科研成果

Ferrihydrite nanoparticles (Fh NPs) are ubiquitous in natural environments. However, their colloidal stability, and fate and transport behavior are difficult to predict in the presence of heterogeneous natural organic matter (NOM) mixtures. Here, we investigated the adsorption and aggregation behavior of Fh NPs exposed to NOM fractions with different molecular weights (MW). The NOM fraction with MW < 3 kDa destabilized the NPs, resulting in accelerated aggregation even at high C/Fe mass ratios, whereas higher MW NOM fractions imparted better colloidal stability with increasing MW and C/Fe ratio. Despite differences in the functional group composition of the bulk (dissolved) NOM fractions, all NOM fractions produced similar adsorbed layer compositions on the NPs, suggesting minimal contribution of chemical properties to the distinctive aggregation behavior. Rather, the higher adsorbed mass and larger size of the higher MW fractions were key factors in stabilizing the NPs through steric repulsion, whereas the lowest MW fraction had low adsorbed mass and was unable to counter electrostatic patch-charge attraction when the NPs are positively charged. This mechanistic understanding helps us predict the transport and fate of Fh NPs and the associated contaminants in natural environments with varying NOM compositions.

Complex networks have been an effective paradigm to represent a variety of complex systems, such as social networks, collaborative networks, and biomolecular networks, where network topology is unkown in advance and has to be inferred with limited observed measurements. Compressive sensing (CS) theory is an efficient technique to achieve accurate network reconstruction in complex networks by formulating the problem as a series of convex optimization models and utilizing the sparsity of networks. However, previous CS-based works have to solve a large number of convex optimization models, which is time-consuming especially when the network scale becomes large. Further, since partial link information shared among multiple convex models, data conflict problem may incur when the derived common variables are inconsistent, which may badly degrade infer precision. To address the issues above, we propose a new model for network reconstruction based on compressive sensing. To be specific, a single convex optimization model is formulated for inferring global network structure by combing the series of convex optimization models, which can effectively improve computation efficiency. Further, we devise a vector to represent the connection states of all the nodes without redundant link information, which is used for representing the unkown topology variables in the proposed optimization model based a devised transformation method. In this way, the proposed model can eliminate data conflict problem and improve infer precision. The comprehensive simulation results shows the superiority of the proposed model compared with the competitive algorithms under a wide variety of scenarios.

Electron emission (EE) from electron sources based on island metal films (IMFs) was mainly attributed to field emission or thermionic emission from IMFs. Here, we propose a new mechanism of EE from IMF-based sources, namely, EE from horizontal tunneling junctions formed in the substrate. The devices with and without IMFs fabricated on silicon oxide substrates are found to exhibit similar EE properties, while the island-metal-film-based devices fabricated on Si3N4/Si substrate show no EE. The comparative results indicate that EE originates from the underlying silicon oxide substrate that was ignored in previous mechanisms, but not from IMFs. EE from the devices is thought to be generated from horizontal tunneling junctions in electroformed silicon oxide substrate due to the rupture of conducting filaments. Even though metal island films are not the origin of EE, they can greatly decrease the forming voltage of the devices. The insights into the emission mechanism are helpful for optimizing the electron source performances.

This paper proposes a new error upper bound formula for the Gaussian integration of the near-singular integral using the Boundary Element Method. First, this study found through numerical tests that the maximum relative error of the Gaussian integration has a downward concave shape but an approximately linear relationship with the relative distance, which is defined as the ratio of the distance from the source point to the element over the element length in a semi-logarithmic plot. Thus, the error upper bound can be defined as a line that closely approaches the computed error data points from the upper side. This line can be obtained by connecting two specified data points that are located outside, but very close to, the considered error range. Further research indicates that one parameter of the fitted line has a linear relationship with the number of Gaussian integration points and singularity orders and the other parameter can be treated as a constant, which together make the proposed Gaussian integration error upper bound formula widely applicable. Compared to the Lachat and Watson criterion, the proposed formula requires fewer integration points when the source point is very close to the element and thus serves to improve computational efficiency. The proposed formula also avoids calculation failure that can occur when using the Davies and Bu criterion. The numerical example results show that the proposed error upper bound formula can evaluate the integration accuracy well and improve computational efficiency when using an adaptive Gaussian integration method.

Perceptual learning, which improves stimulus discrimination, typically results from training with a single stimulus condition. Two major learning mechanisms, early cortical neural plasticity and response reweighting, have been proposed. Here we report a new format of perceptual learning that by design may have bypassed these mechanisms. Instead it is more likely based on abstracted stimulus evidence from multiple stimulus conditions. Specifically, we had observers practice orientation discrimination with Gabors or symmetric dot patterns at up to 47 random or rotating location´orientation conditions. Although each condition received sparse trials (16 trials/session), the practice produced significant orientation learning. Learning also transferred to a Gabor at a single untrained condition with 2-3 time lower orientation thresholds. Moreover, practicing a single stimulus condition with matched trial frequency (16 trials/session) failed to produce significant learning. These results suggested that learning with multiple stimulus conditions may not come from early cortical plasticity or response reweighting with each particular condition. Rather, it may materialize through a new format of perceptual learning, in which orientation evidence invariant to particular orientations and locations is first abstracted from multiple stimulus conditions, and then reweighted by later learning mechanisms. The coarse-to-fine transfer of orientation learning from multiple Gabors or symmetric-dot-patterns to a single Gabor also suggested the involvement of orientation concept learning by the learning mechanisms.

Estimating permeability of carbonate rocks using mercury injection capillary pressure (MICP) data has been carried out by many researchers in the past few decades. However, a major issue with almost all of the existing models is that they focus on a single aperture value from the capillary pressure curve. This study builds a new model to extract permeability from the entire pore throat sizes. Fermic-Dirac function was applied to fit the MICP curve to obtain some critical parameters such as R1 (the large curvature value) and R2 (the small curvature value). Afterwards, the partial least squares regression method was employed to develop a new permeability model. To verify the new model and check other models, we studied ten carbonate rock samples from an Iranian oil reservoir. The results showed that the R1 values vary from 1.00 to 2.73 while R2 values are found between 0.23 and 1.00. The new model performed better than the published models. The idea of building the model for the carbonates can be used in developing the permeability estimating model for shale samples, which could be a new model for the shale permeability estimation.

Patients with central vision loss depend on peripheral vision for everyday functions. A preferred retinal locus (PRL) on the intact retina is commonly trained as a new “fovea” to help. However, reprogramming the fovea-centered oculomotor control is difficult, so saccades often bring the defunct fovea to block the target. Aligning PRL with distant targets also requires multiple saccades and sometimes head movements. To overcome these problems, we attempted to train normal-sighted observers to form a preferred retinal annulus (PRA) around a simulated scotoma, so that they could rely on the same fovea-centered oculomotor system and make short saccades to align PRA with the target. Observers with an invisible simulated central scotoma (5° radius) practiced making saccades to see a tumbling-E target at 10° eccentricity. The otherwise blurred E target became clear when saccades brought a scotoma-abutting clear window (2° radius) to it. The location of the clear window was either fixed for PRL training, or changing among 12 locations for PRA training. Various cues aided the saccades through training. Practice quickly established a PRL or PRA. Comparing to PRL-trained observers whose first saccades persistently blocked the target with scotoma, PRA-trained observers produced more accurate first saccades. The benefits of more accurate PRA-based saccades also outweighed the costs of slower latency. PRA training may provide a very efficient strategy to cope with central vision loss, especially for aging patients who have major difficulties adapting to a PRL.

Silicate Earth is widely considered identical to chondrites in its refractory lithophile element ratios. However, its subchondritic Nb/Ta signature deviates from the chondritic paradigm. To resolve this Nb deficit, its sequestration in Earth's core under very reducing core-forming conditions has been proposed based on low-pressure data. Here, we show that under conditions relevant to core formation Nb is siderophile at high pressures under all redox conditions, corroborating Nb inventory in Earth's core. Further core formation modeling shows that Earth's core could have formed under moderately reducing or oxidizing conditions, whereas highly reducing conditions mismatch the geochemical observables; although Earth may have sampled a variety of reservoirs, it is problematic to accrete primarily from materials as reduced as enstatite chondrites.

In this study, three pilot-scale solid-phase denitrification (SPD) systems filled with poly-3-hydroxybutyrate-co-hyroxyvelate (PHBV), PHBV-Rice hulls (PHBV-RH) and PHBV-Sawdust (PHBV-S) were operated to treat effluent of waste water treatment pangts (WWTPs). The fast start-up and intensified nitrogen removal performance were obtained in PHBV-RH and PHBV-S systems. Besides, the optimal total nitrogen (TN) removal efficiency was obtained in PHBV-S system (91.65 +/- 4.12%) with less ammonia accumulation and dissolved organic carbon (DOC) release. The significant enrichment of amx 16S rRNA and nirS genes in PHBV-RH and PHBV-S systems indicated the possible coexistence of anammox and denitrification. Miseq sequencing analysis exhibited more complex community diversity, more abundant denitrifying and fermenting bacteria in PHBV-RH and PHBV-S systems. The co-existence of denitrification and anammox might contribute to better control of nitrogen and dissolved organic carbon in PHBV-S system. The outcomes provide an economical and eco-friendly alternative to improve nitrogen removal of WWTPs effluent.

The oxidation of nitric oxide to nitrogen dioxide by hydroperoxy (HO2) and organic peroxy radicals (RO2) is responsible for the chemical net ozone production in the troposphere and for the regeneration of hydroxyl radicals, the most important oxidant in the atmosphere. In Summer 2014, a field campaign was conducted in the North China Plain, where increasingly severe ozone pollution has been experienced in the last years. Chemical conditions in the campaign were representative for this area. Radical and trace gas concentrations were measured, allowing for calculating the turnover rates of gas-phase radical reactions. Therefor; the importance of heterogeneous HO(2 )uptake on aerosol could be experimentally determined. HO2 uptake could have suppressed ozone formation at that time because of the competition with gas-phase reactions that produce ozone. The successful reduction of the aerosol load in the North China Plain in the last years could have led to a significant decrease of HO2 loss on particles, so that ozone-forming reactions could have gained importance in the last years. However, the analysis of the measured radical budget in this campaign shows that HO2 aerosol uptake did not impact radical chemistry for chemical conditions in 2014. Therefore, reduced HO2 uptake on aerosol since then is likely not the reason for the increasing number of ozone pollution events in the North China Plain, contradicting conclusions made from model calculations reported in the literature.

During the EXPLORE-YRD campaign (EXPeriment on the eLucidation of the atmospheric Oxidation capacity and aerosol foRmation, and their Effects in Yangtze River Delta) in May June 2018, we measured N2O5, NO2, O-3 and relevant parameters at a regional site in Taizhou, Jiangsu Province. The nocturnal average NO3 production rate was 1.01 +/- 0.47 ppbvh(-1), but the mixing ratio of N2O5 was low, with a maximum of 220 pptv in 1 min, suggesting rapid loss of NO3 and N2O5. The nocturnal steady-state lifetime of N2O5 was 43 + 52 s on average, which may be attributed to the elevated monoterpene and fast N2O5 uptake. VOCs (mainly monoterpenes) dominated daily NO3 loss with the percentage of 36.4% and N2O5 uptake accounted for 14.4%, when taking NO + NO3 and NO3 photolysis into consideration. We demonstrated that the nonnegligible daytime NO3 oxidation of monoterpene in YRD region, which contributes to the daytime formation of organic nitrate and secondary organic aerosol. The daily average NOx consumption rate via rapid NO3 reaction reached 0.63 ppbvh(-1), corresponding to 57.3% NOx loss in comparison with the OH oxidation pathway at this site, highlighting the key role of NO3 and N2O5 in NOx removal and subsequent photochemistry in the YRD region.

Numerous experimental studies suggest that noise is inherent in the human brain. However, the functional importance of noise remains unknown. n particular, from a computational perspective, such stochasticity is potentially harmful to brain function. In machine learning, a large number of saddle points are surrounded by high error plateaus and give the illusion of the existence of local minimum. As a result, being trapped in the saddle points can dramatically impair learning and adding noise will attack such saddle point problems in high-dimensional optimization, especially under the strict saddle condition. Motivated by these arguments, we propose one biologically plausible noise structure and demonstrate that noise can efficiently improve the optimization performance of spiking neural networks based on stochastic gradient descent. The strict saddle condition for synaptic plasticity is deduced, and under such conditions, noise can help optimization escape from saddle points on high dimensional domains. The theoretical results explain the stochasticity of synapses and guide us on how to make use of noise. In addition, we provide biological interpretations of proposed noise structures from two points: one based on the free energy principle in neuroscience and another based on observations of in vivo experiments. Our simulation results manifest that in the learning and test phase, the accuracy of synaptic sampling with noise is almost 20% higher than that without noise for synthesis dataset, and the gain in accuracy with/without noise is at least 10% for the MNIST and CIFAR-10 dataset. Our study provides a new learning framework for the brain and sheds new light on deep noisy spiking neural networks.