Academic Poll: Essentials of Physics

I really don't like the idea of forcing students to take a "Modern Physics" course, maybe just because the one I was forced to take was unrelentingly awful (with the highlight being an explanation of the important role of consciousness in wavefunction collapse, followed by an account of the instructor's own continuing efforts to prove the Heisenberg uncertainty principle wrong -- so I'm sure it was an unusually bad instance of the genre). But in general it seems like these courses are sort of "here's a handwavy explanation of what a reasonably up-to-date physicist might have thought about the world in 1930," which is interesting as history but bad as physics -- it sets up all kinds of misconceptions that have to be knocked down later in a quantum mechanics class. And since most QM classes are "let's solve the Schrödinger equation a lot and try not to think too hard about what it means," the leftover conceptual muck from the "Modern Physics" classes doesn't really get cleared away unless students do it on their own.

@onymous

I agree that most "Modern Physics" courses are poorly taught (our coverage of modern physics is jammed into the last part of the last semester of our freshman/sophomore introductory track, so it usually gets short shrift), but where else are the students gonna learn what the ingredients of the standard model are? Or an understanding of special relativity?

I think a passing knowledge of the fundamental forces & particles, and special relativity qualify as "it would be kind of ridiculous to award a physics degree to a student who [doesn't know this]".

When I took "Modern Physics," it didn't teach the Standard Model at all -- I suspect the latest development that was covered in that course was the Davisson-Germer experiment. I agree that people should learn something about the Standard Model, but I don't really know when. As for special relativity, it really should be taught in a mechanics class (and was, when I took mechanics, if I remember correctly), as well as in E&M classes. I don't see any reason to teach it in a class that's largely about the early development of quantum mechanics.

The additional course I recommend would be a math methods course (Arfken or similar). I took such a course in grad school, but it would have been really nice to see it as a sophomore or junior, before it came up in quantum mechanics and in upper level E&M (FD: my undergraduate alma mater used Jackson for the upper level E&M course, which was clearly inappropriate). That would also reduce the tendency of so many physics courses to turn into thinly disguised math methods courses, and therefore allow more time on actual physics.

Ideally, there would also be a set of elective courses from which students would choose 1-3 courses (depending on schedule constraints) in more specialized topics. Possible choices from your list are the solid state course, the AMO course, nuclear/particle physics, GR/cosmology (at something below Misner/Thorne/Wheeler level), and circuits/electronics. Other options would be plasma physics (Chen or similar) and astrophysics (Shu or similar).

I think there should be a separate list for physics majors and for non-majors. I am with onymous in not seeing the point of physics majors getting a watered down (and often incorrect) version of the material to "prepare" for the real deal, but these courses do make sense as exposure to basic concepts for outsiders. Modern physics is just one particularly bad example of that, but the thermodynamics- stat mech sequence is equally bad. Not sure why I have to spend so much energy when teaching stat mech in fighting all kinds of 19th century intuition the students acquired the previous course (as in the famous "heat is a verb and not a noun"). The root cause is the idea of teaching things in historical order, which makes no sense to me. At least phlogiston is out of the physics curriculum by now...

More controversially, I am wondering if all physics majors have to take the advanced lab courses, electronics etc.. By the time they are junior/seniors many students have a clear idea what they'll do next and maybe it makes sense to let them specialize. I took many required courses which, it was already clear at the time, would serve no purpose for me as a theorist, and in fact they did not. I could have used at least another math methods course though.

I agree with Eric on the need for some ODE and PDE courses. We hide most of our ODEness in our Fowler-level classical mechanics course, but follow it up with a physics PDE coruse.

Moshe, your last controversial part is definitely controversial. I think it is important that physics majors (if we falsely assume that the goal for all of them is grad school) to take advanced lab courses. As painful as those courses can be, we want all of the physics grads to have some idea how experiments are done and the equipment they use. Even if theorists never really set foot in a lab, I think its very beneficial for them to have some understanding of how us experimentalists work, so it can make our communication better.

As a suggestion for courses left off of the list, I would recommend some type of computational physics course, whether it be taught in C++, or even just Matlab. Those courses were offered, but not mandatory in my undergraduate department. After taking computation physics in grad school, then writing a simulation to go with my experiment for my thesis, I found that doing computational work really helps look at problems in a slightly different light, one that I have found very valuable!

Also, let me add to the list a decent course in computational methods/numerical analysis/basic programming. Probably basic knowledge for any science/engineering student.

Steven: I am not entirely committed to this point, but in this spirit it seems to me that what you wrote is a rationale for requiring basic lab courses. This is not something anyone would dispute. Advanced lab courses, and specialized knowledge like electronics and electrical circuits, seems to me potentially less crucial. In a similar way that any experimentalist should know how to quantize the harmonic oscillator, but perhaps does not need to know about Feynman diagrams.

As a confirmed experimentalist, I tend to be in favor of requiring some sort of advanced lab at the undergraduate level, for a few reasons:

1) I've heard from several people that most students who go to graduate school do so under the impression that they will be theorists-- several people have even recommended that students with a demonstrated interest in experimental physics should make a point of highlighting that in graduate applications, because it makes them stand out more-- and a lot of these people are wrong. They end up drifting into experimental groups of one sort or another, and find a home there.

2) Introductory labs tend to be, well, introductory. They don't involve much complicated equipment, and they're often very cookbook-y. Which means they're really not much like experimental physics at all, and don't give students a real idea of what experimental physics is like. Which is part of why so many students go to grad school thinking they're going to be theorists.

3) Advanced lab skills can be very valuable. In much of academia, even theorists have to teach lab sections, and things like knowing how to use an oscilloscope or do data analysis can come in very handy. Outside academia, there are a lot more jobs involving experimental-type skills than pure theory jobs.

For those reasons, I think it's very valuable for students to see an advanced lab course at some point. This can often be combined with something else-- at Williams back in the day, the junior-level quantum course had an advanced lab attached, which served that purpose.

I would also argue for having all physics majors do some sort of numerical modeling/ computation, though that's much easier to integrate with a course in a specific subject area, and thus less likely to need a stand-alone course.

Which means they're really not much like experimental physics at all, and don't give students a real idea of what experimental physics is like. Which is part of why so many students go to grad school thinking they're going to be theorists.

Implicit in this seems to be the assumption that in their non-lab courses students are doing things that give them a real idea of what theoretical physics is like. Which is completely wrong.

onymous, that might be even a better reason why most students going to grad school want to be theorists...

Chad, I think I agree with most of this, except the part about theory skills not being as important outside academia. Put aside the glamorous conceptual stuff and think about practical day-to-day skills needed to be a theorist and you'd find it is not much different from many other jobs in banking, industry, etc. In fact, that is probably the best reason to have a dedicated course for computational methods, this set of skills is even more valuable for those not going to grad school.

My "explained in the comments course" would be

1) the obligatory math methods course (mentioned above)
2) the obligatory computation methods course (mentioned above).

The number of grad students who, in this day and age, still cannot program their way out of a wet paper sack is frightening.

An introductory course in history and philosophy of science is just as important in the "know what you're doing"-sense as lab courses. I'm willing to wager it would save both students and instructors a lot of grief. (And any instructor wanting to disprove Heisenberg can sit in.)

For the GR course, we used Wald in 4th year, but I didn't really get it until my grad work. With the hindsight of another 15 years, I'd suggest something like Hartle instead.

Also, I agree on advanced PDEs and ODEs courses, but there are several other courses that I think are very useful for physics students and still appreciated at the undergrad level:
(1) nonlinear methods
(2) at least one theoretical continuum or fluids course at 3rd or 4th year
(3) at least one functional analysis and/or real analysis course by 4th year, or grad school is going to be too much of a step. This is the bid for more math in a physics program I know, but without those two courses, you tend to be shooting yourself in the foot with a really big gun for grad school
(4) some numerical methods courses and/or programming

Cheers,
Lee.

I'm with Chad on the lab classes. Data rules! The only way to get data is to measure it. I might be biased: in my final year of a four year physics degree, I was assigned the task of using a Simpson double ionization chamber to measure the output of a Ph.D. sudent's UV source. First I had to build (lathe, gas torch, solder, wrenches etc.) the vacuum chamber ...

i would add special relativity as a necessity. GR maybe not so much.

I disagree with the original article by Ives about E+M, because I think you can teach the concepts of Gauss and Ampere without tons of math and I think you need to start trying to grasp those concepts before you take a class that really uses the math.

I would combine math and computational methods as a pre-req for upper level classes, but the key here is to settle on departmental undergrad software package just like you settle on common mathematical methods. Then every mechanics, quantum, and E+M class uses the same math package with the expectation that numerical sims can be done from the first day of class. The choice is less important than that it be made.

As a theorist, I think advanced lab is crucial for everyone going to grad school and certainly everyone who isn't.

Although I vote for a full year of QM and E+M, that really isn't for everyone. You probably need something like a pre-grad school track that definitely has those plus an industry track that substitutes some applied classes in there.

Flaws in modern physics, and there are many, are not an excuse for dropping it -- they are a reason for fixing it.

Thanks for the book recommendations. I've been trying to improve my own schooling, and I was very pleased to discover Griffith's Electrodynamics. Very easy to read: it went over the basics of statics that I knew from my introductory course a decade ago, and then did a thorough explanation of dynamics. The introduction with SR, 4-vectors and tensors was good, too.

I'm looking forward to getting his QM book now. I was taught Gashiorowitz, and that was hard going. A bit easier on my second read last year, though.

I picked up Zee for QFT, but it's too difficult for me. I think I've learned the basics of GR now from Misner, Thorne and Wheeler, but it's still something I need to work harder on. I'm considering buying Carroll as well.

Suggested source: Math methods, e.g. Arfken; or for old-school, Mathews and Walker; Courant and Hilbert; etc. With a second semester option for a toned-down version of topological methods (e.g., an undergraduate version of Nakahara)

I think a year of Griffiths-level quantum is absolutely critical not only because of the material covered but because of the required practice with types of math that show up absolutely everywhere in technical fields. By the time I went to grad school I didn't think of it as a Fourier transform, I thought of it as an integral transform with a Fourier kernel (instead of Mellin, or Laplace, etc). This sort of generalization ability is critical, and I can't think of too many places where I had a chance to work on it during my academic career.

The software part is critical as well, both programming and the ability to adapt to new, complex software packages that you are provided, which in a corporate environment can have features added/removed/changed on a weekly basis when moving to a new version. In my job I feel it's harder to keep up with which particular version is the latest than it is to use the software, which I credit entirely to my time spent in the physics world.

CHECK THIS BOX (for renormalization purposes)