I am a little bit confused about the emptiness problem of the linear bounded automaton (LBA). I know that this problem is undecidable. However if we assume that is decidable, what could be wrong if we reduce it to the acceptance problem of LBA?
Since the acceptance problem for LBAs is decidable, if you could reduce th emptiness problem to the acceptance problem, then the emptiness problem for LBAs would be decidable. However, the emptiness problem for LBAs is not decidable. This shows that we cannot reduce the emptiness problem to the acceptance problem.
It seems that you are confused about the direction of the reduction. When reducing problem A to B, we show how to solve problem A using problem B, not the other way around.