Fw: Re: Twin primes and Pi
- --- On Wed, 11/28/12, gigi_avatar27 wrote:
> Hi folks,Welcome to the list, Robert, great to have a mind as sharp as yours on board!
> I think it's important to use the best tools for theOne might just describe them as merely expressions with higher
> job. I'm glad to hear that RIES is useful, but I'm
> also glad we have ISC too.
> Part of the difficulty with ISC is that if you need it, then
> you *really* need it, but if you don't need it, it's
> difficult to understand why anyone would. ISC is full of
> cryptic, advanced, specialized mathematics.
kolmogorov complexity than what RIES considers.
> So while I did create RIES partly out of frustration withBecause your alphabet is deliberately small. The same principles
> ISC and Simon Plouffe's Inverter, I was far more motivated
> by having fun with maths and helping show people how easy it
> is to match any real number with equations of high-school
work equally well on a bigger alphabet.
And as you open source RIES, there's nothing stopping people from
adding more constants and functions, for which I'm most grateful,
even if I've not actually got around to doing that yet - I certainly
plan to! (However, I'd probably drop phi as a leaf-node too!)
> In a sense, RIES is too frivolous for the problem at hand,Well, even the ISC can't tell you the answer to the actual problem
> namely the infinite product PI[t/(t-1)] for all twin primes t.
> My "Stupid Math Tricks" examples (http://goo.gl/v541k) are
> more in line with the true spirit of RIES.
at hand. However, once you've got a sufficiently accurate
approximation both ISC and RIES can tell you what the expression
might be equal to, *neither* being able to give any justification
stronger than coincidence. In David's and Werner's proven expressions,
*both* ISC and RIES give the same correct answer.
- Thanks, glad to be here.
--- In email@example.com, Phil Carmody <thefatphil@...> wrote:
> And as you open source RIES, there's nothing stopping people from
> adding more constants and functions, for which I'm most grateful,
> even if I've not actually got around to doing that yet - I certainly
> plan to! (However, I'd probably drop phi as a leaf-node too!)
You can already drop phi, if that's all you want to do, by excluding the phi symbol from the search with the option -Nf.
Adding constants and functions was and is a conscious motivation from the very start, and I consider it in any refactoring and enhancements. In the source code you can see that I've already specified the syntax for custom constants and functions through the existing -include mechanism. For example:
--define : EulerGamma # seft-a (constant)
( -- The Euler-Mascheroni constant, 0.57721... )
# 50 digits for when RIES goes to higher precision
This isn't implemented yet but at least I thought of it.