I was reading a definition of pro-Lie group and it spoke of a "Cauchy filter" on an arbitrary topological group even though there was no mention of a metric. Is there some kind of standard meaning for "Cauchy filter" in the context of an arbitrary topological group, or perhaps there was some kind of assumption being made that there is a metric which is compatible with the topology?
The source I was reading is here:
http://www.matha.rwth-aachen.de/publikationen/pumpluen-festschrift/Hofmann_Morris.pdf