Given a paper that introduces a new algorithm and upon publication copyright is transferred to the publisher.
Is an implementation of the presented algorithm considered a derivative of the published article, hence falls under its copyright? Or can someone implement the algorithm at any time making it available under GNU independently of the published paper? What happens if the author published the code under GNU license before submitting the paper that derives and explains the algorithm in detail?
I guess, on a more abstract level, my question is, how paper and code relate to each other?