In today’s lecture we looked at the notion of the null/dual space of a linear subspace and then saw some consequences. In particular, we saw that the existence of a parity check matrix implies polynomial time error detection for linear codes. I typed up some notes for the stuff on null spaces. The notes have a slightly better organization wrt the definition of a dual space and its consequences. The implications for linear codes are from Lecture 5 from Fall 2007.

Also I should correct a brain-freeze I had in the lecture today. While talking about linear subspaces, Jeff noted that a linear subspace and its dual can have the same vector in its basis. Then Dan observed that this is **not** true for linear subspace over the reals. I kind of hedged it in class but Dan was correct. Even though I don’t quite like the reals, I should not have missed the very simple argument behind Dan’s claim– sorry about that! Here is a quick proof: let . Then note that , where the equality only holds when . In other words, (for a linear subspace ) if and only if , which of course means that cannot be an element of a basis.

## Leave a Reply