Friday, December 31, 2021

A book review, and wishes for a happy new year

 I was fortunate enough to receive a copy of Andy Zangwill's recent biography of Phil Anderson, A Mind Over Matter:  Philip Anderson and the Physics of the Very Many.  It's a great book that I would recommend to any physics student or graduate interested in learning about one of the great scientists of the 20th century.  Zangwill does an excellent job with the difficult task of describing (in a way accessible to scientists, if not necessarily always the lay-public) the rise of solid-state physics in the last century and its transformation, with significant guidance from Anderson, into what we now call condensed matter.  This alone is reason to read the book - it's more accessible than the more formally historical (also excellent) Out of the Crystal Maze and a good pairing with Solid State Insurrection (which I discussed here).  

This history seamlessly provides context for the portrait of Anderson, a brilliant, intuitive theorist who prized profound, essential models over computational virtuosity, and who had a litany of achievements that is difficult to list in its entirety.  The person described in the book gibes perfectly with my limited direct interactions with him and the stories I heard from my thesis advisor and other Bell Labs folks.  Some lines ring particularly true (with all that says about the culture of our field):  "Anderson never took very long to decide if a physicist he had just met was worth his time and respect."

On a separate note:  Thanks for reading, and I wish you a very happy new year!  I hope that you and yours have a safe, healthy, and fulfilling 2022.

Monday, December 27, 2021

US News graduate program rankings - bear this in mind

The US News rankings of graduate programs have a surprisingly out-sized influence.  Prospective graduate students seem to pay a lot of attention to these, as do some administrators.  All ranking schemes have issues:  Trying to encapsulate something as complex and multi-variate as research across a whole field + course offerings + student life + etc. in a single number is inherently an oversimplification. 

The USNWR methodology is not secret - here is how they did their 2018 rankings.   As I wrote over a decade ago, it's a survey.  That's all.  No detailed metrics about publications or research impact or funding or awards or graduate rates or post-graduation employment.  It's purely a reputational survey of department chairs/heads and "deans, other administrators and/or senior faculty at schools and programs of Ph.D. Physics programs", to quote the email I received earlier this month.   (It would be nice to know who gets the emails besides chairs - greater transparency would be appreciated.)

This year for physics, they appear to have sent the survey to 188 departments (the ones in the US that granted PhDs in the last five years), and historically the response rate is about 22%.  This implies that the opinions of a distressingly small number of people are driving these rankings, which are going to have a non-perturbative effect on graduate recruiting (for example) for the next several years.   I wish people would keep that in mind when they look at these numbers as if they are holy writ.  

(You also have to be careful about rough analytics-based approaches.  If you ranked departments based purely on publications-per-faculty-member, for example, you would select for departments that are largely made up of particle physics experimentalists.  Also, as the NRC found out the last time they did their decadal rankings, the quality of data entry is incredibly important.)

My advice to students:  Don't place too much emphasis on any particular ranking scheme, and actually look closely at department and research group websites when considering programs.  

Saturday, December 18, 2021

No, a tardigrade was not meaningfully entangled with a qubit

This week this paper appeared on the arxiv, claiming to have entangled a tardigrade with a superconducting transmon qubit system.  My readers know that I very rarely call out a paper in a negative way here, because that's not the point of this blog, but this seems to be getting a lot of attention, including in Physics World and New Scientist.  I also don't know how seriously the authors were about this - it could be a tongue-in-cheek piece.  That said, it's important to point out that the authors did not entangle a tardigrade with a qubit in any meaningful sense.  This is not "quantum biology".

Tardigrades are amazingly robust.  We now have a demonstration that you can cool a tardigrade in high vacuum down to millikelvin temperatures, and if you are sufficiently gentle with the temperature and pressure changes, it is possible to revive the little creature.  

What the authors did here was put a tardigrade on top of the capacitive parts of one of two coupled transmon qubits.  The tardigrade is mostly (frozen) water, and here it acts like a dielectric, shifting the resonance frequency of the one qubit that it sat on.   (It is amazing deep down that one can approximate the response of all the polarizable bits of the tardigrade as a dielectric function, but the same could be said for any material.)

This is not entanglement in any meaningful sense. If it were, you could say by the same reasoning that the qubits are entangled with the macroscopic silicon chip substrate.  The tardigrade does not act as a single quantum object with a small number of degrees of freedom.  The dynamics of the tardigrade's internal degrees of freedom do not act to effectively decohere the qubit (which is what happens when a qubit is entangled with many dynamical degrees of freedom that are then traced over).  

Atoms and molecules in our bodies are constantly entangling at a quantum level with each other and with the environment around us.  Decoherence means that trying to look at these tiny constituents and see coherent quantum processes related to entanglement generally becomes hopeless on very short timescales.  People still argue over exactly how the classical world seems to emerge from this constant churning of entanglement - it is fascinating.  Just nothing to do with the present paper. 

Saturday, December 11, 2021

Real progress on machine learning for density functional theory

(Sorry about the slow pace of posting.  The end of the semester has been very intense, including a faculty retreat for our department last week.)

I've written before (here, here, and here) about density functional theory, arguably one of the most impactful intellectual physics results of 20th century physics.   DFT is one approach to trying to solve the quantum electronic structure problem for molecules or solids containing many electrons.  As explained in the links above, the idea is powerful.  It turns out that the ground state (lowest energy state) electronic density as a function of position \(n(\mathbf{r})\), contains all the information needed to calculate basically anything you could want to know about the ground state.  There is a functional \(E[n(\mathbf{r})]\), for example, that will give you the energy of the full-on, interacting many-electron ground state.  It's possible to do a non-interacting electron model that can get you arbitrarily close to the true, correct \(n(\mathbf{r})\), The tricky bit is, there is no exact analytical expression for the functional \(E[n(\mathbf{r})]\), which includes a particularly tricky contribution called the exchange-correlation part of the functional, \(E_{\mathrm{xc}}[n(\mathbf{r})]\).  Because we are talking about functionals rather than functions,  \(E_{\mathrm{xc}}[n(\mathbf{r})]\) might depend in a non-local way on \(n(\mathbf{r})\) and its derivatives at all points in space - there is no reason to think it will be simple to write down.  


I wrote six years ago about the idea that machine learning techniques might make it possible to get a working version of something close to the exact \(E_{\mathrm{xc}}[n(\mathbf{r})]\) , even if we can't readily write it down in some closed form.  Now it seems that real progress has been made in this direction.  Here is a blog post from the DeepMind team about their paper in Science this week where they demonstrate a new functional that they claim is very good and accurate vs exact calculations on test systems, computationally tractable, and satisfies fundamental properties that have to hold for the true exact functional.  They argue that their code is more than just a fancy look-up table and that it contains generalizable knowledge so that it's useful well beyond their specific training test cases.  

If this is so, then it could be a major step forward in (for some definitions of the term) first-principles calculations of molecular and material properties.  I'm curious about whether the new functional will actually let us gain some physical insight into why physics requires that particular underlying mathematical structure.  Still, even if we end up with a "black box" that allows greatly improved calculations, that would really be something.  I'd appreciate it if knowledgable DFT/electronic structure experts could comment here on how excited we should be about this.