Andrés Tomás

Parallel Arnoldi eigensolvers with enhanced scalability via global communications rearrangement

This paper presents several new variants of the single-vector Arnoldi algorithm for computing approximations to eigenvalues and eigenvectors of a non-symmetric matrix. The context of this work is the efficient implementation of industrial-strength, parallel, sparse eigensolvers, in which robustness is of paramount importance, as well as efficiency. For this reason, Arnoldi variants that employ Gram-Schmidt with iterative reorthogonalization are considered. The proposed algorithms aim at improving the scalability when running in massively parallel platforms with many processors. The main goal is to reduce the performance penalty induced by global communications required in vector inner products and norms. In the proposed algorithms, this is achieved by reorganizing the stages that involve these operations, particularly the orthogonalization and normalization of vectors, in such a way that several global communications are grouped together while guaranteeing that the numerical stability of the process is maintained. The numerical properties of the new algorithms are assessed by means of a large set of test matrices. Also, scalability analyses show a significant improvement in parallel performance.

Evaluation of several variants of explicitly restarted lanczos eigensolvers and their parallel implementations

It is well known that the Lanczos process suffers from loss of orthogonality in the case of finite-precision arithmetic. Several approaches have been proposed in order to address this issue, thus enabling the successful computation of approximate eigensolutions. However, these techniques have been studied mainly in the context of long Lanczos runs, but not for restarted Lanczos eigensolvers. Several variants of the explicitly restarted Lanczos algorithm employing different reorthogonalization strategies have been implemented in SLEPc, the Scalable Library for Eigenvalue Computations. The aim of this work is to assess the numerical robustness of the proposed implementations as well as to study the impact of reorthogonalization in parallel efficiency.

Friendly-Sharing: Improving the Performance of City Sensoring through Contact-Based Messaging Applications

Regular citizens equipped with smart devices are being increasingly used as “sensors” by Smart Cities applications. Using contacts among users, data in the form of messages is obtained and shared. Contact-based messaging applications are based on establishing a short-range communication directly between mobile devices, and on storing the messages in these devices for subsequent delivery to cloud-based services. An effective way to increase the number of messages that can be shared is to increase the contact duration. We thus introduce the Friendly-Sharing diffusion approach, where, during a contact, the users are aware of the time needed to interchange the messages stored in their buffers, and they can thus decide to wait more time in order to increase the message sharing probability. The performance of this approach is anyway closely related to the size of the buffer in the device. We therefore compare various policies either for the message selection at forwarding times and for message dropping when the buffer is full. We evaluate our proposal with a modified version of the Opportunistic Networking Environment (ONE) simulator and using real human mobility traces.

Evaluating the use of sub-gigahertz wireless technologies to improve message delivery in opportunistic networks

The message delivery ratio of mobile opportunistic networks strongly depends on the transmission time, which is closely related either to the mobility of users and to the communication properties of the mobile devices. A larger radio transmission range allows longer contact durations, improving the message dissemination. Furthermore, user mobility is a crucial factor to be considered, especially when the mobile nodes are vehicles, because of their limited freedom of movement and the high relative speed. In this paper, we evaluate the use of a sub-gigahertz wireless technology, namely LoRa (Long Range), to establish links between the mobile users in an opportunistic network in order to augment the number of contacts and their duration. We evaluate the performance of LoRa, comparing it with WiFi, using the Epidemic protocol for message diffusion with realistic vehicular traces. Through simulations, we compare the message delivery probability and the network overhead. These experiments were carried out using the ONE simulator with minor modifications to model the typical behaviour of mobile users. The results show that, in opportunistic networks, increasing the range even while reducing the available bandwidth increases the message delivery ratio.

MuffinEc: Error correction for de Novo assembly via greedy partitioning and sequence alignment

Abstract Error correction is typically the first step of de Novo genome assembly from NGS data. This step has an important impact on the quality and speed of the assembly process. However, the majority of available stand-alone error correction solutions can only detect and correct mismatches. Therefore, these solutions only support correcting reads generated by Illumina sequencers. Several solutions support insertions and deletions (indels) and are capable of working with multiple technologies. However, these solutions are limited by correction performance and resource consumption. In this paper, we introduce MuffinEc, an indel-aware multi-technology correction method for NGS data. This method uses a greedy approach to create groups of reads and subsequently corrects them using their consensus. MuffinEc surpasses existing solutions by offering better correction ratios for multiple technologies. This method also exploits parallel processing via OpenMP and uses less computational resources than similar programs, thereby being capable of handling large datasets. MuffinEc is open source and freely available at http://muffinec.sourceforge.net .

Improving Message Delivery Performance in Opportunistic Networks Using a Forced-Stop Diffusion Scheme

The performance of mobile opportunistic networks strongly depends on contact duration. If the contact lasts less than the required transmission times, some messages will not get delivered, and the whole diffusion scheme will be seriously affected.

Mobility as the Main Enabler of Opportunistic Data Dissemination in Urban Scenarios

The use of opportunistic communications to disseminate common interest messages in an urban scenario have various applications, like sharing traffic status, advertising shop offers, spread alarms, and so on. In this paper, we evaluate the combined use of fixed and mobile nodes to establish an optimal urban opportunistic network aimed at the distribution of general interest data.

Empirical Study and Modeling of Vehicular Communications at Intersections in the 5 GHz Band

Event warnings are critical in the context of ITS, being dependent on reliable and low-delay delivery of messages to nearby vehicles. One of the main challenges to address in this context is intersection management. Since buildings will severely hinder signals in the 5 GHz band, it becomes necessary to transmit at the exact moment a vehicle is at the center of an intersection to maximize delivery chances. However, GPS inaccuracy, among other problems, complicates the achievement of this goal. In this paper we study this problem by first analyzing different intersection types, studying the vehicular communications performance in each type of intersection through real scenario experiments. Obtained results show that intersection-related communications depend on the distances to the intersection and line-of-sight (LOS) conditions. Also, depending on the physical characteristics of intersections, the presented blockages introduce different degrees of hampering to message delivery. Based on the modeling of the different intersection types, we then study the expected success ratio when notifying events at intersections. In general, we find that effective propagation of messages at intersections is possible, even in urban canyons and despite GPS errors, as long as rooftop antennas are used to compensate for poor communication conditions.

Evaluating the Impact of Data Transfer Time and Mobility Patterns in Opportunistic Networks

The Epidemic protocol is an effective way to achieve information diffusion in opportunistic networks. Its performance depends on two key factors: the device mobility pattern,, the message transmission time. The mobility pattern determines the contact time, duration. If contact durations are shorter than the required transmission times, some messages will not get delivered,, the whole diffusion scheme will be seriously hampered. In this paper we evaluate the impact of message transmission times in epidemic diffusion processes. We demonstrate how, when certain conditions hold, forcing devices to stop moving to complete the data delivery process can improve their performance. We implemented this mobility model, called Forced Stop, in the ONE (Opportunistic Networking Environment) simulator,, we show that, for large message sizes, the diffusion performance is increased. These results can be a relevant indication to the designers of opportunistic networks applications that could integrate in their products strategies to inform the user about the need to temporarily stop to increase the overall data delivery.

Fast inexact mapping using advanced tree exploration on backward search methods.

BackgroundShort sequence mapping methods for Next Generation Sequencing consist on a combination of seeding techniques followed by local alignment based on dynamic programming approaches. Most seeding algorithms are based on backward search alignment, using the Burrows Wheeler Transform, the Ferragina and Manzini Index or Suffix Arrays. All these backward search algorithms have excellent performance, but their computational cost highly increases when allowing errors. In this paper, we discuss an inexact mapping algorithm based on pruning strategies for search tree exploration over genomic data.ResultsThe proposed algorithm achieves a 13x speed-up over similar algorithms when allowing 6 base errors, including insertions, deletions and mismatches. This algorithm can deal with 400 bps reads with up to 9 errors in a high quality Illumina dataset. In this example, the algorithm works as a preprocessor that reduces by 55% the number of reads to be aligned. Depending on the aligner the overall execution time is reduced between 20–40%.ConclusionsAlthough not intended as a complete sequence mapping tool, the proposed algorithm could be used as a preprocessing step to modern sequence mappers. This step significantly reduces the number reads to be aligned, accelerating overall alignment time. Furthermore, this algorithm could be used for accelerating the seeding step of already available sequence mappers. In addition, an out-of-core index has been implemented for working with large genomes on systems without expensive memory configurations.