Three-Toed Sloth

Books to Read While the Algae Grow in Your Fur, May 2014

Attention conservation notice: I have no taste.

Robert Hughes, Rome: A Cultural, Visual, and Personal History

As the subtitle suggests, a bit of a grab-bag of Hughes talking about Roman, or Rome-related, subjects, seemingly as they caught his attention. Thus the chapter on ancient Rome contains a mix of recitals of the legends, archaeological findings, the military history of the Punic Wars (including a description of the corvus filled with what I recognize as school-boy enthusiasm), the rise of the Caesars — and then he gets to the art, especially the architecture, of the Augustan age, and takes off, before wandering back into political history (Diocletian--Constantine--Julian). The reader should, in other words, be prepared for a ramble.

Hughes is, unsurprisingly, at his best when talking about art. There is he knowledgeable, clear, sympathetic to a wide range of art but definitely unafraid of rendering judgment. If he doesn't always persuade (I remain completely immune to the charms of Baroque painting and sculpture), he definitely does his best to catalyze an appreciative reaction to the art in his reader, and one can hardly ask more of a critic

He's at his second best in the "personal" parts, conveying his impressions of Rome as he first found it in the early 1960s, and as he left it in the 2000s, to the detriment of the latter. (He's self-aware enough to reflect that some of that is the difference between being a young and an old man.) His ventures into the political and religious history of Rome are not as good — he has nothing new to say — but not bad.

Overall: no masterpiece, but always at least pleasant, and often informative and energizing.

R. A. Scotti, Basilica: The Splendor and the Scandal: Building St. Peter's

Mind candy: engaging-enough popular history, by a very-obviously Catholic writer. (My own first reaction to St. Peter's, on seeing it again for the first time as a grown-up, was that Luther had a point; my second and more charitable reaction was that there was an awesome space beneath the idolatrous and servile rubbish.)

Pacific Rim

Mind candy. While I like giant robots battling giant monsters, and I appreciate playing with elements of the epic (the warrior sulking in his tent; the catalog of ships), I'd have liked it better if the plot made more sense.

Sara Poole, Poisoner and The Borgia Betrayal

Mind candy: decent historical thrillers, though implausibly proto-feminist, philo-Semitic and proto-Enlightenment for the period.

Patrizia Castiglione, Massimo Falcioni, Annick Lesne and Angelo Vulpiani, Chaos and Coarse Graining in Statistical Mechanics

A good modern tour of key issues in what might be called the "Boltzmannian" tradition of work on the foundations of statistical mechanics, emphasizing the importance of understanding what happens in single, very large mechanical assemblages. Both "single" and "very large" here are important, and important by way of contrasts.

The emphasis on the dynamics of single assemblages contrasts with approaches (descending from Gibbs and from Jaynes) emphasizing "ensembles", or probability distributions over assemblages. (Ensembles are still used here, but as ways of simplifying calculations, not fundamental objects.) The entropy one wants to show (usually) grows over time is the Boltzmann entropy of the macrostate, not the Gibbs or Shannon entropy of the ensemble. (Thus studies of the dynamics of ensembles are, pace, e.g., Mackey, irrelevant to this question, whatever their other merits.) One wants to know that a typical microscopic trajectory will (usually) move the assemblage from a low-entropy (low-volume) macrostate to a high-volume macrostate, and moreover that once in the latter region, most trajectories that originated from the low-entropy macrostate will act like ones that began in the equilibrium macrostate. One wants, though I don't recall that they put it this way, a Markov property at the macroscopic level.

Randomizing behavior for macroscopic variables seems to require some amount of instability at the microscopic level, but not necessarily the very strict form of instability, of sensitive dependence on initial conditions, which we've come to call "chaos". Castiglione et al. present a very nice review of the definitions of chaos, the usual measures of chaos (Lyapunov exponents and Kolmogorov-Sinai entropy rate), and "finite-size" or non-asymptotic analogs, in the course of arguing that microscopic chaos is neither necessary nor sufficient for the applicability of statistical mechanics.

The single-assemblage viewpoint on statistical mechanics has often emphasized ergodicity, but Castiglione et al. down-play it. The ergodic property, as that has come to be understood in dynamical systems theory, is both too weak and too strong to really be useful. It's too weak because it doesn't say anything about how quickly time averages converge on expectations. (If it's too slow, it's irrelevant to short-lived creatures like us, but if it's too fast, we should never be able to observe non-equilibrium behavior!) It's too strong in that it applies to all integrable functions of the microscopic state, not just physically relevant ones.

The focus on large assemblages contrasts with low-dimensional dynamical systems. Here the authors closely follow the pioneering work of Khinchin, pointing out that if one has a non-interacting assemblage of particles and considers macroscopic observables which add up over molecules (e.g., total energy), one can prove that they are very close to their expectation values with very high probability. (This is a concentration of measure result, though the authors do not draw connections to that literature.) This still holds even when one relaxes non-interaction to weak, short-range interaction, and from strictly additive observables to ones where each microscopic degree of freedom has only a small influence on the total. (Again, familiar ground for concentration of measure.) This is a distinctly high-dimensional phenomenon, not found in low-dimensional systems even if very chaotic.

Putting these two ingredients — some form of randomizing local instability and high-dimensional concentration of measure — it becomes reasonable to think that something like statistical mechanics, in the microcanonical ensemble, will work. Moreover, for special systems one can actually rigorously prove results like Boltzmann's H-theorem in suitable asymptotic limits. An interesting large-deviations argument by De Roeck et al. suggests that there will generally be an H-theorem when (i) the macroscopic variables evolve autonomously in the large-assemblage limit, and (ii) microscopic phase-space volume is conserved. This conclusion is very congenial to the perspective of this book, but unfortunately the work of De Roeck et al. is not discussed.

One feature which pushes this book beyond being just a careful and judicious defense of the Boltzmannian viewpoint on statistical mechanics is its treatment of how, generally, one might obtain macroscopic dynamics from microscopic physics. This begins with an interesting discussion of multi-scale methods for differential equations, as an alternative to the usual series expansions of perturbation theory. This is then used to give a new-to-me perspective on renormalization, and why differences in microscopic dynamics wash out when it comes to aggregated, macroscopic variables. I found this material intriguing, but not as fully persuasive as the earlier parts.

Conor Fitzgerald, The Dogs of Rome

Mind candy. Police procedural with local color for Rome. (Having an American protagonist seems like a cheat.)

Colin Crouch, Making Capitalism Fit for Society

A plea for an "assertive" rather than a "defensive" social democracy, on the grounds that social democracy has nothing to be defensive about, and that the taming of capitalism is urgently necessary. That neo-liberalism has proved to be a combination of a sham and a disaster, I agree; that lots of his policy proposals, about combining a stronger safety net with more microeconomic flexibility, are both desirable and possible, I agree. But what he seems to skip over, completely, is where the power for this assertion will come from.

Disclaimer: I've never met Prof. Crouch, but he's a friend of a friend.

Elliott Kay, Poor Man's Fight

Mind candy science fiction. Does a good job of presenting the villains sympathetically, from inside their own minds. Also, props for having the hero's step-mother make a prediction of how joining the navy will distort the hero's character, for it coming true, and for the hero realizing it, to his great discomfort. — Sequel.

Books to Read While the Algae Grow in Your Fur; Pleasures of Detection, Portraits of Crime; Scientifiction and Fantastica; Writing for Antiquity; Tales of Our Ancestors; Physics; The Progressive Forces