Pretty much anything in classical linear algebra can be abstracted to something more powerful by taking a swig of the Axiom of Choice.
-
-
But we can restrict the KLT in other useful ways. Polynomials form a linear vector space, so choosing orthogonal polynomials makes sense.
-
As it turns out, the weighting functions in the inner product space for orthogonal polynomials are often closely related to probability fxns
-
Or if we look at a measure theoretic view, certain polynomial families are orthogonal with respect to a probability measure.
-
Norbert Wiener studied this in the 30s and 40s and called it "Polynomial Chaos." It's one of the most powerful unheard of techniques around.
-
It's basically a tool that allows us to write arbitrary (finite variance) distributions with respect to chosen, happy distributions.
-
The math looks a lot different, but it's basically PCA with an (infinite) orthogonal polynomial basis.
-
It's used in CFD somewhat, that's about it. I've never met a data scientist or even a math professor who's heard of it.
-
But it is incredibly powerful and parallelizable, particularly for simulation. Maybe I should write an open source python library someday..
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.