The h Bar

Slereah

14:00

@qwerty Basically anything involving the big python libraries

most of the learning for python is learning how the libraries you use work

think_meaning_builds

pragmatic learning

Silly Goose

for example, it seems like i can frame a physical theory as an estimator of parameters, say the position of a ball subject to a range of conditions.

ACuriousMind

I'm afraid I have no idea what you're talking about

qwerty

@SillyGoose this sounds like model testing?

ACuriousMind

An estimator - in the sense I know the term - is a function that takes a sample that is supposed to follow a certain kind of probability distribution and then outputs an estimate for some characteristic of that distribution (e.g. the sample mean is a common estimator for the expected value of an underlying distribution)

I don't see how this is supposed to be related to physics outside of the cases where physics makes a probabilistic prediction and you compare the observed sample to the expected distribution by using appropriate estimators

Silly Goose

14:11

@qwerty yesh

@ACuriousMind well in say newtonian mechanics, if I throw a ball with initial $(x_0, v_0)$ subject to constant acceleration $a_0$, then I get a predicted trajectory $x(t)$ for the ball. Is this not a model describing the parameter "position of the ball" as following a time-dependent, distribution that says: certainly the ball will occupy position $x(t)$ at time $t$.

qwerty

@SillyGoose no I don't think so? your parameter here is a_0

ACuriousMind

@SillyGoose if you absolutely want to you can say that, but what does that have to do with estimators?

Silly Goose

well maybe i can frame it more generally. experiments measure values. a physical theory is meant to reproduce these values. measurements and physical theories are assumed to be not perfect, so actually a physical theory should only be expected to estimate the experimentally measured values. In this sense, a physical theory should be able to be framed as a theory of systematically constructing parameter estimators that are related to the class of phenomena the physical theory hopes to study.

whether it makes certain predictions (e.g. newtonian mechanics) or whether it makes statistical predictions (e.g. statistical mechanics) does not matter for the above observation, to my understanding

qwerty

this is similar to the discussion NI Slereah and ACM had with me on how science works and popper vs bayesianism

ACuriousMind

@SillyGoose I think you understand something different under "estimator" than I do

qwerty

14:20

the other week

ACuriousMind

Again, an estimator is something which you apply to a sample in order to get back some property of the underlying probability distribution that produced that sample

it's just not just a name for anything that gives you an approximate value (which is what "estimate" would mean colloquially)

Silly Goose

here i am identifying the experimentally measured values as "the sample".

ACuriousMind

no you're not - you're saying the theory "estimates the experimentally measured values" (you above, verbatim)

but the estimator does not "estimate the sample", it uses the sample as input in order to estimate properties of the underlying distribution

qwerty

I agree with ACM that estimator is the incorrect choice of word

Silly Goose

hm i thought that estimation theory includes constructing a probability distribution which is meant to spit out as similar as possible distributions to what is observed in a sample.

qwerty

14:25

but I think a theory is more general than a model

in your description, I would also argue for replacing "theory" with model...

ACuriousMind

@SillyGoose I mean the use of estimators consists of (at least) two steps: 1. Write down a statistical model with some parameters that you think could describe the process that produces your samples. 2. Find estimators that can estimate the parameters of that model from your sample data.

Silly Goose

oh i think i see what is being said now

ACuriousMind

I'm still puzzled how you think this applies to physics outside of the instances where we do statistics and are interested in reconstructing the underlying distribution from observations

Physics may be heavily involved in that first step, but estimators are just one very specific kind of "fitting models to data" that one may use in the course of doing physics

usually you don't have statistical models and distributions and work with estimators, you just have a theoretical prediction and a bunch of data points with error bars and you do some regression fit mimizing $\chi^2$ or whatever

Silly Goose

hm well then i guess i am curious what maybe a generalization of "parameter estimation" would be in the following sense. say you define a physical theory to be a triplet (1) state space, (2) differential equation(s), (3) observable space. out of all of known math there are infinite ways to choose a triplet $(S, D, O)$. But squinting enough, this seems in the same spirit as parameter estimation where you are fitting a model to some data by tuning parameters appropriately.

here, the parameters would be mathematical structures

ACuriousMind

Now that I think about it, the $\chi^2$ technique and other goodness-of-fit methods do conceive of the theoretical model as a kind of random distribution, i.e. "doing a $\chi^2$-minimizing fit" is an estimator

but this is an implementation detail of how we do experimental tests of theories, not an inherent property of the theory

@SillyGoose If I squint that much, my eyes are just closed :P

also that "definition" of physical theory is deeply questionable

we're not doing physics by having hundreds of highly mathematical models compete against each other to see who can fit the data best

Silly Goose

14:41

well it seems quite hard to come up with just one that accomplishes that goal :P. but i am thinking if we did have two we would choose between them by seeing which fits empirical observations best

ACuriousMind

as much as I love mathematical physics, trying to pretend physical theories - the ones that actually exist in the world not the ones you wish there were - are anywhere close to being mathematical objects in any rigorous sense strikes me as deeply misguided

SnoopyKid

Hello,

A static load or a gradual load where the weight start from zero to 2 tonnes to break long bone and a sudden load or an impact load where we instantly put a 1 tonne load on a long bone, the force of both of these are the same. Correct?

Silly Goose

@ACuriousMind well the triplet constructed I intended to be some toy model of the situation to see if any interesting insight could be gained, not so much to faithfully reconstruct reality

ACuriousMind

@SillyGoose yes, but...how often does that happen? The history of science is not the history of people developing theories in a vacuum then testing them against each other during experiment. For most of history, we've had a plethora of observations no one could explain and all the physicists were doing was trying to come up with any model that correctly predicted the strange things we observed

and then the great controversies were often things we can't even really follow today, such as the vis viva discussion where I still don't really understand what the philosophical hang-up was about "true quantities of motion".

@SillyGoose Yes, but even for a toy model I should have some chance of seeing how to extend it to reality

but in reality a physical theory is not an algorithmic thing, there is so much choice and ambiguity involved in when to do what kind of approximation and what to assume about the systems under consideration etc. that I think any attempt to pretend a physical theory is just some kind of mathematical object is just...wrong

it's not an oversimplification, it's a fundamental misunderstanding about how science is practiced

Ryder Rude

15:16

anyone wanna friendly debate on consciousness philosophies?

i will argue for the "consciousness is fundamental" side and u can argue for a different side

@SnoopyKid if the time interval is the same and the impulse is the same, the average force is the same

implulse = total momentum imparted to the long bone

Slereah

15:45

If you want to see some actual examples of "Trying to fit theoretical values to experiments", you can look at the PPN approximation

That's basically how it works in GR

Alternatives to general relativity

Alternatives to general relativity are physical theories that attempt to describe the phenomenon of gravitation in competition with Einstein's theory of general relativity. There have been many different attempts at constructing an ideal theory of gravity. These attempts can be split into four broad categories based on their scope. In this article, straightforward alternatives to general relativity are discussed, which do not involve quantum mechanics or force unification. Other theories which do attempt to construct a theory using the principles of quantum mechanics are known as theories o...

^example

Parameterized post-Newtonian formalism

In physics, precisely in the study of the theory of general relativity and many alternatives to it, the post-Newtonian formalism is a calculational tool that expresses Einstein's (nonlinear) equations of gravity in terms of the lowest-order deviations from Newton's law of universal gravitation. This allows approximations to Einstein's equations to be made in the case of weak fields. Higher-order terms can be added to increase accuracy, but for strong fields, it may be preferable to solve the complete equations numerically. Some of these post-Newtonian approximations are expansions in a smal...

^and the experimental values

naturallyInconsistent

@ACuriousMind At least a big part of it is the "permanence of conservation" kind of arguments coming from not realising that energy has extra ways to be transferred away, which is only to be expected from them being ignorant of. But that then leads to all kinds of "how can you say that something that is only sometimes conserved as being of any fundamental importance?" and then it goes into the argument over "fundamental"

@SnoopyKid no, and that is not how questions of that kind arise.

@SillyGoose even your statement here already assumes that there is time-independence, and just one single distribution to choose from.

@qwerty I don't understand why that ought to make you less patient with code than physics. SQL is not an ugly language; it can be quite nice to use even in raw form. However, you need to have its mindset, and the data must already be organised in a way that naturally fits the database. That is, if you already have the mindset for SQL, then the vast majority of the headaches come from someone having the data stored in a stupid architecture, rather than something inherent to SQL itself.

@qwerty oh dBeaver is quite new! It shouldn't be terrible to use?

SnoopyKid

16:11

@RyderRude thanks ryder

I think this is a good answer "if the time interval is the same and the impulse is the same, the average force is the same
implulse = total momentum imparted to the long bone"

naturallyInconsistent

@SnoopyKid that's an illusion of an answer, of course.

Ryder Rude

@SnoopyKid in general, we have : average force x time interval= momentum imparted

Silly Goose

16:42

so what do you all think physics is all about :P

ACuriousMind

@SillyGoose ...describing the world around us?

constructing working models of nature

Nairit Sahoo

Does anybody know how exactly does the Wick's theorem change for complex scalar field?

Also, I know that the Feynman rules of complex scalar field theory have an orientation but I don't understand why: it wasn't the case for real scalar fields

ACuriousMind

@NairitSahoo I refuse to believe that any text would just state the Feynman rules with orientation without explaining what the orientation represents

Nairit Sahoo

@ACuriousMind It does. It just says that one represents the field and the other one is just the complex conjugate of that.

The incoming lines are the field and the outgoing ones are the complex conjugate of the field

I just don't see how that comes about...

ACuriousMind

What do you mean? You have to visually distinguish $\phi$ and $\phi^\dagger$, no?

choosing to do so by attributing an orientation to the diagrams is as arbitrary as choosing to represent the perturbative integrals with diagrams in the first place

Silly Goose

16:53

@ACuriousMind do you think most participants of physics research today have their work centered around describing the world around us?

ACuriousMind

@SillyGoose usually a very specific and small part of it, but sure

well, I guess for some it's rather a very unspecific and big part of it if you're doing stuff like cosmology :P

Slereah

most of them are just working on very down to earth stuff

like semiconductor or thermal isolation or optics

working on the theory of the mechanical properties of hot dogs

Silly Goose

it has been challenging to find an idea that i can personally believe is worth pursuing. almost everything can be interesting, but few things seem to matter in the end.

ACuriousMind

but the average physics research is less about grand questions and more like "let's see if we can observe this weird prediction for some specific material", "let's see if we can make our laser pulses even shorter" or stuff like that

Slereah

If you want things to matter you should have become an engineer :p

or work on something important and boring like meteorology

Slereah