Electron Band Structure In Germanium, My Ass | Hacker News Reader

Electron Band Structure In Germanium, My Ass

pages.cs.wisc.edu

316 pointsbootload97 comments

Hah, this exactly mirrors my experience in chemistry lessons aged 15. Failing to reproduce experimental results with broken equipment then faking the data with an excel function to get the teachers off my back so I could go back to doing something useful like staring out of the window, or using the magnesium ribbon to heat-seal peoples pencil cases shut.

Good to know that it carries on up to the undergraduate level. And I even managed to associate with women during my eventual CS degree! Feeling pretty smug right now.

Failing to reproduce experimental results with broken equipment then faking the data with an excel function to get the teachers off my back...

I had the exact opposite experience. I once failed to get an experiment working, but kept trying until I was the last person in the lab. Eventually the professor asked why I hadn't finished, and together we discovered the spectrometer was utterly broken.

He gave a a grade of 0-25% to everyone who got the "right" answer, depending on how realistically they faked it (some people didn't bother to add noise or quantize their answers).

Your own example suggests you're in the (vast) minority.

Eh, in my experience, despite the fact that points were not deducted for getting the "wrong" result, and they were frequently told so, students still preferred to falsify data.

I think it is because if they get the "correct" result they can also just copy an analysis from elsewhere, whereas with the "wrong" result they would have to actually do it themselves.

That's so funny because I always loved getting the answer wrong. I could get full credit for explaining possible sources of error, versus having to develop a well-established analysis independently. You'd think more students would think this way.

School isn't about learning, it's about baby sitting and imposing respect for authority and ability to do make work. If you're lucky you'll get an education as a side effect. For myself I went to some of the best public schools in the country and the education I got was generally only comparable to what I could have gotten from self-study with the exception of perhaps one or two excellent teachers.

My EE professor in Dynamic Fields, halfway through a 3 hour lab, announced to the class "Oh, by the way, anyone who gets a right answer gets an F for the lab." Groans from half the class. "The point of the lab is to show how danged difficult it is to get to the right answer".

I looked at my mess of data and chuckled.

It carries on through undergrad and into post-grad. Even the best funded labs still have mountains of crappy equipment that everyone has to use, and one researcher who keeps everything that works under lock and key.

On the plus side, it's a great way to learn to be really ingenious with limited resources. We figured out how to get accurate hysteresis loops for ferrous samples using a breadboard, an ancient laptop, a few clicks of Cu wire and a bucket of miscellaneous components. That was shortly after we flooded the lab testing Reynold's number for shear-resistant fluids with plumbing parts.

I had this insight in Junior High - testing friction by putting blocks on top of other blocks, or in a block train, and using a 'force meter' to see how hard I had to pull. The 'force meter' was a piece of spring steel stuck into another block, with a hook on the end.

It was totally non-linear on every surface I tested. The book said it was supposed to be linear. The students were all furtively fudging it, and eventually the teacher said something like "well, its supposed to be linear so do the best you can".

Insight: this was all a bunch of crap. Turns out that friction is totally non-linear anyway, for most materials, but I didn't read that until 20 years later.

That reminds me of a story about Richard Feynman visiting a physics class outside the US. The diplomat types were very upset with him for pointing out that their books and classes were terrible, teaching things like the energy of a ball going down an incline without taking rotational kinetic energy into account.

http://v.cx/2010/04/feynman-brazil-education

Yes, thank you. That's exactly the one I was thinking of.

He's so totally right about the hand waving approximations used in solid state physics.

All fields of physics need to use some approximations, but those in the solid state were usually the ones with the worst reasons, just things like "it works if we do that" or even "it doesn't if we don't do this".

It went on like that in three courses (two on solid state physics, one on electronics).

Only after that did I happen to come across a decent book which explained some of the approximations in a way that didn't make me cringe, and some of the stuff started to make sense in retrospect. Others are still a mystery to me, and probably always will be.

It's often poorly explained, I completely agree, but people forget that even Planck introduced his constant just because it made the data fit. It turned out to be arguably the most fundamentally important constant (along with its reduced form) in all of physics.

Aren't all fundamental physical constants like that - they are essentially meaningless values plugged into theories to make them align with observations?

yeahbut, some are interrelated. Like the speed of light and the permittivity of freespace (1/c^2 * mu). If you measure that constant and solve back for c, then you better get ~3 * 10^8 m/s.

Mostly because we (humans) made up physical units (i.e. meters) while we were inventing physics. There's nothing keeping you from re-deriving physics equations without the messy constants.

http://en.wikipedia.org/wiki/Natural_units

Even then there are some dimensionless numbers that you can't get rid of (the ratio between the strength of the gravitational and coulomb forces being one obvious example).

Pi being another.

I wouldn't have said Pi was a physical constant - it's not like we actually measure circles to derive it empirically.

True, but it certainly is a dimensionless number that you can't get rid of (and that actually shows up in physical formulae, such as http://en.wikipedia.org/wiki/Coulomb%27s_law). Come to think of it, so is 2.

I guess if you tried to measure π by constructing circles, you'd actually be measuring the curvature tensor of space (which is an experimental observation), not π (which is just π).

π being, simply, the measurement of the curvature tensor of an ideal Euclidean plane. If we lived in a universe that was non-Euclidean at macro-scale, π would be just another irrational number, and some other quantity would be exalted as "fundamental."

Pi would still be fundamental in mathematics whatever the geometry of our universe. (Examples of where it would turn up: consider the differential equation f''=-f; all its solutions are periodic with period 2pi. The series 1-1/3+1/5-1/7... has sum pi/4. The series 1+1/4+1/9+1/16+... has sum pi^2/6. exp(pi sqrt(163)) is ridiculously close to being an integer. There are deep reasons for all these things, and they wouldn't go away if the universe were very far from spatially flat.)

Natural units are nice, but they are completely impractical for dealing with normal Mirkowskian space. I.e., how do I explain to someone who cannot view me how tall I am? Our normal unit system is nice because it's phenomenological, and maybe inconvenient.

Agreed, but the universe is under no obligation to express itself in units that are convenient on a human scale. I like to think of this as an anti-anthropomorphic principle.

I up-voted this comment, because I liked it, but honestly, I first looked for the "like" button. Facebook's got a serious hold on me.

That's what they look like at present, but it's possible that future theories will allow us to calculate their values from first principles.

Do you remember the title or author of that book?

I don't necessarily know what book he's referring to, but the book that taught me solid state was Ashcroft and Mermin.

That's a book that deliberately starts out with the silliest possible model and then gradually introduces more and more sophisticated models until the book runs out of pages or your head explodes, whichever comes first. ;)

An important insight, however, is that the simple models are incredibly useful. When I was a teenager I used to fret that the teachers were showing me the simple models first because they were fools, or because they thought we were fools. But it turns out that they are doing so to call attention to certain important general features without distracting you with irrelevant detail.

My favorite class in all of science was Roald Hoffman's chemistry class, the highest numbered class in the Cornell chemistry department, a class which represented the point where chemistry and physics merge to become the same subject. And Hoffman deliberately used a relatively simple model of molecule-molecule interactions in a solid, the Hückel model:

http://en.wikipedia.org/wiki/Hückel_method

Moreover, Hoffman would communicate in pictures as much as in math.

Have you ever read Feynman's Lectures on Physics Volume 2 (Electricity and Magnetism)? One of the things I love about it is that it starts with Maxwell's equations, in their non-simplified form, then goes into the special cases of electrostatics, magnetostatics, and then uniting them into electrodynamics. Feynman goes over the special cases in a way that you never forget that they're in the end incorrect simplifications, there's even a table at the end of the statics section showing "These equations are false, these are true in general."

I don't mind simplified incorrect models so long as they're presented that way (so I don't want to read a "all that stuff we just covered? Yeah it's wrong"-ish sentence after the fact) and the full truth is eventually revealed in an understandable way. I want a full picture of something, not a partial incorrect picture of something, it scares me that some people never stop thinking of an atom as a small planetary system.

Amen to Ashcroft & Mermin. I found it much easier to use over Kittels solid state text. Or, more accurately, Shittels solid state physics

Indeed, the one solid-state class I took used Kittels, and I quickly ended up borrowing Ashcroft from someone. Kittels was the least "physics-y" physics textbook I ever came across.

"Electronic Transport in Mesoscopic Systems" by Supriyo Datta.

It's a very specific book, but it does explain for example why you can assume that only electrons at the Fermi energy contribute to transport phenomena (aka "current").

I guess for somebody not into mesoscopic transport, only the first three or four chapters will be of interested. But for my diploma thesis (which ended too badly to link it here) it was a real live safer.

If anyone needs a refresher in semiconductor physics, there's a good one at this Britney Spears site: http://britneyspears.ac/

edit: how the heck does a link to an excellent introduction to semiconductor physics get voted down in a thread about an article relating to semiconductor physics?

You're getting voted down because most people see the domain "britneyspears.ac" and assume you're spamming, without even visiting the link.

That site actually does have some decent articles related to semiconductor physics, the whole "Britney Spears" thing is just a gag to show how simple they've made it (though the humor is likely to be appreciated only by Physics geeks :)

Sweet jesus, it's true! http://britneyspears.ac/physics/vcsels/vcsels.htm

While there does appear to be some physics articles on the site, the whole thing appears to be cloaking itself in legitimacy as a scheme to get inbound links and drive up its page rank score. Their strategy is likely to get people linking for the physics so that they can sell links for heating oil prices, concert tickets, physics help, and so on. It's more about the advertising than the physics.

I'm sure there's a less spammy physics site you could have linked to?

On the contrary, the page is a piece of internet history. As you can see from the domain, it was created by an academic in an early example of titlebait.

http://replay.web.archive.org/20001002213854/http://www.brit...

Its basically a reverse rickroll. "Britney Spears" was by far the most searched-for term on the internet for a long time.

.ac is the country code for Ascension Island and domains on that TLD are required to be academic in the same way that Colombia's .co domains are required to be companies (in other words, there is no requirement). There's a page on how to advertise on the site and a merchandise page as well. Sorry, but I'm still not seeing how this wasn't built as primarily a revenue generator.

Hmmm... Well in that case for more evidence I am going to have to fall back on the assertion that I remember hearing a reference to it being set up in the 90s by a physics professor as a joke, in a documentary about memes on BBC Radio 4:

http://www.bbc.co.uk/programmes/b00zlk03

Although you can't stream that programme and I appreciate that my half-recollections are hardly convincing.

What does seem likely is that the site was in fact set up in the 90s by a physics professor as a joke, and since then it has been used to make money, precisely because it was popular and unusual at the time. Maybe he sold it to someone back in the good old 90s bubble, when men were men and a domain name was a business model.

(Fun Fact: That documentary was presented by the woman who wrote the article about LSD that was on here recently.)

Here's a BBC text article about the site: http://news.bbc.co.uk/2/hi/technology/1306364.stm

Kudos to the postgraduate student for getting all that PR coverage. He could be raking in a few hundred to a few thousand dollars per month depending on advertising rates and Britney merchandise sales.

This is a good joke, but too subtle for most of the people who moderate on HN.

My advice if you care about that sort of thing: link in further, e.g., http://britneyspears.ac/lasers.htm

This "lab report" always brings back good memories because it indirectly launched my career in the game industry. I was finishing up my CS degree at UW Madison and working in Mike Gleicher's computer graphics lab in Spring '02. I had previously met Lucas through a fellow CS student (Alex Mohr, now at Pixar). At some point, Lucas was contacted by the AI programmer at Ensemble Studios (Mike Kidd). Mike, a UW alum, had seen the Germanium treatise and wanted Lucas to apply at ES. A month later, knowing how eager I was to join the game industry, Lucas mentioned Mike's email to me and forwarded him my info. Long story short, I got an interview at ES and was hired straight out of college into a dream job.

I look forward to seeing this link pop up again in a few years. :)

I sympathize with the horrible equipment. While getting a physics degree, I had a lab where we measured resistivity. My team was the last to succeed in getting numerical data from the 20-year-old oscilloscope via floppy disk. The students scheduled after us had to take a digital photograph of the oscilloscope screen and reconstruct the data from that.

Broken equipment is not confined to undergraduate classes. A few terms later, I had to use the digital photograph method to get data from a spectrometer in a research lab. If my analysis had succeeded, we would likely have published, and the professor would have had a chance at tenure, all on the strength of data obtained by counting pixels.

Oh, lovely story. I had precisely this happen in Physics at Uni. I wrote down every piece of data, wrote out my computations in full. When I submitted it, I got a D. I challenged my prof to check any one of the data points or computations and exhibit a single error (I knew he wouldn't because I had checked myself). He refused to check even a single one. Every single person for 20 years had fudged the results!!

I read an essay, I think by Isaac Asimov, where he described a graph like this as a "shotgun curve"

Yeah, I remember a number of labs where I didn't get the data that was expected and had to do the experiment again. Knowing the expected result, producing excellent data was easy; instead of a painstaking data-collection process, I outsourced that to a "function" with "noise". Hey look, r=0.999!

I understand this is the same process by which the top quark was discovered.

Hey look, r=0.999!

I had to grade physics labs one semester -- this was always an instant tell that students were making shit up. :)

always an instant tell that students were making shit up

Including the correlation coefficient!

I wish I'd even once had students who understood the experiment well enough to fabricate data.

"Going into physics was the biggest mistake of my life. I should've declared CS. I still wouldn't have any women, but at least I'd be rolling in cash."

This advice should be told to all high schoolers.

Well, at least male ones.

Or do both! I always cringe when I meet good programmers that have no understanding of what is going on at the hardware level.

Because knowing the hardware is, of course, essential to make a well-functioning CRUD app.

A physics degree can get you a mid-six-figure job as a quant on Wall Street.

What's a 'mid-six-figure'? 150000? 500000?

Does "mid-six-figure" mean $550k (midpoint of the six-figure range), $300k (ditto but using the geometric mean), or $150k (midpoint of the $100k-$200k range)?

(My impression is that $150k is a plausible entry-level quant salary, but it seems a bit strange to describe that as "mid-six-figure".)

The "x-figure" notation is essentially logarithmic with an offset: "6-figures" = 10^5. Therefore, "mid-six-figures" should be roughly 10^5.5 or $316227.

I believe that's the geometric mean he's referring to.

  (a*b)^1/2 # geometric mean, square root of the product of the two numbers
  (10^5 * 10^6)^(1/2) # substitute in for 10^5 and 10^6, the limits of the six-figure range
  (10^(5+6))^(1/2) # simplify inner product
  (10^11)^(1/2) # more simplify
  10^(11 * 1/2) # move the 1/2 in, and ...
  10^(11/2) # simplify
  10^5.5 # and finally you get the same result

The geometric mean always struct me as an interesting idea in mathematics

Curious: how could $150k be mid six-figure? I can accept the roughly 500K or 300K meanings, but if someone told me mid six figure, and they meant roughly $150K, I'd feel like they deliberately mislead or lied about the salary. There's no way I can see a reasonable person expecting $150K when someone is referring to "mid six-figure".

But the people writing the CRUD apps get that too.

If the goal of a CS course is to teach how to make a well-functioning CRUD app, I despair!

CRUD apps are bread and butter, essential, but unhealthy foods.

no way butter is unhealthy, I can't tell about bread though.

If the goal of a CS course is to in any way make you reliant on hardware or even physics, I despair.

What use is physics when you have an infinite tape?

If you're a programmer and your program relies on the premise that you have infinite storage then your program is probably not going to work too well. Presumably you're assuming zero seek time too?

Wasn't that the GP's point that to make the transition from CS to programming some knowledge of physical realities of computers might be useful??

Programming isn't strictly computer science. Programming is programming. Computer science is a theoretical discipline.

A couple of the best programmers I've ever met majored in physics.

While I would like to think you were talking about myself, I'm guessing you're thinking of people more along the lines of Dennis Ritchie.

Go into chemistry, instead. Lots of nice women. [Edit: It is also a cool subject. And you'll get access to lots of ethanol.]

Aren't they brutal to the grad students working as research assistants in the lab?

Where are they not brutal to grad students? (Depends on which country we talk about, afaik.)

In computer science, you will sometimes get to choose which 20 hours of the day you want to spend in the lab. Luxury!

Physics, surprisingly enough.

Which country is this? Surely not the US.

Looks like I did well: studied physics and chemistry, got MS in physics and astronomy. Dropped out of PhD in astronomy — was already working as a programmer. To tell the truth I still feel sorry for not getting that PhD, but on the other hand I was not going to be an astronomer anyway, so what's the point.

I don't think it would have increased how many women you get by all that much either ;-)

There really is no reason to earn an undergraduate degree in physics (as I learned the hard way). It is just as easy to get into a physics graduate program with an ECE or EECS degree plus a couple semesters of upper division quantum mechanics and maybe a semester of stat mech - and you have the option of getting a decent job instead.

Yea perfect case in point. I graduate this Friday with a physics undergrad. Class of 33, 31 males. 2 females :(

This is why the only lab I enjoyed was in Materials Science. The goal of the labs was never to try to get the data to match some pre-defined equation. Instead, the goal was to test materials and interpret the results. Plus, we got to break stuff. :)

As a materials science major, this comment made me smile. All my labs at Carnegie Mellon were like that, it was(is) great. We just take a bunch of materials and then argue which one would be the best choice. The only time I had to produce results was when we synthesized a known superconducting structure and just had to show that it superconducted (how strongly wasn't important). Good times that sound much better than people's experiences in physics and chemistry. Vindication!

Funny as hell when I was a physics undergraduate 10 years ago, and still funny now. I still routinely use the phrase "to first order" when justifying horrific approximations.

The crap results were probably due to the soldering messing around with the crystal properties. Using a pressure-contacts approach instead seemed to do the trick:

http://tinyrock.com/resources/4000918/Bandgap-Measurements-I...

Has it really been 10 years!

I see the test results for germanium but where are the results for his ass?

I don't understand a wee bit of that, so Go Computer Science! :)

He's unit-testing Germanium.resistivity(intensity, temperature) like so:

    fixed_i = lambda t: Germanium.resistivity(42, t)
    for t in range(100, 350, step=10):
        assertEqual(fixed_I(t), k*math.expr(t/tref))

Unfortunately fixed_i returns results akin to /dev/urandom because of broken hardware.

Awesome explanation! Thanks.

That is an absolutely fantastic summary. I upvoted the parent just to get this one more attention :D

Good questions deserve an upvote just as much as getting answered :)

If only high-school Chemistry was this easy, I would go back and do it again, just to be fascinated by nature.

This pops up every few years. I always forget it and laugh out loud brand new when re-reading it.

An oldie, but a goodie. Easily my favorite scientific report paper of all time.

>Banking on my hopes that whoever grades this will just look at the pictures, I drew an exponential through my noise. I believe the apparent legitimacy is enhanced by the fact that I used a complicated computer program to make the fit. I understand this is the same process by which the top quark was discovered.

Absolutely brilliant.

The way I have always approached labs is a thorough report on my procedure, complete results (totally wrong or not) and an analysis of Whether I got good results, Why I got the wrong results, How might one improve the experiment to get better result...

That has always served me well. I don't usually get A+'s on lab reports, but I think that's my fault and for an unrelated reason.

It certainly seems to me reflection on what was done and thoughtful analysis on why it was good/how it could be fixed (with specific statements or suggestions, not general ones)/what went wrong demonstrates understanding of the material that simple results do not. I guess if you are being trained primarily to be a lab worker who's job is to produce accurate numbers, that's important, but if the labs are for learning...

Hilarious! First time I've seen this and I laughed so hard I had tears running down my face. I think it was figure 1 "check this shit out" that made me really lose it. Got some odd looks on the train.

During my physics practicals for my high school final exam, I had to measure the voltage of a Daniel Cell using a Dry Cell (1.5V), a 2metre length of wire and a bunch of other stuff.

The only problem... the dry cell was dead, so only carried 0.8V or something. So, my results consistently showed that the Daniel Cell had a PD of 1.7V (it's supposed to be ~1.1V).

The examiner came up with a brilliant solution to "fix" my data. Switch the labels on the data columns.

All scientific results should be reported this bluntly.

Ah, physics, the slower cousin of mathematics. How I weep for you.

Reply on news.ycombinator.com