In-vivo entropy production of A. subaru
Yu Fu, Emmy Dobson, Benjamin B. Machta, Michael C. Abbott · 2026
Read In-vivo entropy production of A. subaru by Yu Fu, Emmy Dobson, Benjamin B. Machta, Michael C. Abbott (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-00453.txt.
Introduction
Fu and colleagues measure entropy production in a living system whose true energy use is also known. Entropy production often stands in for the energy consumption of a system out of equilibrium, and lower bounds can be estimated from coarse observations. Applying these tools to A. subaru, they find the bound falls short of saturation by some 25 orders of magnitude. To be sure of the gap, they survey several irreversibility estimators and write down a new kNN estimator.
Catalog record
| CREATOR | Yu Fu, Emmy Dobson, Benjamin B. Machta, Michael C. Abbott |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21887 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.00453
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
In-vivo entropy production of A. subaru Yu Fu, Emmy Dobson, Benjamin B. Machta, Michael C. Abbott Abstract Entropy production is often used as a proxy for energy consumption of a non-equilibrium system. Lower bounds can be estimated from coarse-grained observations, and this has been done for various biological systems. Here, we apply these tools to a more macroscopic system whose true energy consumption is also known. We find that while entropy production does give a lower bound, it is some 25 orders of magnitude away from being saturated. To be certain of this result, we survey different methods of estimating irreversibility, and write down a novel kNN estimator. 1 Introduction Life is a struggle against the inevitable increase of entropy, enabled by the consumption of free energy from the sun, or more proximally by food, or by ATP arriving from elsewhere in the cell. A prominent signal of such energy consumption is irreversibility: a living process looks quite different if we play the movie backwards, unlike any system at equilibrium, for which we cannot tell. Irreversibility \,\Sigma is a statistical concept which we take to have units of bits per second. When calculated from coarse-grained observables, it provides a lower bound on the steady-state rate at which free energy must be consumed to drive the system, W in Joules per second [ 1 ] : k_{B}T\,\Sigma\leq W (1) where k_{B}=1.38\times 10^{-23}\, J/K, and T is the temperature. Yet not everything in life is driven. For instance our hair contains no muscles, although our arms do. One application of these ideas to biology is as a way to demonstrate that certain microscopic protrusions from a bacterium are more like arms than hairs. An observation of nonzero \,\Sigma proves that the system must be powered [ 2 ] , and such a proof holds even if we observe only a few of the degrees of freedom [ 3 ] . For more macroscopic living systems, we are seldom in doubt about whether they are driven, but we can still estimate their irreversibility from data. What can be learned in such cases? Inspired by Bennett et. al. [ 4 ] , this letter explores the meaning of macroscopic entropy production by studying experimental data from a rather large model organism, shown in figure 1 . Of course we do not observe all of its degrees of freedom, but working from what we do have, we can estimate the entropy production (using methods described in results) to be at least \,\Sigma\approx 0.5\,\text{bits/s}. (2) Unlike some microscopic biological systems, here we have little doubt that this system is driven — not only as a matter of natural law, but also that of the state legislature [ 5 ] . We can estimate its rate of work by measuring its diet (as described in the methods section), and converting to SI units we have: k_{B}T\,\Sigma\approx 2\times 10^{-21}\,\text{J/s}\qquad\leq W\approx 7\times 10^{4}\,\text{ J/s}. (3) We observe that the inequality is quite far from being saturated. The gap obtained in equation ( 3 ) is, to our knowledge, an increase on any so far reported in the literature. A gap of 8 orders of magnitude was reported by [ 6 ] for neurons, and 4 orders of magnitude in [ 7 ] , studying auditory hair cells. We can compute similar gaps for many other biological systems where a paper has reported something like \,\Sigma , by estimating their power W by independent means. The resulting comparisons are shown in figure 2 . The next section looks at how we arrived at equation ( 2 ) from data, and the uncertainties of its estimation. We return to questions of biology in the conclusion. Figure 1: (A) Automobilus subaru (left) is a large quadrirotus creature, originally from Japan. Shown with H. sapiens (centre) and S. demersus (right) for scale. (B) Raw data time-trace, showing speed, engine RPM, and coolant temperature. These three numbers are recorded about every 140 ms, using a device [ 8 ] which plugs into the OBD-II port [ 9 ] . We regard the temperature as a constant, and use the other two as x(t)\in\mathbb{R}^{2} for our EPR calculations (figures 3 , 4 ). 2 Results Because so many ways of estimating irreversibility have been proposed, we thought it might be a service to the field to apply many of them to the same system. As is traditional, we also introduce our own, superior, estimator, which we presume will soon be widely adopted. Alongside each formula, we give a numerical estimate read off figure 4 ; these involve some choices described below. Suppose that X(t) is a random variable taking values x\in\mathbb{R}^{d} at each time 0\leq t\leq t_{\mathrm{max}} . If we write its time-reverse as \tilde{x}(t)=x(t_{\mathrm{max}}-t) , then the statistical irreversibility is defined as the KL divergence between forward and reversed trajectories: \,\Sigma=\lim_{t_{\mathrm{max}}\to\infty}\frac{1}{t_{\mathrm{max}}}D_{\mathrm{KL}}\big[p(X)\big\|p(\tilde{X})\big]. (4) This quantity is often called the entropy production rate (EPR). The true microscopic EPR will saturate ( 3 ), and we write this as \dot{S} . The irreversibility \,\Sigma calculated from macroscopic variables may be viewed as one way to estimate a coarse-grained EPR. But there are other such estimates, independent of irreversibility. Below we calculate one, \dot{S}_{\mathrm{TUR}} , and discuss some others. The difficulty of estimating irreversibility \,\Sigma from data is that the space of possible trajectories integrated over in ( 4 ) is extremely sparsely covered. One solution is to assume a data-generating process for which we can calculate p(X) everywhere, and then fit the parameters of this process to the data. The first such method we use assumes Gaussian statistics, with continuous space and time. For this case, Seara et al. [ 10 ] derived: \,\Sigma_{\mathrm{gauss}}=\int\frac{d\omega}{4\pi}\mathop{\mathrm{Tr}}\left\{\left[C^{-1}(-\omega)-C^{-1}(\omega)\right]C(\omega)\right\}\qquad\Rightarrow 0.5\,\text{bits/s}. (5) Here C(\omega) is the d\times d covariance matrix, and \omega is frequency. If the covariance is estimated from a finite time-series, this formula will tend to over-estimate the true entropy production. The method of [ 10 ] uses a Gaussian smoothing \sigma in frequency space, for which they were able to calculate the \,\Sigma_{\mathrm{gauss}} for uncorrelated noise, finding \,\Sigma_{\mathrm{bias}}=d(d-1)N\sqrt{\pi}/(2t_{\mathrm{max}}^{2}\sigma) for N samples. Subtracting this bias leads to an improved estimate. Figure 2: Entropy production provides a bound on the rate of energy consumption, at a given temperature. Besides our work (black star), we plot some competing papers; only the one on L. catesbeianus hair cell bundles [ 7 ] provides figures for both axes (red star). For all other points, we have estimated energy consumption (as described in Methods) and plot irreversibility / entropy production rates taken from the following papers: Bos taurus [ 11 ] , H. sapiens [ 12 ] , Sturnus vulgaris [ 13 ] , microtubules and HEK cells [ 14 ] , neurons [ 6 ] , and the Turin public transit system [ 15 ] . The two red points are sub-cellular scale, while the green points are single cells. Another simple data-generating process is a Markov chain, with discrete time-steps, and the simplest case to consider has discrete state space too. If the flux from state i to state j is J_{ji} , then the irreversibility is: \,\Sigma_{\text{discrete}}=\sum_{i<j}(J_{ij}-J_{ji})\log\frac{J_{ij}}{J_{ji}}\qquad\Rightarrow 0.03\,\text{bits/s}. (6) Continuous data can be clustered into k discrete states, and for small enough k