A test of pre-requisite knowledge and skills

Today I gave my PHYS4030 Condensed Matter Physics students a Pre-test, a 50 minute exam to test whether they have some of the background knowledge and skills I consider necessary for the course.

I am pretty happy with the test I came up with as it tested some basic skills as well as knowledge. But I am sure it can be improved.
I welcome comments on it and examples of alternatives.

I marked the exams and gave them back to the students. I did not keep a record of the marks and they don't count at all for the final grade.
In case some students felt they did poorly they can take the exam home, do some review, and then submit a new set of answers I will mark.

Most of the marks were in the 50-85% range, which suggests to me the test was set at about the right level. Although, of course, I really wish they were all scoring above 80%.

A few random thoughts:

1. I think this exercise also has value as a bit of a reality check and wake-up call at the beginning of the semester. It reminds students that they are going to have to do an exam in the subject. They are also going to need a lot of stuff they may have forgotten or should have mastered in the past. Normally, some students don't get the wake up call until the mid-semester exam.

2. It helps students get familiar [in a non-threatening context] with the kind of exams I set and how I mark them.

3. As I have posted before I still find it disturbing that there are final year undergrads who struggle with basic things such as

• working with physical units
• sketching graphs
• basic calculus and analysis
• interpreting a graph of experimental data 

4. If students fail the test the message should be clear: either drop the course or expect to do an extra-ordinary amount of work to catch up. Otherwise, you are going to fail. Don't say you were not warned.

Why is condensed matter physics so hard for undergraduates?

The new semester has started here and I am helping teaching PHYS4030 Condensed Matter Physics. We basically cover some fraction of Ashcroft and Mermin. For some students, particularly those with weak backgrounds, this is a difficult course.
Some find it much harder than other courses. Why?
I can think of a few reasons of why CMP can be more demanding than other courses.

It requires a working knowledge of basic thermodynamics, statistical mechanics, kinetic theory, electromagnetism, and quantum mechanics. Weak students do o.k. in some subjects but poorly in others. Weak students do not remember much from previous courses and struggle to apply what they do learn in new contexts. Hence, CMP really exposes some of these weaknesses.
Furthermore, one has to understand how to integrate all this knowledge.

There is an emphasis on

• orders of magnitude estimates 
• making approximations
• relating theory and experiment

On the other hand, for these same reasons motivated and well prepared students can find a condensed matter course very stimulating and interesting. Such a course also provides some skills (model building and testing, synthesis, estimates) which are useful in much broader contexts.

I welcome alternative thoughts.

What is the ground state of solid hydrogen?

The Journal of Chemical Physics website has a fascinating podcast with Roald Hoffmann, Neil Ashcroft, and Vanessa Labet talking about a series of 4 papers they have just published about "molecular" hydrogen under pressure. They illustrate some very rich and subtle physics and chemistry.

It highlights the importance of both physical and chemical insight, simple models, and how there are still these old problems waiting to be solved.

Questionable paper titles

There is an interesting article by Ben Goldacre in the Guardian newspaper which nicely summarises research on the following questions:

Will asking a question in the title get your paper cited more?
No. But it will be downloaded more!

What is the evidence that having your paper mentioned in the New York Times will increase  its citation rate?

Overdoped cuprates are an anisotropic marginal Fermi liquid II

Jure Kokalj, Nigel Hussey, and I have just completed a paper, Transport properties of the metallic state of overdoped cuprate superconductors from an anisotropic marginal Fermi liquid model.

We show how a relatively simple model self-energy [considered earlier in this PRL] gives a nice quantitative description of a wide range of experimental results on Tl2201 including intra-layer resistivity, frequency-dependent conductivity, and the Hall resistance. No new parameters are introduced beyond those needed to describe angle-dependent magnetoresistance experiments from Nigel's group.

One thing I found striking was just how sensitive the Hall conductivity is to anisotropies in the Fermi surface and the scattering rate [a point emphasized by Ong with his beautiful geometric interpretation].

We also show that our model self-energy successfully describes both the resistivity (with a significant linear in temperature T dependence) and the Hall angle ( ~T^2) without invoking exotic new theories.

A key outstanding challenge is to connect our model self-energy [which is valid in the overdoped region] to possible forms for the underdoped region where the pseudogap occurs.

We welcome comments.

Strongly correlated electron systems in high magnetic fields IV

The observed sensitivity of strongly correlated metals to laboratory magnetic fields of the order of 5-50 Tesla presents a significant theoretical puzzle and challenge. There have been very few calculations on lattice models such as the Hubbard model in a magnetic field. The few calculations that have been done only see very small perturbative effects on the scale of laboratory fields. They require huge magnetic fields of the order of a thousand Tesla for any significant effect, such as a change in ground state.

In terms of coupling of the magnetic field to the orbital degrees of freedom, most studies have been on ladder models (e.g. this PRB), at zero temperature, and only see significant effects at extremely high fields, of the order of thousands of Tesla, when there is a quantum of magnetic flux through a single lattice plaquette. The smaller field scale of the upper critical field for superconductivity corresponds to the longer length scale of a superconducting coherence length. This longer scale may only be accessible on sufficiently large square lattices.

How many metrics do you need?

Different metrics claiming to measure research impact, such as the h-index, are receiving increasing prominence in grant and job applications. I have written before that I think they have some merit as a blunt instrument to filter applications, particularly for people at later career stages.
However, I am noticing an increasingly silly trend to cite a whole range of metrics, where I think one or two  (probably the h-index and m-index=h-index/no. of years since Ph.D) would suffice. I have seen not just a paragraph, but a whole page! of analysis citing all sorts of metrics [e.g. comparing an authors citation rate for a particular journal to the impact journal factor, no's of papers with more than 50 citations, citation rate relative to others in the field, .... the list goes on and on...]. Don't people have better things to do with their time?

In the end it becomes like university rankings. Every university seems to cite the one in which they rank the highest.