Why Floating-Point Still Surprises People
A practical walk through binary fractions, rounding modes, and why 0.1 + 0.2 is not broken.
Exploring the intersection of Math, Art, and Code.
I am a software engineer who spends too much time thinking about P vs NP problems. Originally from Tokyo, now writing code in the Bay Area.
My professional work involves backend optimization, but my passion lies in Computational Geometry and Cryptographic Primitives. I believe that code is just math that actually does something.
Currently, I am researching efficient ways to generate Voronoi diagrams in non-Euclidean spaces. If you want to talk about category theory or why 0.1 + 0.2 != 0.3 in floating-point arithmetic, buy me a coffee.
Efficient Voronoi construction in non-Euclidean spaces, numerically stable geometric predicates, and the quiet pleasure of well-behaved backend systems.
Also: category theory, exact arithmetic, generative art, and why tiny implementation details always become philosophical eventually.
A practical walk through binary fractions, rounding modes, and why 0.1 + 0.2 is not broken.
What changes when the notion of distance itself changes?
Immutability is not only about purity. It can also be a performance tool.
You do not need to worship abstractions to find them useful.
Durable tools tend to be boring in the best possible way.
Less glamour, more measurement.
Randomness is rarely the point. Constraint is.
The more elegant a primitive looks, the less casually you should touch it.
Some ideas need friction before they deserve a compiler.
Why geometric algorithms feel different from other kinds of code.