PDA

View Full Version : Glib Empirical Idea

ASEI
2007-Apr-22, 03:23 PM
With so many new theories like string theory, ect, it seems there is a whole lot of wild visualization being attempted. Rather than fundamental variables, they are now trying to replace these with extra unobserved dimensions (seems like trading an unexplained quantity for an unobservable physical axis to me, but I admit not having seen much of the math behind it).

I was wondering – if you wanted to get a theory that was completely consistent with all of your current observations and contained little in the way of extra theoretical/narrative fluff, rather than embarking on extended voyages of visualization, could you do something like the following?

Just set up the variables, rates, dependencies, ect out to whatever highest order you are expecting out into a gigantic state variable vector.

{dxi/dxj} = [K]{xi},

Gather all known experimental correlations, solve for the K matrix, and translate the K matrix back out into nice diffeqs? It should give you a set of physical laws that reproduce everything experimentally observed up until this point.

(Not too useful for projecting new expected phenomena, but it should resolve any lingering problems with our models of current phenomena).

Nereid
2007-Apr-26, 03:46 PM
I think you'll find that this is a variation on, or a subset of, Ken G's "Google science". There is at least one thread, in another section of BAUT, that discusses this idea ... from the POV of the nature of science, Occam's razor, etc.

papageno
2007-Apr-26, 07:34 PM
Just set up the variables, rates, dependencies, ect out to whatever highest order you are expecting out into a gigantic state variable vector.

{dxi/dxj} = [K]{xi},

Gather all known experimental correlations, solve for the K matrix, and translate the K matrix back out into nice diffeqs? It should give you a set of physical laws that reproduce everything experimentally observed up until this point.

This is not a theory-theory, but phenomenology (which, by the way, is quite common).

Instead of starting from a set of postulates and definitions (like Newton's mechanics), you are simply describing observed behaviour (like Kepler's laws).

I don't see how it would help solving the "lingering problems" of our current models, because a phenomenological theory usually does not provide a new insight into the phenomena.

Again the example of Kepler's laws and Newton's mechanics and law of gravitation.
Kepler's laws describe nicely the orbits of the planets. But it was Newton who showed us that those laws were the direct consequence of his dynamics and law of gravitation.

Dr Nigel
2007-May-04, 04:40 PM
IIUC, the models of string theory predict the properties of the subatomic particles. The problem is, there are too many versions of string theory that do this, so it is not possible to choose between them.