- Leiden University, Institute of Environmental Sciences, Faculty Memberadd
Last several years the ship accidents caused the serious ecological catastrophes in the seaside of different countries. To prevent or, at least, reduce the possible damage due to accidents we need to develop efficient and accurate model... more
Last several years the ship accidents caused the serious ecological catastrophes in the seaside of different countries. To prevent or, at least, reduce the possible damage due to accidents we need to develop efficient and accurate model for the simulation of the pollution spreading. The movement of the pollutant can be modeled by using a random walk model. Here the trajectory of a particle of the pollutant is simulated with the help of the appropriate system of the stochastic differential equations. By averaging the positions of many particles the concentration of the pollutant can be found. For a number of application,it is not necessary to simulate the concentration in the whole domain of the problem For these kind of problem the forward- reverse estimator can be applied. This estimator has recently been introduced by Milstein, Schoenmakers and Spokoiny and is based on realizations of original forward system and also on realizations of reverse time system derived from original for...
Research Interests:
Research Interests:
Research Interests:
In this paper we describe the design and implementation of a parallel algorithm for data assimilation with ensemble Kalman filter (EnKF) for oil reservoir history matching problem. The use of large number of observations from time-lapse... more
In this paper we describe the design and implementation of a parallel algorithm for data assimilation with ensemble Kalman filter (EnKF) for oil reservoir history matching problem. The use of large number of observations from time-lapse seismic leads to a large turnaround time for the analysis step, in addition to the time consuming simulations of the realizations. For efficient parallelization it is important to consider parallel computation at the analysis step. Our experiments show that parallelization of the analysis step in addition to the forecast step has good scalability, exploiting the same set of resources with some additional efforts.
Research Interests:
Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep concentration... more
Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep concentration gradients after the pollutant has just entered into the water, whereas conventional numerical methods such as finite difference and finite volume methods may have difficulties. Since the computation time in a particle model increases linearly with the number of particles, this often forms a limiting factor. We consider the parallelization of the particle model SIMPAR. Different load balancing and communication optimization possibilities are investigated. Some experiments with the parallel implementation of the SIMPAR model on a cluster of workstations and the Cray T3E are also reported. Introduction Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. These methods are es...
Last several years the ship accidents caused the serious ecological catastrophes in the seaside of different countries. To prevent or, at least, reduce the possible damage due to accidents we need to develop efficient and accurate model... more
Last several years the ship accidents caused the serious ecological catastrophes in the seaside of different countries. To prevent or, at least, reduce the possible damage due to accidents we need to develop efficient and accurate model for the simulation of the pollution spreading. The movement of the pollutant can be modeled by using a random walk model. Here the trajectory of a particle of the pollutant is simulated with the help of the appropriate system of the stochastic differential equations. By averaging the positions of many particles the concentration of the pollutant can be found. For a number of application,it is not necessary to simulate the concentration in the whole domain of the problem For these kind of problem the forward- reverse estimator can be applied. This estimator has recently been introduced by Milstein, Schoenmakers and Spokoiny and is based on realizations of original forward system and also on realizations of reverse time system derived from original for...
Research Interests:
Research Interests:
In this paper we discuss a classic clustering algorithm that can be used to segment images and a recently developed active contour image segmentation model. We propose integrating aspects of the classic algorithm to improve the active... more
In this paper we discuss a classic clustering algorithm that can be used to segment images and a recently developed active contour image segmentation model. We propose integrating aspects of the classic algorithm to improve the active contour model. For the resulting CVK and B-means segmentation algorithms we examine methods to decrease the size of the image domain. The CVK method has been implemented to run on parallel and distributed computers. By changing the order of updating the pixels, it was possible to replace synchronous communication with asynchronous communication and subsequently the parallel efficiency is improved.
Research Interests:
Research Interests:
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a... more
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a careful study on the characteristics of the ensemble ash state, we propose a mask-state algorithm which records the sparsity information of the full ensemble state matrix and transforms the full matrix into a relatively small one. This will reduce the computational cost in the analysis step. Experimental results show the mask-state algorithm significantly speeds up the expensive analysis step. Subsequently, the total amount of computing time for volcanic ash data assimilation is reduced to an acceptable level, which is important for providing timely and accurate aviation advices. The mask-state algorithm is generic and thus can be embedded in any ensemble-based data assimilation framework. Moreover, ensemble-based data assimilation with the mask-state a...
In this paper we discuss a classic clustering algorithm that can be used to segment images and a recently developed active contour image segmentation model. We propose integrating aspects of the classic algorithm to improve the active... more
In this paper we discuss a classic clustering algorithm that can be used to segment images and a recently developed active contour image segmentation model. We propose integrating aspects of the classic algorithm to improve the active contour model. For the resulting CVK and B-means segmentation algorithms we examine methods to decrease the size of the image domain. The CVK method has been implemented to run on parallel and distributed computers. By changing the order of updating the pixels, it was possible to replace synchronous communication with asynchronous communication and subsequently the parallel efficiency is improved.
Research Interests:
Remote sensing, as a powerful tool for monitoring atmospheric phenomena, has been playing an increasingly important role in inverse modeling. Remote sensing instruments measure quantities that often combine several state variables as one.... more
Remote sensing, as a powerful tool for monitoring atmospheric phenomena, has been playing an increasingly important role in inverse modeling. Remote sensing instruments measure quantities that often combine several state variables as one. This creates very strong correlations between the state variables that share the same observation variable. This may cause numerical problems resulting in a low convergence rate or inaccurate estimates in gradient-based variational assimilation if improper error statistics are used. In this paper, two criteria or scoring rules are proposed to quantify the numerical robustness of assimilating a specific set of remote sensing observations and to quantify the reliability of the estimates of the parameters. The criteria are derived by analyzing how the correlations are created via shared observation data and how they may influence the process of variational data assimilation. Experimental tests are conducted and show a good level of agreement with theo...
Research Interests:
Algorithms are often parallelized based on data dependence analysis manually or by means of parallel compilers. Some vector/matrix computations such as the matrix-vector products with simple data dependence structures (data parallelism)... more
Algorithms are often parallelized based on data dependence analysis manually or by means of parallel compilers. Some vector/matrix computations such as the matrix-vector products with simple data dependence structures (data parallelism) can be easily parallelized. For problems with more complicated data dependence structures, parallelization is less straightforward. The data dependence graph is a powerful means for designing and analyzing parallel algorithms. However, for sparse matrix computations, parallelization based on solely exploiting the existing parallelism in an algorithm does not always give satisfactory results. For example, the conventional Gaussian elimination algorithm for the solution of a tri-diagonal system is inherently sequential, so algorithms specially for parallel computation has to be designed. After briefly reviewing different parallelization approaches, a powerful graph formalism for designing parallel algorithms is introduced. This formalism will be discus...
Research Interests:
Research Interests:
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a... more
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a careful study on the characteristics of the ensemble ash state, we propose a mask-state algorithm which records the sparsity information of the full ensemble state matrix and transforms the full matrix into a relatively small one. This will reduce the computational cost in the analysis step. Experimental results show the mask-state algorithm significantly speeds up the expensive analysis step. Subsequently, the total amount of computing time for volcanic ash data assimilation is reduced to an acceptable level, which is important for providing timely and accurate aviation advices. The mask-state algorithm is generic and thus can be embedded in any ensemble-based data assimilation framework. Moreover, ensemble-based data assimilation with the mask-state a...
Research Interests: Computer Science and DAG
In this study, we investigate a strategy to accelerate the data assimilation (DA) algorithm. Based on evaluations of the computational time, the analysis step of the assimilation turns out to be the most expensive part. After a study of... more
In this study, we investigate a strategy to accelerate the data assimilation (DA) algorithm. Based on evaluations of the computational time, the analysis step of the assimilation turns out to be the most expensive part. After a study of the characteristics of the ensemble ash state, we propose a mask-state algorithm which records the sparsity information of the full ensemble state matrix and transforms the full matrix into a relatively small one. This will reduce the computational cost in the analysis step. Experimental results show the mask-state algorithm significantly speeds up the analysis step. Subsequently, the total amount of computing time for volcanic ash DA is reduced to an acceptable level. The mask-state algorithm is generic and thus can be embedded in any ensemble-based DA framework. Moreover, ensemble-based DA with the mask-state algorithm is promising and flexible, because it implements exactly the standard DA without any approximation and it realizes the satisfying p...
Research Interests:
Remote sensing, as a powerful tool for monitoring atmospheric phenomena, has been playing an increasingly important role in inverse modeling. Remote sensing instruments measure quantities that often combine several state variables as one.... more
Remote sensing, as a powerful tool for monitoring atmospheric phenomena, has been playing an increasingly important role in inverse modeling. Remote sensing instruments measure quantities that often combine several state variables as one. This creates very strong correlations between the state variables that share the same observation variable. This may cause numerical problems resulting in a low convergence rate or inaccurate estimates in gradient-based variational assimilation if improper error statistics are used. In this paper, two criteria or scoring rules are proposed to quantify the numerical robustness of assimilating a specific set of remote sensing observations and to quantify the reliability of the estimates of the parameters. The criteria are derived by analyzing how the correlations are created via shared observation data and how they may influence the process of variational data assimilation. Experimental tests are conducted and show a good level of agreement with theo...
Research Interests:
ABSTRACT
Research Interests:
Research Interests:
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a... more
In this study, we investigate strategies for accelerating data assimilation on volcanic ash forecasts. Based on evaluations of computational time, the analysis step of the assimilation is evaluated as the most expensive part. After a careful study on the characteristics of the ensemble ash state, we propose a mask-state algorithm which records the sparsity information of the full ensemble state matrix and transforms the full matrix into a relatively small one. This will reduce the computational cost in the analysis step. Experimental results show the mask-state algorithm significantly speeds up the expensive analysis step. Subsequently, the total amount of computing time for volcanic ash data assimilation is reduced to an acceptable level, which is important for providing timely and accurate aviation advices. The mask-state algorithm is generic and thus can be embedded in any ensemble-based data assimilation framework. Moreover, ensemble-based data assimilation with the mask-state a...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
ABSTRACT Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep... more
ABSTRACT Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep concentration gradients after the pollutant has just entered into the water, whereas conventional numerical methods such as finite difference and finite volume methods may have difficulties. Since the computation time in a particle model increases linearly with the number of particles, this often forms a limiting factor. We consider the parallelization of the particle model SIMPAR. Different load balancing and communication optimization possibilities are investigated. Some experiments with the parallel implementation of the SIMPAR model on a cluster of workstations and the Cray T3E are also reported.
Research Interests:
Research Interests:
Research Interests: Computer Science and DAG
ABSTRACT
Research Interests:
Research Interests: Computer Science and DAG
ABSTRACT
Research Interests:
Research Interests:
ABSTRACT Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep... more
ABSTRACT Particle methods are an important technique for simulating transport phenomena such as the transport of pollutants in coastal waters. Particle methods can quite accurately predict the pollutant transport in cases of steep concentration gradients after the pollutant has just entered into the water, whereas conventional numerical methods such as finite difference and finite volume methods may have difficulties. Since the computation time in a particle model increases linearly with the number of particles, this often forms a limiting factor. We consider the parallelization of the particle model SIMPAR. Different load balancing and communication optimization possibilities are investigated. Some experiments with the parallel implementation of the SIMPAR model on a cluster of workstations and the Cray T3E are also reported.
Research Interests:
Research Interests:
ABSTRACT
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Algorithms are often parallelized based on data dependence analysis manually or by means of parallel compilers. Some vector/matrix computations such as the matrix-vector products with simple data dependence structures (data parallelism)... more
Algorithms are often parallelized based on data dependence analysis manually or by means of parallel compilers. Some vector/matrix computations such as the matrix-vector products with simple data dependence structures (data parallelism) can be easily parallelized. For problems with more complicated data dependence structures, parallelization is less straightforward. The data dependence graph is a powerful means for designing and analyzing parallel
Research Interests:
Research Interests:
ABSTRACT This paper deals with the simulation of transport of pollutants in shallow water using random walk models and develops several computation techniques to speed up the numerical integration of the stochastic differential equations... more
ABSTRACT This paper deals with the simulation of transport of pollutants in shallow water using random walk models and develops several computation techniques to speed up the numerical integration of the stochastic differential equations (SDEs). This is achieved by using both random time stepping and parallel processing.We start by considering a basic stochastic Euler scheme for integration of the diffusion and drift terms of the SDEs, with a strong order 1 in the strong sense. The errors due to this scheme depend on the location of the pollutant; it is dominated by the diffusion term near boundaries, and by the deterministic drift further away from the boundaries. Using a pair of integration schemes, one of strong order 1.5 near the boundary and one of strong order 2.0 elsewhere, we can estimate the error and approximate an optimal step size for a given error tolerance. The resulting algorithm is developed such that it allows for complete flexibility of the step size, while guaranteeing the correct Brownian behaviour.Modelling pollutants by non-interacting particles enables the use of parallel processing in the simulation. We take advantage of this by implementing the algorithm using the MPI library. The inherent asynchronic nature of the particle simulation, in addition to the parallel processing, makes it difficult to get a coherent picture of the results at any given points. However, by inserting internal synchronisation points in the temporal discretisation, the code allows pollution snapshots and particle counts to be made at times specified by the user.
Research Interests: Applied Mathematics, Parallel Processing, Random Walk, Stochastic differential equation, Numerical Integration, and 7 moreTransportation Problem, Mathematical and Computer Modelling, Numerical Analysis and Computational Mathematics, Shallow Water, Variable Step Size, Euler scheme, and Computational Techniques
Research Interests:
ABSTRACT This paper deals with the simulation of transport of pollutants in shallow water using random walk models and develops several computation techniques to speed up the numerical integration of the stochastic differential equations... more
ABSTRACT This paper deals with the simulation of transport of pollutants in shallow water using random walk models and develops several computation techniques to speed up the numerical integration of the stochastic differential equations (SDEs). This is achieved by using both random time stepping and parallel processing.We start by considering a basic stochastic Euler scheme for integration of the diffusion and drift terms of the SDEs, with a strong order 1 in the strong sense. The errors due to this scheme depend on the location of the pollutant; it is dominated by the diffusion term near boundaries, and by the deterministic drift further away from the boundaries. Using a pair of integration schemes, one of strong order 1.5 near the boundary and one of strong order 2.0 elsewhere, we can estimate the error and approximate an optimal step size for a given error tolerance. The resulting algorithm is developed such that it allows for complete flexibility of the step size, while guaranteeing the correct Brownian behaviour.Modelling pollutants by non-interacting particles enables the use of parallel processing in the simulation. We take advantage of this by implementing the algorithm using the MPI library. The inherent asynchronic nature of the particle simulation, in addition to the parallel processing, makes it difficult to get a coherent picture of the results at any given points. However, by inserting internal synchronisation points in the temporal discretisation, the code allows pollution snapshots and particle counts to be made at times specified by the user.