Niels G. W. Serup
I'm 26 years old.
I have a master's degree in computer science from the University of Copenhagen (DIKU). During the day I work as a software developer for Jobindex.dk, and by night I play with fun computer science problems.
I like to work in the cross section between interesting theory and practical implementations.
URLs for Niels G. W. Serup
Banko is a game of chance, often played in a social setting. It is somewhat similar to the U.S. variant of Bingo, but has its own rules. The basics is that each player has one or more banko cards -- each card having 15 numbers -- and must cover a set amount of randomly picked numbers to win, at which point they yell "BANKO!".
For this talk I want to first explore existing problems related to scaling banko games, quickly discover that none exist, and then spend the remaining time describing made-up problems and solutions to them.
In particular, the umbrella term 'Big Banko' covers such diverse areas as:
Standardisation: How do you transfer banko cards between different entities and not get confused? I present a well-defined and open banko card standard.
Visualisation: What is the most human-friendly non-print visualisation of one or more banko cards? I showcase several approaches and pick the best one.
Validation: When you're in a hurry, how do you quickly check the validity of millions of banko cards? I present a GPU-accelerated solution that outperforms a human by several orders of magnitude.
Compression: Just how much can you compress (the digital representation of) a banko card? Whenever someone needs to transfer billions of banko cards over a slow internet connection, they need a fast and reliable compression/decompression scheme. There are several low-hanging fruits in the quest for a good lossless compressor, but it is a surprisingly complex and math-dense endavour to exploit all the banko card invariants and reach optimality. I present a variety of algorithms and discuss their bit usage.
Simulation: Is it possible to remove the human element from banko altogether, and just let a computer play it? Yes, of course it is.
This talk is introductory in nature and is suited for all audiences, including:
- intermediate banko enthusiasts wanting to gain an edge;
- banko business-doers wishing to modernise their enterprises; and
- banko newcomers working on becoming cool.
The contents of the talk will occasionally require a good understanding of computer science, but I will make sure to always give the big picture as well.
This is joint work with Troels Henriksen, postdoc at DIKU, and David Udsen, software developer. Our Big Banko tools can be found at https://github.com/diku-dk/openbanko and are licensed under BSD 2. Part of our work is based on the 2003 scientific paper "Hvor mange bankoplader er der?" (eng: "How many banko cards exist?") by Nils Andersen, professor emeritus at DIKU.
Scheduled Instances of "Big Banko: Scaling for the Elder Generation"
- Saturday Aug. 18 13:00 - 14:00