Martin Josef Geiger

 orcid.org/0000-0003-1797-957X

 https://www.kaggle.com/mjgeiger

 https://github.com/MartinJGeiger

I am a scientist with an academic background in Economics, Business Administration, and Computer Science.
My current position is here: HSU Hamburg.

Research interests

1. Combinatorial optimization, with applications in vehicle routing, scheduling, timetabling, etc.

   Here, the objective is to find (optimal) values for a vector of decision variables $X\text{, }X=\left(x_1\text{, ..., }{x}_n\right)$ such that some function $f\left(X\right)$ assumes a minimal value. In most cases, $f\left(X\right)$ is a cost-function, e.g. measuring the costs of trucks making deliveries, of a production schedule, or some other practice-related management objective. As we typically have a good understanding of the optimality criterion, we can tell whether a solution is optimal (or at least of a certain quality – by comparison with lower bounds or best-known upper bounds).

   Employed methods stem from mathematical optimization, graph theory, operations research, and related disciplines. Often, (meta-)heuristics play an important role, as optimal solutions cannot be obtained within the available computing times.

2. Multi-criteria decision making under uncertainty, with applications in real-world dynamic capital allocation.

   There, the objective is to find values for a vector of decision variables $X$, such that some future outcome $F\left(X\right)=\left(f_1\right(X\left)\text{, ..., }{f}_k\right(X\left)\right)$ is optimized. Contrary to case (i.) above, the optimality criteria are less obvious and open for discussion/ interpretation, hence the multi-objective formulation. In all cases, they will express some sort of payoff-/ cashflow-/ profit-function, complemented with some appropriate risk measure(s). Moreover, the actual outcomes are typically revealed over time, as certain scenarios play out (while others do not). It follows that often expected values $E\left(f_i\right(X\left)\right)$ are considered.

   Relevant methods require a thorough understanding of risk management and the underlying economics of the problem at hand. Studying probability theory makes sense, also, as it is important to know how to use it and when to better avoid it.

Programming competitions

My most recent contribution was in June 2024 to the Heuristic Track of PACE 2024.

More on programming competitions can be found here.