# Lowerbound to sample complexity or run-time of doing sparse-coding

I noticed this interesting analogous discussion that happened here,

Lower bound proof for compressive sensing (Gel'fand widths)?

• What is the closest analogue know for such a result for sparse-coding?