Dynamical Mechanism of Sampling-Based Probabilistic Inference under Probabilistic Population Codes

Animals make efficient probabilistic inferences based on uncertain and noisy information from the outside environment. It is known that probabilistic population codes, which have been proposed as a neural basis for encoding probability distributions, allow general neural networks (NNs) to perform near-optimal point estimation. However, the mechanism of sampling-based probabilistic inference has not been clarified. In this study, we trained two types of artificial NNs, feedforward NN (FFNN) and recurrent NN (RNN), to perform sampling-based probabilistic inference. Then we analyzed and compared their mechanisms of sampling. We found that sampling in RNN was performed by a mechanism that efficiently uses the properties of dynamical systems, unlike FFNN. In addition, we found that sampling in RNNs acted as an inductive bias, enabling a more accurate estimation than in maximum a posteriori estimation. These results provide important arguments for discussing the relationship between dynamical systems and information processing in NNs.

Negative self-feedback induces enhancement and transition of spiking activity for class 3 excitability

Post-inhibitory rebound (PIR) spike, which has been widely observed in diverse nervous systems with different physiological functions and simulated in theoretical models with class 2 excitability, presents a counterintuitive nonlinear phenomenon in that the inhibitory effect can facilitate neural firing behavior. In this study, a PIR spike induced by inhibitory stimulation from the resting state corresponding to class 3 excitability that is not related to bifurcation is simulated in the Morris-Lecar neuron. Additionally, the inhibitory self-feedback mediated by an autapse with time delay can evoke tonic/repetitive spiking from phasic/transient spiking. The dynamical mechanism for the PIR spike and the tonic/repetitive spiking is acquired with the phase plane analysis and the shape of the quasi-separatrix curve. The result extends the counterintuitive phenomenon induced by inhibition to class 3 excitability, which presents a potential function of inhibitory autapse and class 3 neuron in many neuronal systems such as the auditory system.

Mechanical Mechanism Analysis of Warping and Bulging Of Track Slab of High-Speed Railway in Service

Due to warpage and ballooning of high-speed rail track slab caused by debonding and environmental temperature changes during the deterioration of CA mortar layer being extremely concerned in practical engineering, it is based on numerical simulation of typical working conditions and analysis of test model that the deformation dynamics mechanism of track slab of high-speed rail in service is studied in this paper. Firstly, three kinds of damage conditions of CA mortar layer are designed to simulate the partial stress state of track slab under normal, warping and bulging conditions, and the results of model test are compared with those of finite element analysis so that the accuracy and credibility of the numerical simulation method and results are verified. Then, through finite element numerical simulation, the dynamical mechanism of actual full-scale high-speed rail track slab under vibration load is studied. The results show that the warping deformation around the track slab and the bulging deformation in the middle part under the action of positive and negative temperature gradient load caused by environmental temperature change will have a great impact on the structural performance of itself and CA mortar layer; Bulging deformation of track slab is more destructive to its structure than warping deformation. It is of great practical significance to further study the critical position of track plate warpage and bulging deformation, and to optimize and strengthen the structure of this part; The research results are of great significance to further study the deterioration.

The Identification and Prediction of Mesoscale Eddy Variation via Memory in Memory With Scheduled Sampling for Sea Level Anomaly

Ocean mesoscale eddies are ubiquitous in world ocean and account for 90% oceanic kinetic energy, which dominate the upper ocean flow field. Accurately predicting the variation of ocean mesoscale eddies is the key to understand the oceanic flow field and circulation system. In this article, we propose to make an initial attempt to explore spatio-temporal predictability of mesoscale eddies, employing deep learning architecture, which primarily establishes Memory In Memory (MIM) for sea level anomaly (SLA) prediction, combined with the existing mesoscale eddy detection. Oriented to the western Pacific ocean (125°−137.5°E and 15°−27.5°N), we quantitatively investigate the historic daily SLA variability at a 0.25° spatial resolution from 2000 to 2018, derived by satellite altimetry. We develop the enhanced MIM prediction strategies, equipped with Gated Recurrent Unit (GRU) and spatial attention module, in a scheduled sampling manner, which overcomes the gradient vanishing and complements to strengthen spatio-temporal features for long-term dependencies. At the early stage, the real value SLA input guides the model training process for initialization, while the scheduled sampling intentionally feeds the newly predicted value, to resolve the distribution inconsistency of inference. It has been demonstrated in our experiment results that our proposed prediction scheme outperformed the state-of-art approaches for SLA time series, with MAPE, RMSE of the 14-day prediction duration, respectively, 5.1%, 0.023 m on average, even up to 4.6%, 0.018 m for the effective sub-regions, compared to 19.8%, 0.086 m in ConvLSTM and 8.3%, 0.040 m in original MIM, which greatly facilitated the mesoscale eddy prediction. This proposed scheme will be beneficial to understand of the underlying dynamical mechanism behind the predictability of mesoscale eddies in the future, and help the deployment of ARGO, glider, AUV and other observational platforms.

Turbulence in Large-Scale Two-Dimensional Balanced Hard Sphere Gas Flow

In recent works, we developed a model of balanced gas flow, where the momentum equation possesses an additional mean field forcing term, which originates from the hard sphere interaction potential between the gas particles. We demonstrated that, in our model, a turbulent gas flow with a Kolmogorov kinetic energy spectrum develops from an otherwise laminar initial jet. In the current work, we investigate the possibility of a similar turbulent flow developing in a large-scale two-dimensional setting, where a strong external acceleration compresses the gas into a relatively thin slab along the third dimension. The main motivation behind the current work is the following. According to observations, horizontal turbulent motions in the Earth atmosphere manifest in a wide range of spatial scales, from hundreds of meters to thousands of kilometers. However, the air density rapidly decays with altitude, roughly by an order of magnitude each 15–20 km. This naturally raises the question as to whether or not there exists a dynamical mechanism which can produce large-scale turbulence within a purely two-dimensional gas flow. To our surprise, we discover that our model indeed produces turbulent flows and the corresponding Kolmogorov energy spectra in such a two-dimensional setting.

Analysis of the Neuron Dynamics in Thalamic Reticular Nucleus by a Reduced Model

Strategically located between the thalamus and the cortex, the inhibitory thalamic reticular nucleus (TRN) is a hub to regulate selective attention during wakefulness and control the thalamic and cortical oscillations during sleep. A salient feature of TRN neurons contributing to these functions is their characteristic firing patterns, ranging in a continuum from tonic spiking to bursting spiking. However, the dynamical mechanism under these firing behaviors is not well understood. In this study, by applying a reduction method to a full conductance-based neuron model, we construct a reduced three-variable model to investigate the dynamics of TRN neurons. We show that the reduced model can effectively reproduce the spiking patterns of TRN neurons as observed in vivo and in vitro experiments, and meanwhile allow us to perform bifurcation analysis of the spiking dynamics. Specifically, we demonstrate that the rebound bursting of a TRN neuron is a type of “fold/homo-clinic” bifurcation, and the tonic spiking is the fold cycle bifurcation. Further one-parameter bifurcation analysis reveals that the transition between these discharge patterns can be controlled by the external current. We expect that this reduced neuron model will help us to further study the complicated dynamics and functions of the TRN network.

Transition from Anti-coherence resonance to coherence resonance for mixed-mode oscillations and period-1 firing of nervous system

Identification of dynamics of the mixed-mode oscillations (MMOs), which exhibit transition between oscillations with large and small amplitudes, is very important for nonlinear physics. In this paper, the MMOs with transition between subthreshold oscillations and spikes are investigated in a neuron model. In the absence of noise, the MMOs appear between the resting state and period-1 firing with increasing depolarization current. After introducing white noise, coherence resonance (CR) is evoked from the resting state and non-CR is induced from period-1 firing far from the MMOs, which is consistent with the traditional viewpoint. However, an interesting result that a transition from anti-CR to CR is evoked by noise from both the MMOs and the period-1 firing near the MMOs is acquired, which is characterized by the increase, decrease and increase again of the coefficient of variations of interspike intervals (ISIs) with increasing noise intensity. At small noise intensity, more subthreshold oscillations are evoked by noise to reduce the firing frequency, resulting in faster increase of standard deviation (SD) of ISIs than that of mean value of ISIs, which is the cause for the anti-CR. The decrease of SD is faster for middle noise intensity and is lower for strong noise intensity, which is the cause for the CR. The different stochastic responses of MMOs and period-1 firing nearby at different levels of noise insanity are the dynamical mechanism for the transition from anti-CR to CR. Such results present potential functions of the MMOs and period-1 firing on information processing in the nervous system with noise and extend the conditions for the CR and anti-CR phenomena, which enriches the contents of nonlinear dynamics.

Spin Experimentation with Unpolarized Colliding Beams at the LHC

A brief recollection of the problems related to a significant hyperon polarization observed in pp-collisions is given with an emphasis on the general role of spin in the dynamics of hadron interactions. The old, unsolved problem of the observation of a significant hyperon polarization can provide new insights as a result of the measurements of energies at the LHC; in combination with other measurements, these can be used to tag the QGP formation in pp-collisions with colliding beams. Polarization studies in the processes of hyperon production do not require the use of polarized beams or targets and can be performed in the existing experimental environment at the LHC. Model predictions based on the chiral dynamics and pictures of the impact parameter are presented for the illustration of a possible dynamical mechanism that leads to a hyperon polarization.