Travelling Salesman Relaxation



Hello,

Can anybody suggest (and describe) a simple relaxation to the
travelling salesman problem?

I'm only interested in solving small instances and intend to use a
branch-and-bound algorithm.

Regards,
--markus

.