Search

If dark matter is composed of axions, then axion stars form in the cores of dark matter halos. These stars are unstable above a critical mass, decaying to radio photons that heat the intergalactic medium, offering a new channel for axion indirect detection. We recently provided the first accurate calculation of the axion decay rate due to axion star mergers. In this work we show how existing data concerning the CMB optical depth leads to strong constraints on the axion photon coupling in the mass range $10^{-14}\,{\rm eV}\lesssim m_a\lesssim 10^{-8}\,{\rm eV}$. Axion star decays lead to efficient reionization of the intergalactic medium during the dark ages. By comparing this non-standard reionization with Planck legacy measurements of the Thompson optical width, we show that couplings in the range $10^{-14}\,{\rm GeV}^{-1} \lesssim g_{a\gamma\gamma} \lesssim 10^{-10}\,{\rm GeV}^{-1}$ are excluded for our benchmark model of axion star abundance. Future measurements of the 21cm emission of neutral hydrogen at high redshift could improve this limit by an order of magnitude or more, providing complementary indirect constraints on axion dark matter in parameter space also targeted by direct detection haloscopes., Comment: 5 pages, 4 figures + 8 pages of appendix. v2: matches published version. Added collisional excitations as a cooling channel, bounds are relaxed by a factor of ~ 2 wrt v1. Clarified assumptions upon which the bounds rest on. Improved discussion about the application of Planck bounds on tau. Conclusions remain unchanged

Solitons are observed to form in simulations of dark matter (DM) halos consisting of bosonic fields. We use the extended Press-Schechter formalism to compute the mass function of solitons, assuming various forms for the relationship between halo mass and soliton mass. We further provide a new calculation of the rate of soliton major mergers. Solitons composed of axion DM are unstable above a critical mass, and decay to either relativistic axions or photons, depending on the values of the coupling constants. We use the computed soliton major merger rate to predict the enhanced DM decay rate due to soliton instability. For certain values of currently allowed axion parameters, the energy injection into the intergalactic medium from soliton decays to photons is comparable to or larger than the energy injection due to core collapse supernovae at $z>10$. A companion paper explores the phenomenology of such an energy injection., Comment: 18 pages, 18 figures. Version 2: matches published version. Add an appendix discussing the timescale of axion star merging process due to dynamical friction

Post-training, layer-wise quantization is preferable because it is free from retraining and is hardware-friendly. Nevertheless, accuracy degradation has occurred when a neural network model has a big difference of per-out-channel weight ranges. In particular, the MobileNet family has a tragedy drop in top-1 accuracy from 70.60% ~ 71.87% to 0.1% on the ImageNet dataset after 8-bit weight quantization. To mitigate this significant accuracy reduction, we propose a new weight equalizing shift scaler, i.e. rescaling the weight range per channel by a 4-bit binary shift, prior to a layer-wise quantization. To recover the original output range, inverse binary shifting is efficiently fused to the existing per-layer scale compounding in the fixed-computing convolutional operator of the custom neural processing unit. The binary shift is a key feature of our algorithm, which significantly improved the accuracy performance without impeding the memory footprint. As a result, our proposed method achieved a top-1 accuracy of 69.78% ~ 70.96% in MobileNets and showed robust performance in varying network models and tasks, which is competitive to channel-wise quantization results., Comment: 9 pages, 4 figures, 4 tables

The distribution of natural vegetation on North Merritt Island, Florida, was studied by computer analysis of ERTS satellite multispectral-scanner data. The boundaries of six distinct plant associations were located in photos produced on the image analyzer with an insignificant mean error of -24.38 meters. The six plant associations are described as: aquatic estuarine association, mangrove, spartina swamp, wooded swamp, sabal hammock, and oak-palmetto. The difference in average reflectance 'grey level' between the lowest of the four spectral scanning bands and the highest spectral scanning band for each of the six vegetation types is described. The decreasing trend of the differences is strongly negatively correlated with height of land, the coefficient of correlation being -0.9696.