Random Thread

[bradrn]

But given this, I still think the mathematics is fundamental to physics, in a way that it isn’t to other subjects.

I don't think we need to argue about how fundamental is is, but just to tease you, I'll mention one discipline where mathematics is central, and it sucks: economics.

I don’t know enough about economics as a field to properly discuss it.

Oh, and if you listen to Hossenfelder, string theory.

Yes, this is correct. String theory is a very elegant mathematical theory, which at the moment has absolutely no relationship to the real world other than the fact that it’s consistent with quantum field theory. And although string theory gets all the press, it’s far from the only field of ‘physics’ which is basically just maths — my department has a research group specialising in quantum information theory and black holes, and from what I’ve seen it’s not that different there.

That being said, being ‘basically just maths’ doesn’t invalidate a research area in any way. Pretty much the entirety of theoretical quantum computation is physically unrealisable at the moment, but that doesn’t mean it can’t be applied one day. I know quite a few people in quantum optics, and they spend most if not all of their time doing theory too. This kind of thing is why I feel that mathematics is so important in actually doing physics. I might summarise my position as: ‘physics is the mathematics of the real world’.

[Raphael]

For what it's worth, I find the results of the first few centuries of research in physics very interesting, and they are an important part of helping me understand the world in which I live, but the cutting-edge research in physics during my own lifetime seems to be mainly concerned with things that are either too small for me to find them interesting or to large for me to find them interesting.

[Moose-tache]

So I've been kiling time looking at minicars and autorickshaws. For example, this charming little meep-meep has a hook! Scroll down to spot the "white monkey" working at Known Motors, and the trophy case that apparently includes an Oscar. The common thread of these kinds of vehicles is that they all have 60V 1kW engines, less than half the power you'd typically find in a golf cart, or those little 50cc scooters you see Filipino kids zipping around on. And they run on lead acid batteries. This used to be a thing, back in the 70s: just string a bunch of ordinary car batteries together to run a motor. I actually don't hate that, as absurd as it is. The rise of lithium ion, sodium ion, LiFe, etc., is all about making EVs compete with gasoline on its own terms, while lead acid batteries will get the job done just fine if you accept that "the job" is getting you to the chippy, back home, and then immediately dying. Anyway, I found this gem and thought "who makes this?" I googled JuYou, and found out they primarily make food processing machines, like vegetable peelers. One of their biggest sellers is a pork sausage maker, which could also be the car with the seatbelts removed, I dunno. You usually have to google the brand, because they're extremely coy about it. But I googled too soon, because if you keep scrolling, the Alibaba product page admits that they make "vegetable machines," with photographs, and starts bragging about their great customer satisfaction. They even let you in on some conversations where their customer service rep "Betty" gets customer feedback on their garlic peelers, like "not bad."
Should a vegetable processer manufacturer be making EVs? Should they be simultaneously cagey and proud of the fact that you can also buy a potato peeler from them? Is it alarming that they have to estimate the door swing angle when they, my brothers in Christ, made the door? Should they ask obviously real and extant anglophonic employee "Betty" to fix the grammar mistakes in their own copy? I can't answer any of these questions, and I hope I never can.

[Ares Land]

I am suddenly overcome with the need to buy one of these things as soon as possible.

[Moose-tache]

Right? Alibaba is basically Koyanisqatsi with all the morals reversed.

[rotting bones]

This must be a trending topic or something... I just read a book on dark matter, and also recently discovered Hossenfelder.

I've been irregularly watching Sabine Hossenfelder for around 4-5 years now. I watch other physics channels too. There's one called Anton Petrov that's not skeptical at all, but disseminates news about the latest cosmology papers.

Here's Hossenfelder on dark matter. It's a good backgrounder: she gives the evidence for dark matter, also some things particle theories don't explain.

Sabine Hossenfelder is an advocate for MOND, Modified Newtonian Dynamics. (I didn't remember this. She said it again in a recent video.)

IIRC the reason to think dark matter exists is that the velocities of stars on the peripheries of galaxies is higher than gravitational force can explain.

The dark matter hypothesis is that invisible ("dark") matter present in galaxies is causing the peripheral stars to accelerate.

According to MOND, there is no invisible matter. The law of gravitation we're working with is wrong. Eg. If the gravitational force varies inverse linearly with the distance over very large distances instead of the inverse square of the distance, the math works out very nicely. One such proposal is RMOND.

Of course experimentation is important! I I’ve spent a lot of time over the past year or two working in various labs (in both chemistry and physics), sometimes quite intensively, and I’m very familiar with what goes on there. My current research group is mostly experimental, too. (As a matter of fact I helped out in the lab just today.) And personally I enjoy experiment quite as much as theory, if not more.

Your account of math doesn't seem to address the problem with overfitting: https://en.wikipedia.org/wiki/Overfitting If the math fits the data even better, isn't that always a good thing? Sadly, it's not. There was one German physicist who tried to create brand new equations that fit the data even more closely than the standard models. Of course, it all failed to generalize. Statistics shows very clearly how the models are distinct from the data.

[bradrn]

IIRC the reason to think dark matter exists is that the velocities of stars on the peripheries of galaxies is higher than gravitational force can explain.

That’s one reason dark matter might exist. There are many others: https://en.wikipedia.org/wiki/Dark_matt ... l_evidence. My understanding is that dark matter is more widely accepted as it provides an explanation for all these facts at once, whereas MOND only explains a few of them.

Of course experimentation is important! I I’ve spent a lot of time over the past year or two working in various labs (in both chemistry and physics), sometimes quite intensively, and I’m very familiar with what goes on there. My current research group is mostly experimental, too. (As a matter of fact I helped out in the lab just today.) And personally I enjoy experiment quite as much as theory, if not more.

Your account of math doesn't seem to address the problem with overfitting: https://en.wikipedia.org/wiki/Overfitting If the math fits the data even better, isn't that always a good thing? Sadly, it's not. There was one German physicist who tried to create brand new equations that fit the data even more closely than the standard models. Of course, it all failed to generalize. Statistics shows very clearly how the models are distinct from the data.

Overfitting doesn’t need any specific explanation. If a model is disproved by further evidence, it’s wrong irrespective of whether it was ‘overfitted’ or not; similarly if a model’s predictions are true, that’s evidence for its truth in any case.

[rotting bones]

That’s one reason dark matter might exist. There are many others: https://en.wikipedia.org/wiki/Dark_matt ... l_evidence. My understanding is that dark matter is more widely accepted as it provides an explanation for all these facts at once, whereas MOND only explains a few of them.

I'm only reporting Sabine Hossenfelder's view. She might have answers to these that I can't remember.

Overfitting doesn’t need any specific explanation. If a model is disproved by further evidence, it’s wrong irrespective of whether it was ‘overfitted’ or not; similarly if a model’s predictions are true, that’s evidence for its truth in any case.

You can explicitly reason about whether a model is overfitting or not: https://en.wikipedia.org/wiki/Statistic ... larization

[bradrn]

Overfitting doesn’t need any specific explanation. If a model is disproved by further evidence, it’s wrong irrespective of whether it was ‘overfitted’ or not; similarly if a model’s predictions are true, that’s evidence for its truth in any case.

You can explicitly reason about whether a model is overfitting or not: https://en.wikipedia.org/wiki/Statistic ... larization

Sure, and intuitively it’s often obvious too. But that has nothing to do with whether a model is wrong or not! There’s a correlation, of course, but ultimately it’s up to seeing whether its predictions are verified or not.

[zompist]

IIRC the reason to think dark matter exists is that the velocities of stars on the peripheries of galaxies is higher than gravitational force can explain.

That’s one reason dark matter might exist. There are many others: https://en.wikipedia.org/wiki/Dark_matt ... l_evidence. My understanding is that dark matter is more widely accepted as it provides an explanation for all these facts at once, whereas MOND only explains a few of them.

It's not quite that clear-cut. MOND explains galaxy rotation better, but it explains many of the other things less, or not at all.

[bradrn]

IIRC the reason to think dark matter exists is that the velocities of stars on the peripheries of galaxies is higher than gravitational force can explain.

That’s one reason dark matter might exist. There are many others: https://en.wikipedia.org/wiki/Dark_matt ... l_evidence. My understanding is that dark matter is more widely accepted as it provides an explanation for all these facts at once, whereas MOND only explains a few of them.

It's not quite that clear-cut. MOND explains galaxy rotation better, but it explains many of the other things less, or not at all.

Maybe I should have been more clear, but that’s essentially what I was trying to say.

[rotting bones]

Sure, and intuitively it’s often obvious too. But that has nothing to do with whether a model is wrong or not! There’s a correlation, of course, but ultimately it’s up to seeing whether its predictions are verified or not.

This makes the model less like a description and more like a prediction machine.

[bradrn]

Sure, and intuitively it’s often obvious too. But that has nothing to do with whether a model is wrong or not! There’s a correlation, of course, but ultimately it’s up to seeing whether its predictions are verified or not.

This makes the model less like a description and more like a prediction machine.

The whole point of physics is to make prediction machines, though. A model’s no use if it can only tell you stuff you’ve already verified. (This is the reason why many people don’t like string theory etc.)

[Raphael]

And another blog post by me. I should warn you that this one might hurt some readers' religious feelings.

https://guessishouldputthisupsomewhere. ... on-belief/

[Raphael]

And another blog post by me. I should warn you that this one might hurt some readers' religious feelings.

https://guessishouldputthisupsomewhere. ... on-belief/

Slightly provocative follow-up:

https://guessishouldputthisupsomewhere. ... -morality/

[doctor shark]

Sure, and intuitively it’s often obvious too. But that has nothing to do with whether a model is wrong or not! There’s a correlation, of course, but ultimately it’s up to seeing whether its predictions are verified or not.

This makes the model less like a description and more like a prediction machine.

The whole point of physics is to make prediction machines, though. A model’s no use if it can only tell you stuff you’ve already verified. (This is the reason why many people don’t like string theory etc.)

I mean, that's Science in a nutshell: devising hypotheses and extensively testing them to see if there are holes, but even there being holes is not inherently fatal (just limiting their applicability).

More: show

For example, Onsager theory, a classic liquid crystal theory for phase transitions, assumes noninteracting rod-shaped particles (which is very much not the case in real life). And it generally works quite well for many systems, though not for all. Similarly, the Frank free energy density equation can be either expanded or simplified from its "classical" three-constant form, either including a fourth term (for a very strange case of deformations), reducing it to one elastic constant, or adding other terms, depending on what's needed.

[bradrn]

This makes the model less like a description and more like a prediction machine.

The whole point of physics is to make prediction machines, though. A model’s no use if it can only tell you stuff you’ve already verified. (This is the reason why many people don’t like string theory etc.)

I mean, that's Science in a nutshell: devising hypotheses and extensively testing them to see if there are holes, but even there being holes is not inherently fatal (just limiting their applicability).

More: show

For example, Onsager theory, a classic liquid crystal theory for phase transitions, assumes noninteracting rod-shaped particles (which is very much not the case in real life). And it generally works quite well for many systems, though not for all. Similarly, the Frank free energy density equation can be either expanded or simplified from its "classical" three-constant form, either including a fourth term (for a very strange case of deformations), reducing it to one elastic constant, or adding other terms, depending on what's needed.

I can’t tell… are you agreeing with me, or with rotting bones? (Or with neither?) But I accept that many — no, all — models have limited applicability; you may not be able to predict facts outside that area, but within the area it works fine.

[doctor shark]

The whole point of physics is to make prediction machines, though. A model’s no use if it can only tell you stuff you’ve already verified. (This is the reason why many people don’t like string theory etc.)

I mean, that's Science in a nutshell: devising hypotheses and extensively testing them to see if there are holes, but even there being holes is not inherently fatal (just limiting their applicability).

More: show

For example, Onsager theory, a classic liquid crystal theory for phase transitions, assumes noninteracting rod-shaped particles (which is very much not the case in real life). And it generally works quite well for many systems, though not for all. Similarly, the Frank free energy density equation can be either expanded or simplified from its "classical" three-constant form, either including a fourth term (for a very strange case of deformations), reducing it to one elastic constant, or adding other terms, depending on what's needed.

I can’t tell… are you agreeing with me, or with rotting bones? (Or with neither?) But I accept that many — no, all — models have limited applicability; you may not be able to predict facts outside that area, but within the area it works fine.

Agreeing with you, My comment was mostly pointing out that models often make a lot of assumptions that limit their general applicability. And they can be both "descriptions" and "prediction machines" (by using the description to predict), but, again, everything has limitations as to where it can be used best.

[Raphael]

https://en.wikipedia.org/wiki/Mummies_(2023_film)

Looks like they finally made a movie with good guy mummies. After all those books and movies and TV shows with vampires as protagonists, I had seriously been waiting for this. I mean, say about mummies what you want, at least they don't go around sucking people's blood. I think before this movie, the only work in the "mummies as sympathetic characters" subgenre was one old episode of Chip 'n Dale: Rescue Rangers.

[malloc]

An odd question but something I have always wondered. Plenty of chemical substances cause euphoria or pleasure when consumed. But does anything have the opposite effect, making you miserable? It has always seemed odd to me that drugs with largely opposite effects, like stimulants versus depressants, nonetheless both feel good when consumed.