Lost in Math? • Hacker News Board

Re: Lost in Math?

#71

Post by dnautics » Fri, Mar 01, 2019, 1:46 AM UTC

tathougies wrote: post #68

chopin wrote: post #64

Earlier quoted context omitted.

I agree as it is important to know your boundaries. But I also like the GP prefer legible over performant code whenever it is obvious that I am not going to need more performance. In almost all cases I encountered the necessity to optimize code the reason was I/O bound. I don't recall any instance of algorithmic performance being a problem.

Tco has little to do with timing performance and all to do with keeping code legible in functional languages.

In FP that implements actor model, for example, you will be in a world of hurt if you don't TCO your actor - has nothing at all to do with legibility.

Re: Lost in Math?

#72

Post by neokantian » Fri, Mar 01, 2019, 2:51 AM UTC

Nasrudith wrote: post #67

neokantian wrote: post #52

Earlier quoted context omitted.

> Most of the evidence points to the economists abusing assumptions, which is hardly a mathematics problem. Agreed. Economics is about the real world. Therefore, it has to be empirical. That means that axiomatically deriving conclusions from assumptions is not legitimate in economics. Still, in the context of empirical knowledge, we have only two usable methods: the scientific or the historical one. Economics cannot…

There is a difference between non-repeatable in same state and non-scientific. Applying absolute standards of rigor is ironically also unscientific. We know that hyperinflation is a way to screw over an economy utterly. It can and will fail and in the best case be the equivalent of dissolving the currency and going bankrupt. The most benign form of it that may not techically count would involve massive growth as well…

> There is a difference between non-repeatable in same state and non-scientific.

No, there isn't.

> Applying absolute standards of rigor is ironically also unscientific.

The rules governing science are not determined by science itself. Science experimentally tests propositions about facts. Rules about science are propositions about other propositions. Hence, science has absolutely nothing to say about its own rules. Therefore, propositions about the scientific method are necessarily unscientific.

Re: Lost in Math?

#73

Post by dllthomas » Tue, Mar 05, 2019, 4:03 PM UTC

bonoboTP wrote: post #62

tikhonj wrote: post #37

Earlier quoted context omitted.

SAT solvers only scale sometimes . Complexity theory is a starting point for understanding why this "sometimes" is inevitable, but it never even implies that NP-complete problems are never tractable. Complexity theory gives us an understanding of how powerful and expressive SAT is which very much does inform real development. Arguing that "SAT is NP-complete and therefore useless" is not misusing complexity theory, i…

It's common. My lecturer in the first lecture gave a motivation why we learn about this saying:imagine your boss proposes you compute , then you - informed by this lecture - will recognize it belongs to an infeasible complexity class and can tell your boss it won't be possible. We did learn the definitions but "worst case" was never highlighted as such, it was naturally assumed. The closest we got was the discussion…

> tell your boss it won't be possible

I mean, if your boss is demanding a solution that's both accurate and fast on every input, you can tell your boss that showing P=NP is probably out of scope of your project. But that's the start of a discussion about how to step back from that ideal, not the end of a discussion about the potential product.