a theorem of Haefliger and Weber, which characterizes embeddability using the deleted product obstruction. Our reductions are based on examples, due to Segal, Spiez, Freedman, Krushkal, Teichner, and Skopenkov, showing that outside the metastable range the deleted product obstruction is not sufficient to characterize embeddability.