arXiv subjects

Ian Wright

Publications and source records attributed to Ian Wright.

7 recordsLinked to original sources

arXiv2023

$\partial\mathbb{B}$ nets: learning discrete functions by gradient descent

$\partial\mathbb{B}$ nets are differentiable neural networks that learn discrete boolean-valued functions by gradient descent. $\partial\mathbb{B}$ nets have two semantically equivalent aspects: a differentiable soft-net, with real weights, and a non-differentiable hard-net, with boolean weights. We train the soft-net by backpropagation and then `harden' the learned weights to yield boolean weights that bind with the hard-net. The result is a learned discrete function. `Hardening' involves no loss of accuracy, unlike existing approaches to neural network binarization. Preliminary experiments demonstrate that $\partial\mathbb{B}$ nets achieve comparable performance on standard machine learning problems yet are compact (due to 1-bit weights) and interpretable (due to the logical nature of the learnt functions).

cs.LG↗

arXiv2023

Self-Sustained Oscillations in a Low Viscosity Round Jet

This experimental study investigates the effects of viscosity contrast between a saltwater jet and its high-viscosity propylene glycol surroundings. Using density-matched fluids in a gravity-driven flow, Jet Reynolds numbers (Re) from 1600 to 3400 and ambient-to-jet viscosity ratios (M) from 1 to 50 were examined. Observations indicate that low viscosity ratios lead to axisymmetric jet breakdown, while higher ratios result in helical modes. The study employs various diagnostic tools to delineate this transition. Hot film anemometry reveals a discrete peak in the velocity fluctuation frequency spectrum, marking the onset of helical modes. This peak shows minimal spatial variation downstream. Laser-induced fluorescence (LIF) was utilized to distinguish the jet boundary. High-speed LIF imaging facilitated the determination of wave growth rates on the jet boundary and oscillation frequencies of the interface. These frequencies, consistent with self-sustained oscillation behavior, depend on the viscosity ratio and align with absolutely unstable modes from spatio-temporal linear stability theory. Spectral Proper Orthogonal Decomposition of the images identifies several spatial modes, predominantly a single dominant mode. These findings suggest the presence of a global mode, likely stemming from absolute instability in velocity and viscosity profiles near the nozzle exit, providing significant insights into fluid dynamics at varying viscosity contrasts.

physics.flu-dyn↗

arXiv2019

The standard coder: a machine learning approach to measuring the effort required to produce source code change

We apply machine learning to version control data to measure the quantity of effort required to produce source code changes. We construct a model of a `standard coder' trained from examples of code changes produced by actual software developers together with the labor time they supplied. The effort of a code change is then defined as the labor hours supplied by the standard coder to produce that change. We therefore reduce heterogeneous, structured code changes to a scalar measure of effort derived from large quantities of empirical data on the coding behavior of software developers. The standard coder replaces traditional metrics, such as lines-of-code or function point analysis, and yields new insights into what code changes require more or less effort.

cs.SE↗

arXiv2017

Decay of COSAC and Ptolemy Mass Spectra at Comet 67P/Churyumov-Gerasimenko

The Rosetta lander Philae successfully landed on the nucleus of comet 67P/Churyumov-Gerasimenko on 12 November 2014. Philae is equipped with two gas analyzers: The Cometary Sampling and Composition experiment (COSAC) and the gas chromatograph and mass spectrometer Ptolemy. On 12 to 14 November 2014 both instruments measured the organic composition of the comet nucleus material through seven measurements in sniffing mode during Philae's hopping and at its final landing site Abydos. We compare the temporal evolution of intensities of several ion species identified by both mass spectrometers. For COSAC this is the first analysis of the temporal behaviour of the measured ion species. All ion species showed the highest intensities in the first spectra measured by both instruments about 20 to 30 minutes after Philae's first touchdown at Agilkia, and a decay during the six consecutive measurements at Abydos. Both instruments measured a nearly identical decay of the water peak (m/z 18), and also CO (m/z 28) behaved similarly. In the COSAC measurements the peak at m/z 44 decays much slower than all the other ion species, including the water peak. In particular, the m/z 44 peak decays much slower in the COSAC measurements than in the Ptolemy data. This supports our earlier interpretation that COSAC for the first time analyzed a regolith sample from a cometary nucleus in situ, while Ptolemy measured cometary gas from the ambient coma. The m/z 44 peak measured by COSAC was likely dominated by organic species, whereas the peak measured by Ptolemy was interpreted to be mostly due to $CO_2$. Ion species heavier than m/z 30 tend to decay somewhat slower in the COSAC measurements than in the Ptolemy data, which may be related to differences in the exhaust designs between both instruments.

astro-ph.EP↗

arXiv2011

A conjecture on the distribution of firm profit

A common assumption of political economy is that profit rates across firms or sectors tend to uniformity, and often models are formulated in which this tendency is assumed to have been realised. But in reality this tendency is never realised and the distribution of firm profits is not degenerate but skewed to the right. The mode is less than the mean and super-profits are present. To understand the distribution of firm profits a general probabilistic argument is sketched that yields a candidate functional form. The overall properties of the derived distribution are qualitatively consistent with empirical measures, although there is more work to be done.

cond-mat.other↗

arXiv2011

The duration of recessions follows an exponential not a power law

Ormerod and Mounfield analysed GDP data of 17 leading capitalist economies from 1870 to 1994 and concluded that the frequency of the duration of recessions is consistent with a power-law. But in fact the data is consistent with an exponential (Boltzmann-Gibbs) law.

cond-mat↗

arXiv2011

The Social Architecture of Capitalism

A dynamic model of the social relations between workers and capitalists is introduced. The model is deduced from the assumption that the law of value is an organising principle of modern economies. The model self-organises into a dynamic equilibrium with statistical properties that are in close qualitative and in many cases quantitative agreement with a broad range of known empirical distributions of developed capitalism, including the power-law distribution of firm size, the Laplace distribution of firm and GDP growth, the lognormal distribution of firm demises, the exponential distribution of the duration of recessions, the lognormal-Pareto distribution of income, and the gamma-like distribution of the rate-of-profit of firms. Normally these distributions are studied in isolation, but this model unifies and connects them within a single causal framework. In addition, the model generates business cycle phenomena, including fluctuating wage and profit shares in national income about values consistent with empirical studies. A testable consequence of the model is a conjecture that the rate-of-profit distribution is consistent with a parameter-mix of a ratio of normal variates with means and variances that depend on a firm size parameter that is distributed according to a power-law.

cond-mat↗