Steepest Descent

Created in 10 days for Scream Jam 2020, hosted by Caveware Digital.

See the author's notes for this game here.

See the source code for this game here.

This is a short horror game based on a well-known analogy from mathematical optimization for describing certain local search algorithms (notably gradient descent and hill climb). If our goal is to minimize the value of a cost function, then we can imagine that the objective is like a foggy mountain range and that our search process is a mountaineer trying to get down. Because of the fog the mountaineer's vision is limited, and so they have to make their decisions about which direction to move in based solely on their immediate surroundings.

This scenario struck me as one rife with horror potential, and I thought it would be kind of funny to try to make a horror game based on mathematical concepts. To help flesh things out there are also a few other... things... to discover on the mountain, based on a couple of local search-based metaheuristics that I've used during my doctoral research.