Deep learning techniques for solving elliptic interface problems have gained significant attentions.In this paper,we introduce a hybrid residual and weak form(HRW)loss aimed at mitigating the challenge of model traini...Deep learning techniques for solving elliptic interface problems have gained significant attentions.In this paper,we introduce a hybrid residual and weak form(HRW)loss aimed at mitigating the challenge of model training.HRW utilizes the functions residual loss and Ritz method in an adversary-system,which enhances the probability of jumping out of the local optimum even when the loss landscape comprises multiple soft constraints(regularization terms),thus improving model’s capability and robustness.For the problem with interface conditions,unlike existing methods that use the domain decomposition,we design a Pre-activated ResNet of ResNet(PRoR)network structure employing a single network to feed both coordinates and corresponding subdomain indicators,thus reduces the number of parameters.The effectiveness and improvements of the PRoR with HRW are verified on two-dimensional interface problems with regular or irregular interfaces.We then apply the PRoR with HRW to solve the size-modified Poisson-Boltzmann equation,an improved dielectric continuum model for predicting the electrostatic potentials in an ionic solvent by considering the steric effects.Our findings demonstrate that the PRoR with HRW accurately approximates solvation free-energies of three proteins with irregular interfaces,showing the competitive results compared to the ones obtained using the finite element method.展开更多
An improved quantum-behaved particle swarm optimization （IQPSO） algorithm is employed to deter- mine aerosol size distribution （ASD）. The direct problem is solved using the anomalous diffraction approximation and ...An improved quantum-behaved particle swarm optimization （IQPSO） algorithm is employed to deter- mine aerosol size distribution （ASD）. The direct problem is solved using the anomalous diffraction approximation and Lambert-Beer＇s Law. Compared with the standard particle swarm optimization algo- rithm, the stochastic particle size optimization algorithm and the original QPSO, our IQPSO has faster convergence speed and higher accuracy within a smaller number of generations. Optimization param- eters for the IQPSO were also evaluated; we recommend using four measurement wavelengths and S0 particles. Size distributions of various aerosol types were estimated using the IQPSO under dependent and independent models. Finally, experimental ASDs at different locations in Harbin were recovered using the IQPSO. All our results confirm that the IQpSO algorithm is an effective and reliable technique for estimatinz ASD.展开更多
The angular light-scattering measurement（ALSM） method combined with an improved artificial bee colony algorithm is introduced to determine aerosol optical constants and aerosol size distribution（ASD） simultaneousl...The angular light-scattering measurement（ALSM） method combined with an improved artificial bee colony algorithm is introduced to determine aerosol optical constants and aerosol size distribution（ASD） simultaneously. Meanwhile, an optimized selection principle of the ALSM signals based on the sensitivity analysis and principle component analysis（PCA）is proposed to improve the accuracy of the retrieval results. The sensitivity analysis of the ALSM signals to the optical constants or characteristic parameters in the ASD is studied first to find the optimized selection region of measurement angles. Then, the PCA is adopted to select the optimized measurement angles within the optimized selection region obtained by sensitivity analysis. The investigation reveals that, compared with random selection measurement angles, the optimized selection measurement angles can provide more useful measurement information to ensure the retrieval accuracy. Finally,the aerosol optical constants and the ASDs are reconstructed simultaneously. The results show that the retrieval accuracy of refractive indices is better than that of absorption indices, while the characteristic parameters in ASDs have similar retrieval accuracy. Moreover, the retrieval accuracy in studying L-N distribution is a little better than that in studying Gamma distribution for the difference of corresponding correlation coefficient matrixes of the ALSM signals. All the results confirm that the proposed technique is an effective and reliable technique in estimating the aerosol optical constants and ASD simultaneously.展开更多
文摘Deep learning techniques for solving elliptic interface problems have gained significant attentions.In this paper,we introduce a hybrid residual and weak form(HRW)loss aimed at mitigating the challenge of model training.HRW utilizes the functions residual loss and Ritz method in an adversary-system,which enhances the probability of jumping out of the local optimum even when the loss landscape comprises multiple soft constraints(regularization terms),thus improving model’s capability and robustness.For the problem with interface conditions,unlike existing methods that use the domain decomposition,we design a Pre-activated ResNet of ResNet(PRoR)network structure employing a single network to feed both coordinates and corresponding subdomain indicators,thus reduces the number of parameters.The effectiveness and improvements of the PRoR with HRW are verified on two-dimensional interface problems with regular or irregular interfaces.We then apply the PRoR with HRW to solve the size-modified Poisson-Boltzmann equation,an improved dielectric continuum model for predicting the electrostatic potentials in an ionic solvent by considering the steric effects.Our findings demonstrate that the PRoR with HRW accurately approximates solvation free-energies of three proteins with irregular interfaces,showing the competitive results compared to the ones obtained using the finite element method.
基金Support from the National Natural Science Foundation of China （No. 51476043）, the Major National Scientific Instruments and Equipment Development Special Foundation of China （No. 51327803） and the Foundation for Innovative Research Groups of the National Natural Science Foundation of China （No. 51421063） is gratefully acknowledged.
文摘An improved quantum-behaved particle swarm optimization （IQPSO） algorithm is employed to deter- mine aerosol size distribution （ASD）. The direct problem is solved using the anomalous diffraction approximation and Lambert-Beer＇s Law. Compared with the standard particle swarm optimization algo- rithm, the stochastic particle size optimization algorithm and the original QPSO, our IQPSO has faster convergence speed and higher accuracy within a smaller number of generations. Optimization param- eters for the IQPSO were also evaluated; we recommend using four measurement wavelengths and S0 particles. Size distributions of various aerosol types were estimated using the IQPSO under dependent and independent models. Finally, experimental ASDs at different locations in Harbin were recovered using the IQPSO. All our results confirm that the IQpSO algorithm is an effective and reliable technique for estimatinz ASD.
基金Project supported by the Jiangsu Provincial Natural Science Foundation,China(Grant Nos.BK20170800 and BK20160794)the National Natural Science Foundation of China(Grant No.51606095)
文摘The angular light-scattering measurement（ALSM） method combined with an improved artificial bee colony algorithm is introduced to determine aerosol optical constants and aerosol size distribution（ASD） simultaneously. Meanwhile, an optimized selection principle of the ALSM signals based on the sensitivity analysis and principle component analysis（PCA）is proposed to improve the accuracy of the retrieval results. The sensitivity analysis of the ALSM signals to the optical constants or characteristic parameters in the ASD is studied first to find the optimized selection region of measurement angles. Then, the PCA is adopted to select the optimized measurement angles within the optimized selection region obtained by sensitivity analysis. The investigation reveals that, compared with random selection measurement angles, the optimized selection measurement angles can provide more useful measurement information to ensure the retrieval accuracy. Finally,the aerosol optical constants and the ASDs are reconstructed simultaneously. The results show that the retrieval accuracy of refractive indices is better than that of absorption indices, while the characteristic parameters in ASDs have similar retrieval accuracy. Moreover, the retrieval accuracy in studying L-N distribution is a little better than that in studying Gamma distribution for the difference of corresponding correlation coefficient matrixes of the ALSM signals. All the results confirm that the proposed technique is an effective and reliable technique in estimating the aerosol optical constants and ASD simultaneously.
