Learning to learn. Physics PhD style.

[RD]

*Please move this to the most appropriate sub-forum, if this fits better somewhere else.

I'm kinda a fan of rationality and critical thinking.
And it seems (from the many posts by Jacob so far), there's a certain clarity of thinking that comes along with going through and learning/practicing of the scientific method/rigor.

I'd like to gain that kind of "mental powers" and have been expanding my areas of reading and forming my mini latticework framework. (Cognitive biases seems to have high mileage returns and many more to go)

So the question, since I'm quite weak in the sciences, how should I go about getting the (mental) education of having gone through the physics PhD path, DIY style.
My experience of physics from school is really bad, since it's boring and no way relevant to real world calculations (from the bare bits I recall, gross estimation models and not really teaching the principles of physics, more like manipulating formulas with no understanding of it's derivation)

I do want to learn about Physics, and not just the "scientific mental powers" bestowed.

I have this feeling I'm not being very clear on my goal, but please question away and I hope to use the forum as a sounding board to gain further clarity on this quest for clarity

[jacob]

First read this 10 times! Then think about it

http://www.colorado.edu/physics/phys481 ... gs/vh1.pdf

I have worked and studied at several universities and I never seen anyone teaching qualitative understanding and a focus on abstract fundamental principles as the primary purpuse.

It is, as you said, a focus on equations and particulars backed by little but hot air.

The normal educational process is a selection process in which people [hopefully] mature with experience. By this I mean having seen lots of different examples of the same thing, people start picking out the similarities rather than the differences. People start to recognize what's important about a question and what's just details. Normally this is a process that takes a good 10 years of study, at least!

You can tell inexperienced people from experienced people by the former's focus on equations---"The answer is given by these [complicated] equations"---and the latter's focus on the physical problem---"This diagram shows the relevant degrees of freedom from which it's easy to see that X and Z are conserved quantities." Unfortunately, the former group suffer a bit from the Dunning-Kruger effect confusing their technical ability to manipulate and compute equations without an actual understanding [of the potential/full scope] of the problem. They are often right, but when they're wrong, they're really really wrong. Conversely, the experienced physicist is almost never wrong and if they are, it's not by much.

Part of the physics curriculum is to start from the simple models and then move onto ever more complex models that removes more and more assumptions of the simple model. This teaches the physicist that "the problem is probably never as simple as initially assumed". Conversely, engineers, premeds, biologists, geologists, etc. who take a few courses in the physics department tend to miss this "meta-message" and consequentially think they know everything there is to know about the subject---that they got the whole story, you know, "the one true answer!"---based on a couple of courses. What is worse is that they have no fear of using this incomplete knowledge to slap on an equation, compute a number, derive wide-ranging conclusions, and claim they're doing science. The recent thread on global warming showed some examples [of using equations and graphs to cover a lack of qualitative understanding] in the links posted by Seneca and Steveo73.

Basically, you want either enough qualitative understanding to use technical skills with authority (some PhDs, a few MScs) or just enough to understand when you're on thin ice and should avoid making definite conclusions (some MScs, most physics BSc), but not so little as to be confident in your technical ability alone (Dunning-Kruger effect).

Unfortunately, I don't know of any way to skip the technical/quantitative training and go directly the qualitative education. Doing so risks being able to blabber without really having any quantitative understanding. Kinda like a high school nerd who has read too many Brian Greene books on string theory ;-P This is just as bad because you won't know whether your qualitative understanding is quantitatively correct. The general approach is thus to do lots of technical practice and the hope that somehow the person will put the pieces together down the line.

[slowth]

RD, we're in the same boat. I have an interest in physics, but it is completely unrelated to my job, so the year of physics courses in undergrad were lacking, to say the least. At least we realize this, and aren't going into it full Dunning-Kruger style.

Since we have to start somewhere, I think the Feynman lectures are reasonable compromise of quantitative/qualitative instruction. He doesn't just bring up a specific situation and then poof here comes an equation from thin air to solve this specific situation. No, quite the contrary, he actually derives the formulas and includes much qualitative explanation. Since Jacob is the expert, I'm curious what he has to say about Feynman, but I think it's a great place to begin.

Feynman Lectures Vol. I

[jacob]

First, I think critical thinking and the kind of "thinking maturity" can be developed in any field if one is introspective and reflective. The transition/development is made a'la the six Cs I discuss in the book. Eventually one realizes that all those techniques tend to build on a much more limited set of basic principles. Whether it's physics, swordfighting, or sailing, this holds.

In my opinion and experience, physics has some advantages over other fields though.
1) It's based on math which is the language of quantitative reasoning.
2) It's based on measurable reality. (The scientific method enters here).
3) It's the most fundamental of all sciences in that it deals with the basic laws of the universe that all other fields take from and build on.

Physics has 1-2-3. Compare to math which only has 1 and 3; this can lead mathematicians to some funny conclusions when it comes to reality. (I had a bunch of math grad students in my nuclear physics class once. Brilliant technicians. Utterly clueless about what they were actually deriving.) Or compare to engineering which only has 1 and 2 which can lead to the Dunning-Kruger effect described above.

It's a natural philosophy. It's natural in the sense that it deals with nature---the real world---not imaginated or constructed worlds. It's philosophical in the sense that it deals with ____how___ to think about it. Of course, physics is not great at all when it comes to disciplines that don't appear to be abstractable to fundamental principles, e.g. paintings, literature, ... (There are examples of physicists who have turned to those for some interesting results, e.g. Alan Lightman who's a writer, and Feynman who painted.)

Books, books, books, ...

Read the link I posted at the top again. If the goal is to gain a qualitative understanding and learn "how to think like a physicist" rather than learn a bunch of equations that can be unthinkingly applied to various situations, I would focus on trying to solve lots and lots and lots of problems within some relatively easy field.

I would also strongly recommend against simply memorizing how to derive various theorems. This only teaches you how to BS your way through an exam. At one of the universities I worked at this was the standard philosophy. The students got through some advanced material this way but 80% of them couldn't solve the simplest of problems.

If, however, you can develop your own proof/demonstration instead of simply regurgitating the one in the textbook, this is certainly to be commended and shows a great deal of insight!

Classical mechanics is usually the first topic taught. This is the book I learned from. We spent an entire year on it. I think it is good, but only if you try and successfully solve most of the exercises
http://www.amazon.com/Introduction-Mech ... 521198119/

If classical mechanics isn't your thing, focusing on electric circuits and basic electromagnetics is another good field to master, e.g. moving wire with resistance R in a magnetic field subjected to initial impulse, graph the voltage as a function of time, etc.

I would stay away from quantum mechanics. It takes a long time to finally "get it". I bet 95% of those who pass advanced QM can't explain to their mother (or father) what the experimental interpretation of a commutator is and what implications that has. This is one of the most fundamental principles of QM and yet few will have grokked why that is. This didn't dawn on me until well into grad school when I accidentally read a more philosophical treatise on QM. I told the other grad student next to me about it and I think I just blew his mind. Before that it had just been a bunch of crazy-ass group theory rules that I thought had been handed down from above, literally and figuratively.

By practicing, you should basically reach a situation in which if I give you a list of objects, e.g. a bar, a ball, a magnet, ... and tell you that the bar is spinning, the ball is magnetic, the velocities are so and so, ... then you can solve for the dynamics of the situations.

You should absolutely focus on text problems that are not standard. Practice practice practice. If something can be answered with a ScanTron, you're not really learning anything. You're just testing your IQ and memorization ability.

Frankly, I don't think the Feynman lectures are going to give you the required exercise. I'd rather read those to consolidate my understanding and renew the enthusiasm AFTER having fully understood the basics.

For a much more thorough treatment, I would recommend Landau and Lifshitz 10 book series on theoretical physics. When it comes to physics, pretty much anything that came out of Russia is superb.

For German speakers, Walter Greiner (and his dynasty) wrote a very similar series which I think is even better since it contains a great deal of calculated examples.

I would not use any of those two series on freshmen or juniors though. Too heavy.

TL;DR - The key is practice practice practice. Reading books ain't gonna cut it. The best books are the ones that get you to practice more.

[ebast]

Curious about:

enough understanding to use technical skills with authority (some PhDs, a few MScs)

what happens with those PhDs/MScs who don't have that understanding? I assume they're still out there doing work too? Writing textbooks?

Is there a dead giveaway when you meet one at a party?

[jacob]

@ebast - They become management consultants And yes, they'll be wearing a suit and drinking Heineken.

[RD]

It's gonna take me awhile to digest all that was written and the reference links.
So this will me a short reply.

@Bigato, I'm already doing software in my day job, and have developed a sense to seeing underlying principles (there aren't really that many in typical CRUD software even with scale) and finding similar patterns in the real world.

@Jacob, digesting in progress... ;p

[Seneca]

Part of the physics curriculum is to start from the simple models and then move onto ever more complex models that removes more and more assumptions of the simple model. This teaches the physicist that "the problem is probably never as simple as initially assumed". Conversely, engineers, premeds, biologists, geologists, etc. who take a few courses in the physics department tend to miss this "meta-message" and consequentially think they know everything there is to know about the subject---that they got the whole story, you know, "the one true answer!"---based on a couple of courses. What is worse is that they have no fear of using this incomplete knowledge to slap on an equation, compute a number, derive wide-ranging conclusions, and claim they're doing science. The recent thread on global warming showed some examples [of using equations and graphs to cover a lack of qualitative understanding] in the links posted by Seneca and Steveo73.

Come on Jacob, this smacks of the old PhD Physicist "holier than thou" bullshit.

Engineering school does exactly the same thing you describe physics does. You take 3 semesters of calc based physics as pre-reqs, and those physics classes have idealized problems full of assumptions to make them manageable for learning. Engineers then move on to engineering-based problems that start removing the assumptions, and add inconvenient little things like friction.

I'm not a physicist, so I don't really know what the last two years of a physics BS look like. But my perception is physics dives more in to derivations where engineers spend that time removing assumptions and learning how to apply equations to problems without as much emphasis on where the equation came from. This seems to match what you're saying to a point, but it ignores why engineers are trained to work this way.

I work with a bunch of physicists, but I don't find the guys working in industry to have this opinion of engineers, and never had the discussion where we dove in to the details of the differences in education. Usually we're too busy getting our asses kicked by the same problems

In the semi industry, physicists tend to end up silicon device modelers, because...within silicon electrons behave much closer to modeled data than they do once they leave the chip. I definitely find them no better with their "math tool", and in fact, if anything, find physicists more likely to rely on equations than the lab data engineers usually favor when solving a real world problem. I do know from experience, you spend a lot more time arguing with PhDs, and especially academics, (keep these guys off direct projects with real time lines at all costs!), when real world data doesn't fit models.

The GW discussion is something I'm simply not invested in and I was very up front in that thread I've never made a project of it.

[jacob]

@Seneca -

The difference I think is that the physics path goes like this, for example,
Classical mechanics -> Special Relativity -> Quantum Mechanics
The focus is mainly on how to _build_ a theory and that "theories come about based on certain assumptions about what's being modelled". So a physicist would be inclined to question whether they got the big picture, whether they're missing some aspect of reality in their modeling, whether there are latent variables or unconsidered factors.

Whereas in engineering, the path goes like this
classical mechanics -> classical mechanics with friction -> classical mechanics with friction and bending and stretching ...
The focus here is on how to _use_ a theory. Does this apply to this situation. Is my zero-friction assumption correct (not is the speed of light identical regardless of which inertial frame it's measured it)?

My point above was the engineering approach will be more correct until you build a lightspeed spaceship. So scientists are needed to say "Wait a minute! You're entering a domain where your entire theory is wrong".

This is what sent up red flags when I read the GW links: "Hey wait, this person hasn't considered that the absorbing atom is in a gas. Wait, this calculation ignores the boundary conditions." My background caused me to notice incomplete description which apparently the authors were happily blind to.

Another way I'd put it is that physics (or science) starts from reality and goes to theory. Engineering starts from theory and goes to reality. For practical reasons, each use a bit from the other side. In order to do science, you often have to do some engineering. In order to do engineering, you often have to some science. The question is where the main focus is. For example, you say you'd be happy to take a precise heuristic from the lab data over some equation. This is because you need a useful answer NOW. Scientists would consider it unsatisfactory to only have a theory based on heuristics because they consider their understanding (which is what's important to them) incomplete. Hence heuristics = much work to be done still.

[Seneca]

This post I agree with completely. (EDIT- except for the boundary conditions comment. Any engineer that has done real world modeling knows the sensitivity of this, and almost surely better than a theoretical scientist!)

There are differences in the educations for a good reason.

I guess we are down to arguing over what "science" is. I'd argue applied science is still science, but there is definitely a big difference between a theoretical scientist and an engineer.

I'm very practical minded BS engineer and early PhD dropout though...and thus probably don't care nearly as much as the physicists do.

[Seneca]

Since you've brought it up a couple of times now, that engineers use equations with "no fear" or don't respect boundary conditions and assumptions, I'll give you an example of just how serious engineers learn boundary conditions are.

In my world we frequently use the Arrhenius Equation for estimating (not predicting!!) time to failure over changes in temperature for electronics.

Just yesterday I was talking to some colleagues about a problem where a couple of engineers didn't have all the data but had to come up with something to give to a customer to decide whether or not the part would fail in the expected lifetime of the product. The boundary conditions were not easily known, and thus the coefficients they used to solve it were wrong, and because the equation is a power law, the result was way wrong. This led to a >$10mil problem, lawsuits, making the MSM with their fuck up when the failures started to cascade etc etc. There is no hiding or denying this mistake. Believe me...an engineer that has been around the block will have learned to respect the interplay of models/equations/boundary conditions/assumptions deeply and unforgettably.

In comparison, what happens if a scientist gets a boundary condition wrong in a GW model? Would she even know? Is it possible she ever be forced to admit she was wrong by incontrovertible proof like the Challenger exploding?

[jacob]

Here's the example (of ignoring boundary conditions) I was talking about:

Wrong/incomplete way to think about it:
CO2 is very absorbent in a particular wavelength that depends exclusively on the quantum dynamical properties of the CO2 molecule. Once radiation at that wavelength is fully absorbed, there's no more radiation to absorb. We can therefore compute the length is takes before, say, 99.99% of the radiation is gone. That length is X. The first conclusion to be drawn is that beyond the X distance, we can pile on as much CO2 as we like and it makes no difference because the radiation is already absorbed.
We can therefore compute how much many degrees of heating this absorption causes in the first X meters of the atmosphere since there is no heating above X. This number is very low and thus proves that CO2 does not cause global warming.

Right way to think about it:
The atmosphere consists of several layers worth a nominal X meters and layers of the atmosphere are not perfectly isolated from each other. If you heat X meters of the atmosphere, that X meters will diffusive heat the to the layer above it and heat that layer up as well. That layer, in turn, will heat the layer above it, etc. An analogy is a person under a blanket in a bed. Putting a blanket on heats the blanket until the emission from the blanket going out into the room matches the heat comes from below from the person. Throw another blanket on top of the first and the equilibrium adjusts again: between the top blanket and the room and the middle blanket; the middle vs the top and the person. The more blankets that are thrown on top, the higher the temperature at the person. The temperature at the bottom with the person never peaks out but keeps increasing with the number of blankets.

Empirical example: This is why Venus is so warm. Use the wrong boundary condition and the result would indicate that Venus is not warm at all. If you sent astronauts to Venus based on this theory of balmy conditions, they would die.

[Sclass]

So the question, since I'm quite weak in the sciences, how should I go about getting the (mental) education of having gone through the physics PhD path, DIY style.)

Wow. Okay I thought I'd try to pitch in some advice here.

As a bystander reading both Seneca and Jacob's posts on and off this thread I conclude that both of their fine intellects were not formed exclusively in some stuffy university environment even though they're quick to bash one another.

Perhaps you're looking in the wrong place to develop your intellect. If you want some of Jacob's mojo I'm not sure you'll find it by learning more physics or going into servitude for the grant mob.

Seneca and Jacob both challenge what most people take for granted. Whether it is climate change or just how society tells us how we must "win" they not only won't take what is served up, they try to reason through alternative ideas. I don't think they became who they are by just earning a diploma.

I liked the last two books by Nassim Taleb in this regard. They opened my eyes a bit. Depending on how well you teach yourself (flashbacks of self teaching tai Bo in the park come to mind) you may benefit by doing some reading.

There are all kinds of physicists and engineers. Good ones, enlightened ones, intellectually deep and shallow idiotic. I had a lot of shallow colleagues in science school. I encountered naive scientists in industry. I knew talented and idiotic engineers. I knew brilliant physicists who could work through the most unengineerable connundrum to just f.ck it up with an unrealistic execution. The only thing I can say is the best ones didn't learn all their tricks from school. They were always in learn mode.

I'm not sure it has a lot to do with a few years of intellectual force feeding at the university...but more how much reality one has been exposed to AND how much meaningful knowledge is gained from the experience. You may be able to DIY a lot of your enlightenment (hence your post is in the right place ).

[jacob]

As a bystander reading both Seneca and Jacob's posts on and off this thread I conclude that both of their fine intellects were not formed exclusively in some stuffy university environment even though they're quick to bash one another.

Absolutely. Here's a brief story of how I formed my so-called fine intellect (I take it as a compliment ).

I got my no-BS attitude from my dad although not nearly to the same degree. He is also not afraid of calling it as he sees it despite status/authority/etc.

I don't think I was very different from any other person when I graduated with a MSc in terms of attitudes and world-views, but due to the physics education I was able to break complicated problems down to their most fundamental aspects and make conclusions about them.

I retrospect, though, I do believe that even at that point, I have read far more books than most people. Furthermore, most of the books I read are non-fiction and I read many different categories. At my current age, I estimate I have read somewhere around 2500 books +/- 1000. A more accurate estimate is probably that I read about 500-750 pages per week. Now multiply that by 25 years. That's a lot.

A paradigm shift occurred in graduate school. This is when I discovered peak oil, anti-consumerism, reserve banking, ... which destroyed my standard world of consumerism, economic and technological growth, and a fair system.

I also learned a very important lesson doing actual research: "It was no longer possible to find an expert/professor that knew the answer if I was stuck".

So if you put all this together, it comes down to
1) No BS attitude
2) The ability to break problems down in parts and build them back up.
3) Having read a ton of material---a very broad informal education.
4) Knowing that nobody really have the answer and if I want an answer I have to find it myself.

ERE essentially came out of my realization that what I had been told about "life" was BS and my attempt to deconstruct and reconstruct it into something that made more sense to me.

[Seneca]

Wow, thanks Sclass.

I'll be the first to admit, I don't learn "PhD style". My style is decidedly "dirty fingernail"-

Try doing something I think I might be interested in.
Suck at it/fail
Decide if it is worth time.
Read on the topic a lot, keep doing/failing to truly figure out what is important and focus study.
Curse myself when I forget my book lessons "in the field" as I meld the two.
Apply professionally/get paid (when possible, which is almost always)

[Sclass]

You are welcome.

We share some common components on the BOM.

Learning is an ongoing process and it can happen in all kinds of places. Like prison.

Speaking for myself I didn't learn to learn by learning physics in school. I was a physics major for undergrad and I was very average. I had a bike accident without a helmet and had to become an engineer. To be fair though, I probably would not be where I am without my education. It's just that it is a small part of many lessons I've picked up. The whole world is a school. And I've been "schooled".

[jacob]

Unfortunately, "PhD style" is no longer well-defined. It's supposed to be an apprentice style education in "how to do independent research". In some cases this becomes more of a question of being thrown in at the deep end to learn how to swim, but that's fine if the personality for it is there (this was the way I learned, very hands-off).

Unfortunately, there are also PhDs which can be best characterized as glorified technicians. Cheap labor of the professor (a Masters for $25k/year is a good deal!) who is trained to become expert on one of the laboratory gadgets just executing whatever instructions the prof. hands down. This eventually results in the credentials, but in my opinion it was a waste of 4-6 years. Might as well have been paid twice that to do the same thing in industry.

My point: The PhD style education is essentially about "solving some original problem independently". Sometimes people get the education without the PhD and sometimes they get the PhD without the education. It's possible to get a PhD style education outside academia. You can do it on your own. You can do it in industry. The important aspects, I think, is in solving some problem that nobody else knows the answer to. Academia can help point in the direction of unknown answers but frequently, it's more about unknown questions---which you then answer

[Sclass]

Unfortunately, "PhD style" is no longer well-defined. It's supposed to be an apprentice style education in "how to do independent research". In some cases this becomes more of a question of being thrown in at the deep end to learn how to swim, but that's fine if the personality for it is there (this was the way I learned, very hands-off).

Unfortunately, there are also PhDs which can be best characterized as glorified technicians. Cheap labor of the professor (a Masters for $25k/year is a good deal!) who is trained to become expert on one of the laboratory gadgets just executing whatever instructions the prof. hands down. This eventually results in the credentials, but in my opinion it was a waste of 4-6 years. Might as well have been paid twice that to do the same thing in industry.)

Oh ouch. Yes. That sounds like my graduate experience. I actually built the scientific gadget. In the process I realized I was more interested in the machine than the science. So I went off to make a living designing signal processing systems. My thesis adviser was a grant kingpin, so he just let me do whatever I wanted while he lobbied in D.C. getting our huge research group money. I was managed by his loser graduates who never got out on their own...perpetual post docs. I guess what I failed to learn from my adviser was how to dominate big science. I did build an awesome machine that put the group on the forefront for many years. The kids at the university still use a hunk of my hardware twenty years later. The best thing about my arrangement is I had my own office, lab and machine shop where nobody would check up on me for months. I made the best of that. I built a lot of stuff back then. EE and physics had great stock rooms and I had a blank check.

My brother had the latter treatment. He was truly mentored. He would meet his professor (math) everyday at 8am and work equations with him till 12noon. Then his adviser would say, take that home and we'll see where you get by tomorrow. He did this four years. At graduation his research crashed. I spoke to him over a beer one night complaining about how I'd been abandoned in grad school and had to hack my way to a career. He said, "at least George taught you how to think for yourself". My brother had hit a wall. His adviser died. He was totally lost and went into a tech company to solve complex math problems that other people drop on his desk. He's literally a mental prostitute...or a human calculator whatever. And they do come by the next day to see where he is.

So it is fitting how you describe the importance of self teaching. A good mentor isn't always around. I didn't quite understand what "Ph.D style" meant.

[riparian]

I'm learning to learn qualitative social science style.

But there's so much reading. I'm a relatively fast reader, but I'm apparently a slower thinker than I'd thought. Every chapter is full of theories and ideas that I have to pause and consider the applications of. It seems that I'm learning more than the average bear, but I probably can't keep up the pace of reading/thinking for much longer. In the past I've gotten A's with skimming and memorizing, but everything I'm learning now is actually important.

Help?

[jacob]

@riparian -

Not much to do about it.

We used to warn prospects(*) that achievement is a function of time and brainpower (whatever particulars is required for a given field) and that it's possible to compensate for one with the other until you reach a point where you're thinking round the clock. And people do tend to reach that point if they keep going onward and upwards.

(*) Someone made the mistake of asking "how smart one needed to be to study physics" )

As they say in professional sports, it never gets easier, you just go faster. It's the same in academic pursuits. There's are no known short-cuts to 10,000 quality hours. However, if you already have some of the hours down, you can somewhat translate them to other fields, e.g. you wouldn't have to relearn calculus when going from chemistry to biology. In addition, the more fields you know, the more you can translate too.