Optimal transportation plays a fundamental role in many fi elds in engineering and medicine,including surface parameterization in graphics,registration in computer vision,and generative models in deep learning.For qua...Optimal transportation plays a fundamental role in many fi elds in engineering and medicine,including surface parameterization in graphics,registration in computer vision,and generative models in deep learning.For quadratic distance cost,optimal transportation map is the gradient of the Brenier potential,which can be obtained by solving the Monge-Ampère equation.Furthermore,it is induced to a geometric convex optimization problem.The Monge-Ampère equation is highly non-linear,and during the solving process,the intermediate solutions have to be strictly convex.Specifi cally,the accuracy of the discrete solution heavily depends on the sampling pattern of the target measure.In this work,we propose a self-adaptive sampling algorithm which greatly reduces the sampling bias and improves the accuracy and robustness of the discrete solutions.Experimental results demonstrate the efficiency and efficacy of our method.展开更多
An orthogonal 2D training image is constructed from the geological analysis results of well logs and sedimentary facies;the 2 D probabilities in three directions are obtained through linear pooling method and then agg...An orthogonal 2D training image is constructed from the geological analysis results of well logs and sedimentary facies;the 2 D probabilities in three directions are obtained through linear pooling method and then aggregated by the logarithmic linear pooling to determine the 3 D multi-point pattern probabilities at the unknown points,to realize the reconstruction of a 3 D model from 2D cross-section.To solve the problems of reducing pattern variability in the 2 D training image and increasing sampling uncertainty,an adaptive spatial sampling method is introduced,and an iterative simulation strategy is adopted,in which sample points from the region with higher reliability of the previous simulation results are extracted to be additional condition points in the following simulation to improve the pattern probability sampling stability.The comparison of lateral accretion layer conceptual models shows that the reconstructing algorithm using self-adaptive spatial sampling can improve the accuracy of pattern sampling and rationality of spatial structure characteristics,and accurately reflect the morphology and distribution pattern of the lateral accretion layer.Application of the method in reconstructing the meandering river reservoir of the Cretaceous McMurray Formation in Canada shows that the new method can accurately reproduce the shape,spatial distribution pattern and development features of complex lateral accretion layers in the meandering river reservoir under tide effect.The test by sparse wells shows that the simulation accuracy is above 85%,and the coincidence rate of interpretation and prediction results of newly drilled horizontal wells is up to 80%.展开更多
基金the National Numerical Wind Tunnel Project,China(No.NNW2019ZT5-B13)the National Natural Science Foundation of China(Nos.61907005,61772105,61936002,and 61720106005)。
文摘Optimal transportation plays a fundamental role in many fi elds in engineering and medicine,including surface parameterization in graphics,registration in computer vision,and generative models in deep learning.For quadratic distance cost,optimal transportation map is the gradient of the Brenier potential,which can be obtained by solving the Monge-Ampère equation.Furthermore,it is induced to a geometric convex optimization problem.The Monge-Ampère equation is highly non-linear,and during the solving process,the intermediate solutions have to be strictly convex.Specifi cally,the accuracy of the discrete solution heavily depends on the sampling pattern of the target measure.In this work,we propose a self-adaptive sampling algorithm which greatly reduces the sampling bias and improves the accuracy and robustness of the discrete solutions.Experimental results demonstrate the efficiency and efficacy of our method.
基金Supported by the China National Science and Technology Major Project(2017ZX05005-004-002,2016ZX05031-002-001)National Natural Science Foundation of China(41872138)Open Foundation of Top Disciplines in Yangtze University(2019KFJJ0818029)。
文摘An orthogonal 2D training image is constructed from the geological analysis results of well logs and sedimentary facies;the 2 D probabilities in three directions are obtained through linear pooling method and then aggregated by the logarithmic linear pooling to determine the 3 D multi-point pattern probabilities at the unknown points,to realize the reconstruction of a 3 D model from 2D cross-section.To solve the problems of reducing pattern variability in the 2 D training image and increasing sampling uncertainty,an adaptive spatial sampling method is introduced,and an iterative simulation strategy is adopted,in which sample points from the region with higher reliability of the previous simulation results are extracted to be additional condition points in the following simulation to improve the pattern probability sampling stability.The comparison of lateral accretion layer conceptual models shows that the reconstructing algorithm using self-adaptive spatial sampling can improve the accuracy of pattern sampling and rationality of spatial structure characteristics,and accurately reflect the morphology and distribution pattern of the lateral accretion layer.Application of the method in reconstructing the meandering river reservoir of the Cretaceous McMurray Formation in Canada shows that the new method can accurately reproduce the shape,spatial distribution pattern and development features of complex lateral accretion layers in the meandering river reservoir under tide effect.The test by sparse wells shows that the simulation accuracy is above 85%,and the coincidence rate of interpretation and prediction results of newly drilled horizontal wells is up to 80%.
