Essentially, first you need to work out a direction.
If you do have a game in mind then find something to start on, given your maths experience you might want to tackle the game physics/movement/whatever.
If you don't have an idea, this again is where your maths might help. Back in the day folks would mess around with the computer, trying to do different animated effects, etc. Those with math would usually go toward animated graphing functions and whatnot, once they get something that's interesting, they mess around on gamifying the hack, making it a playfield, character or goal... and build out a game from that.
The third method is well known to artists, start by emulating something you already like and put your spin into it. Got a game you like but have an idea that you could improve upon the concept? then maybe that's the route to start.
Once you have the spark of what to do, now you work on making it work (what platform do you want to see it on, go to that platform forum; need 3d support? go to 3d forums; etc). Most of all while doing this is actually writing code and testing out your ideas.
Without an idea of what you want, the most folks could help you with is directing you to some type-in game tutorial...