Synergistic Execution of Stream Programs on Multicores with Accelerators

The StreamIt programming model has been proposed to exploit parallelism in streaming applications oil general purpose multicore architectures. The StreamIt graphs describe task, data and pipeline parallelism which can be exploited on accelerators such as Graphics Processing Units (GPUs) or CellBE which support abundant parallelism in hardware. In this paper, we describe a novel method to orchestrate the execution of if StreamIt program oil a multicore platform equipped with an accelerator. The proposed approach identifies, using profiling, the relative benefits of executing a task oil the superscalar CPU cores and the accelerator. We formulate the problem of partitioning the work between the CPU cores and the GPU, taking into account the latencies for data transfers and the required buffer layout transformations associated with the partitioning, as all integrated Integer Linear Program (ILP) which can then be solved by an ILP solver. We also propose an efficient heuristic algorithm for the work-partitioning between the CPU and the GPU, which provides solutions which are within 9.05% of the optimal solution on an average across the benchmark Suite. The partitioned tasks are then software pipelined to execute oil the multiple CPU cores and the Streaming Multiprocessors (SMs) of the GPU. The software pipelining algorithm orchestrates the execution between CPU cores and the GPU by emitting the code for the CPU and the GPU, and the code for the required data transfers. Our experiments on a platform with 8 CPU cores and a GeForce 8800 GTS 512 GPU show a geometric mean speedup of 6.94X with it maximum of 51.96X over it single threaded CPU execution across the StreamIt benchmarks. This is a 18.9% improvement over it partitioning strategy that maps only the filters that cannot be executed oil the GPU - the filters with state that is persistent across firings - onto the CPU.

Application of Mathematical Programming to the Plywood Design and Manufacturing Problem

Plywood manufacture includes two fundamental stages. The first is to peel or separate logs into veneer sheets of different thicknesses. The second is to assemble veneer sheets into finished plywood products. At the first stage a decision must be made as to the number of different veneer thicknesses to be peeled and what these thicknesses should be. At the second stage, choices must be made as to how these veneers will be assembled into final products to meet certain constraints while minimizing wood loss. These decisions present a fundamental management dilemma. Costs of peeling, drying, storage, handling, etc. can be reduced by decreasing the number of veneer thicknesses peeled. However, a reduced set of thickness options may make it infeasible to produce the variety of products demanded by the market or increase wood loss by requiring less efficient selection of thicknesses for assembly. In this paper the joint problem of veneer choice and plywood construction is formulated as a nonlinear integer programming problem. A relatively simple optimal solution procedure is developed that exploits special problem structure. This procedure is examined on data from a British Columbia plywood mill. Restricted to the existing set of veneer thicknesses and plywood designs used by that mill, the procedure generated a solution that reduced wood loss by 79 percent, thereby increasing net revenue by 6.86 percent. Additional experiments were performed that examined the consequences of changing the number of veneer thicknesses used. Extensions are discussed that permit the consideration of more than one wood species.

Suomen maatalouden tuottamat kasvihuonekaasujen päästöt eri politiikkaskenaarioissa

Climate change is the single biggest environmental problem in the world at the moment. Although the effects are still not fully understood and there is considerable amount of uncertainty, many na-tions have decided to mitigate the change. On the societal level, a planner who tries to find an eco-nomically optimal solution to an environmental pollution problem seeks to reduce pollution from the sources where reductions are most cost-effective. This study aims to find out how effective the instruments of the agricultural policy are in the case of climate change mitigation in Finland. The theoretical base of this study is the neoclassical economic theory that is based on the assumption of a rational economic agent who maximizes his own utility. This theoretical base has been widened towards the direction clearly essential to the matter: the theory of environmental eco-nomics. Deeply relevant to this problem and central in the theory of environmental economics are the concepts of externalities and public goods. What are also relevant are the problems of global pollution and non-point-source pollution. Econometric modelling was the method that was applied to this study. The Finnish part of the AGMEMOD-model, covering the whole EU, was used for the estimation of the development of pollution. This model is a seemingly recursive, partially dynamic partial-equilibrium model that was constructed to predict the development of Finnish agricultural production of the most important products. For the study, I personally updated the model and also widened its scope in some relevant matters. Also, I devised a table that can calculate the emissions of greenhouse gases according to the rules set by the IPCC. With the model I investigated five alternative scenarios in comparison to the base-line scenario of Agenda 2000 agricultural policy. The alternative scenarios were: 1) the CAP reform of 2003, 2) free trade on agricultural commodities, 3) technological change, 4) banning the cultivation of organic soils and 5) the combination of the last three scenarios as the maximal achievement in reduction. The maximal achievement in the alternative scenario 5 was 1/3 of the level achieved on the base-line scenario. CAP reform caused only a minor reduction when com-pared to the base-line scenario. Instead, the free trade scenario and the scenario of technological change alone caused a significant reduction. The biggest single reduction was achieved by banning the cultivation of organic land. However, this was also the most questionable scenario to be real-ized, the reasons for this are further elaborated in the paper. The maximal reduction that can be achieved in the Finnish agricultural sector is about 11 % of the emission reduction that is needed to comply with the Kyoto protocol.

Problems and Algorithms for Sequence Segmentations

The analysis of sequential data is required in many diverse areas such as telecommunications, stock market analysis, and bioinformatics. A basic problem related to the analysis of sequential data is the sequence segmentation problem. A sequence segmentation is a partition of the sequence into a number of non-overlapping segments that cover all data points, such that each segment is as homogeneous as possible. This problem can be solved optimally using a standard dynamic programming algorithm. In the first part of the thesis, we present a new approximation algorithm for the sequence segmentation problem. This algorithm has smaller running time than the optimal dynamic programming algorithm, while it has bounded approximation ratio. The basic idea is to divide the input sequence into subsequences, solve the problem optimally in each subsequence, and then appropriately combine the solutions to the subproblems into one final solution. In the second part of the thesis, we study alternative segmentation models that are devised to better fit the data. More specifically, we focus on clustered segmentations and segmentations with rearrangements. While in the standard segmentation of a multidimensional sequence all dimensions share the same segment boundaries, in a clustered segmentation the multidimensional sequence is segmented in such a way that dimensions are allowed to form clusters. Each cluster of dimensions is then segmented separately. We formally define the problem of clustered segmentations and we experimentally show that segmenting sequences using this segmentation model, leads to solutions with smaller error for the same model cost. Segmentation with rearrangements is a novel variation to the segmentation problem: in addition to partitioning the sequence we also seek to apply a limited amount of reordering, so that the overall representation error is minimized. We formulate the problem of segmentation with rearrangements and we show that it is an NP-hard problem to solve or even to approximate. We devise effective algorithms for the proposed problem, combining ideas from dynamic programming and outlier detection algorithms in sequences. In the final part of the thesis, we discuss the problem of aggregating results of segmentation algorithms on the same set of data points. In this case, we are interested in producing a partitioning of the data that agrees as much as possible with the input partitions. We show that this problem can be solved optimally in polynomial time using dynamic programming. Furthermore, we show that not all data points are candidates for segment boundaries in the optimal solution.

Application of the Conjugate Gradient Method to a Problem on Minimum Time Orbit Transfer

This correspondence considers the problem of optimally controlling the thrust steering angle of an ion-propelled spaceship so as to effect a minimum time coplanar orbit transfer from the mean orbital distance of Earth to mean Martian and Venusian orbital distances. This problem has been modelled as a free terminal time-optimal control problem with unbounded control variable and with state variable equality constraints at the final time. The problem has been solved by the penalty function approach, using the conjugate gradient algorithm. In general, the optimal solution shows a significant departure from earlier work. In particular, the optimal control in the case of Earth-Mars orbit transfer, during the initial phase of the spaceship's flight, is found to be negative, resulting in the motion of the spaceship within the Earth's orbit for a significant fraction of the total optimized orbit transfer time. Such a feature exhibited by the optimal solution has not been reported at all by earlier investigators of this problem.

Convergence analysis of the Newton algorithm and a pseudo-time marching scheme for diffuse correlation tomography

We propose a self-regularized pseudo-time marching scheme to solve the ill-posed, nonlinear inverse problem associated with diffuse propagation of coherent light in a tissuelike object. In particular, in the context of diffuse correlation tomography (DCT), we consider the recovery of mechanical property distributions from partial and noisy boundary measurements of light intensity autocorrelation. We prove the existence of a minimizer for the Newton algorithm after establishing the existence of weak solutions for the forward equation of light amplitude autocorrelation and its Frechet derivative and adjoint. The asymptotic stability of the solution of the ordinary differential equation obtained through the introduction of the pseudo-time is also analyzed. We show that the asymptotic solution obtained through the pseudo-time marching converges to that optimal solution provided the Hessian of the forward equation is positive definite in the neighborhood of optimal solution. The superior noise tolerance and regularization-insensitive nature of pseudo-dynamic strategy are proved through numerical simulations in the context of both DCT and diffuse optical tomography. (C) 2010 Optical Society of America.

Interactive Distributed Source Coding in Asymmetric Communication Scenarios

We provide a new unified framework, called "multiple correlated informants - single recipient" communication, to address the variations of the traditional Distributed Source Coding (DSC) problem. Different combinations of the assumptions about the communication scenarios and the objectives of communication result in different variations of the DSC problem. For each of these variations, the complexities of communication and computation of the optimal solution is determined by the combination of the underlying assumptions. In the proposed framework, we address the asymmetric, interactive, and lossless variant of the DSC problem, with various objectives of communication and provide optimal solutions for those. Also, we consider both, the worst-case and average-case scenarios.

A genetic algorithm with characteristic amplification through multiple geographically isolated populations and varied fitness landscapes

This paper proposes a new approach, wherein multiple populations are evolved on different landscapes. The problem statement is broken down, to describe discrete characteristics. Each landscape, described by its fitness landscape is used to optimize or amplify a certain characteristic or set of characteristics. Individuals from each of these populations are kept geographically isolated from each other Each population is evolved individually. After a predetermined number of evolutions, the system of populations is analysed against a normalized fitness function. Depending on this score and a predefined merging scheme, the populations are merged, one at a time, while continuing evolution. Merging continues until only one final population remains. This population is then evolved, following which the resulting population will contain the optimal solution. The final resulting population will contain individuals which have been optimized against all characteristics as desired by the problem statement. Each individual population is optimized for a local maxima. Thus when populations are merged, the effect is to produce a new population which is closer to the global maxima.

A genetic algorithm with characteristic amplification through multiple geographically isolated populations and varied fitness landscapes

This paper proposes a new approach, wherein multiple populations are evolved on different landscapes. The problem statement is broken down, to describe discrete characteristics. Each landscape, described by its fitness landscape is used to optimize or amplify a certain characteristic or set of characteristics. Individuals from each of these populations are kept geographically isolated from each other Each population is evolved individually. After a predetermined number of evolutions, the system of populations is analysed against a normalized fitness function. Depending on this score and a predefined merging scheme, the populations are merged, one at a time, while continuing evolution. Merging continues until only one final population remains. This population is then evolved, following which the resulting population will contain the optimal solution. The final resulting population will contain individuals which have been optimized against all characteristics as desired by the problem statement. Each individual population is optimized for a local maxima. Thus when populations are merged, the effect is to produce a new population which is closer to the global maxima.

Minimizing total weighted tardiness on a batch-processing machine with non-agreeable release times and due dates

This study considers the scheduling problem observed in the burn-in operation of semiconductor final testing, where jobs are associated with release times, due dates, processing times, sizes, and non-agreeable release times and due dates. The burn-in oven is modeled as a batch-processing machine which can process a batch of several jobs as long as the total sizes of the jobs do not exceed the machine capacity and the processing time of a batch is equal to the longest time among all the jobs in the batch. Due to the importance of on-time delivery in semiconductor manufacturing, the objective measure of this problem is to minimize total weighted tardiness. We have formulated the scheduling problem into an integer linear programming model and empirically show its computational intractability. Due to the computational intractability, we propose a few simple greedy heuristic algorithms and meta-heuristic algorithm, simulated annealing (SA). A series of computational experiments are conducted to evaluate the performance of the proposed heuristic algorithms in comparison with exact solution on various small-size problem instances and in comparison with estimated optimal solution on various real-life large size problem instances. The computational results show that the SA algorithm, with initial solution obtained using our own proposed greedy heuristic algorithm, consistently finds a robust solution in a reasonable amount of computation time.

Minimum enclosing spheres formulations for support vector ordinal regression

We present two new support vector approaches for ordinal regression. These approaches find the concentric spheres with minimum volume that contain most of the training samples. Both approaches guarantee that the radii of the spheres are properly ordered at the optimal solution. The size of the optimization problem is linear in the number of training samples. The popular SMO algorithm is adapted to solve the resulting optimization problem. Numerical experiments on some real-world data sets verify the usefulness of our approaches for data mining.

Joint Power Control, Scheduling and Routing for Multihop Energy Harvesting Sensor Networks

We study wireless multihop energy harvesting sensor networks employed for random field estimation. The sensors sense the random field and generate data that is to be sent to a fusion node for estimation. Each sensor has an energy harvesting source and can operate in two modes: Wake and Sleep. We consider the problem of obtaining jointly optimal power control, routing and scheduling policies that ensure a fair utilization of network resources. This problem has a high computational complexity. Therefore, we develop a computationally efficient suboptimal approach to obtain good solutions to this problem. We study the optimal solution and performance of the suboptimal approach through some numerical examples.

A Novel, Balanced, and Energy-Efficient Training Method for Receive Antenna Selection

In receive antenna selection (AS), only signals from a subset of the antennas are processed at any time by the limited number of radio frequency (RF) chains available at the receiver. Hence, the transmitter needs to send pilots multiple times to enable the receiver to estimate the channel state of all the antennas and select the best subset. Conventionally, the sensitivity of coherent reception to channel estimation errors has been tackled by boosting the energy allocated to all pilots to ensure accurate channel estimates for all antennas. Energy for pilots received by unselected antennas is mostly wasted, especially since the selection process is robust to estimation errors. In this paper, we propose a novel training method uniquely tailored for AS that transmits one extra pilot symbol that generates accurate channel estimates for the antenna subset that actually receives data. Consequently, the transmitter can selectively boost the energy allocated to the extra pilot. We derive closed-form expressions for the proposed scheme's symbol error probability for MPSK and MQAM, and optimize the energy allocated to pilot and data symbols. Through an insightful asymptotic analysis, we show that the optimal solution achieves full diversity and is better than the conventional method.

Heuristic algorithms for scheduling of a batch processor in automobile gear manufacturing

In this paper we address a scheduling problem for minimising total weighted tardiness. The motivation for the paper comes from the automobile gear manufacturing process. We consider the bottleneck operation of heat treatment stage of gear manufacturing. Real life scenarios like unequal release times, incompatible job families, non-identical job sizes and allowance for job splitting have been considered. A mathematical model taking into account dynamic starting conditions has been developed. Due to the NP-hard nature of the problem, a few heuristic algorithms have been proposed. The performance of the proposed heuristic algorithms is evaluated: (a) in comparison with optimal solution for small size problem instances, and (b) in comparison with `estimated optimal solution' for large size problem instances. Extensive computational analyses reveal that the proposed heuristic algorithms are capable of consistently obtaining near-optimal solutions (that is, statistically estimated one) in very reasonable computational time.

Rectilinear Path Following in 3D Space

This paper addresses the problem of determining an optimal (shortest) path in three dimensional space for a constant speed and turn-rate constrained aerial vehicle, that would enable the vehicle to converge to a rectilinear path, starting from any arbitrary initial position and orientation. Based on 3D geometry, we propose an optimal and also a suboptimal path planning approach. Unlike the existing numerical methods which are computationally intensive, this optimal geometrical method generates an optimal solution in lesser time. The suboptimal solution approach is comparatively more efficient and gives a solution that is very close to the optimal one. Due to its simplicity and low computational requirements this approach can be implemented on an aerial vehicle with constrained turn radius to reach a straight line with a prescribed orientation as required in several applications. But, if the distance between the initial point and the straight line to be followed along the vertical axis is high, then the generated path may not be flyable for an aerial vehicle with limited range of flight path angle and we resort to a numerical method for obtaining the optimal solution. The numerical method used here for simulation is based on multiple shooting and is found to be comparatively more efficient than other methods for solving such two point boundary value problem.