ABSTRACT. Hybrid deterministic-stochastic methods present an efficient alternative to a purely stochastic treatment of biochemical models. Traditionally, those methods split biochemical reaction networks into one set of slow reactions that is computed stochastically and one set of fast reactions that is computed deterministically. Applying those methods to multi-level models with dynamic nestings requires to cope with dynamic reaction networks changing over time. In addition, in case of large populations of nested entities, stochastic events can still decrease the runtime performance significantly, as reactions of dynamically nested entities are inherently stochastic. In this paper, we present a hybrid deterministic-stochastic simulation algorithm for multi-level models exploiting an adaptive approximation control and applying an additional approximation by executing multiple independent stochastic events simultaneously in one simulation step. The algorithm has been realized in the rule-based multi-level modeling language ML-Rules. Its impact on speed and accuracy is evaluated based on simulations performed with a model of Dictyostelium discoideum amoebas.

ABSTRACT. The Functional Mock-up Interface (FMI) standard is becoming an essential solution for co-simulation. In this paper, we address a specific issue which arises in the context of Distributed Cyber-Physical System (DCPS) co-simulation where Functional Mock-up Units (FMU) need to interact with distributed application models. The core of the problem is that, in general, complex distributed application behaviors cannot be easily and accurately captured by a modeling formalism but are instead directly specified using a standard programming language. As a consequence, the model of a distributed application is itself a distributed program. The challenge is then to bridge the gap between this programmatic description and the equation-based framework of FMI in order to make FMUs interact with distributed programs. In this article, we show how we use the unique model of execution of the SimGrid simulation platform to tackle this issue. The platform manages the co-evolution and the interaction between IT models and the different concurrent processes which compose a distributed application code. Thus, SimGrid offers a framework to mix models and distributed programs. We show then how we specify an FMU as a SimGrid model in order to solve the DCPS co-simulation issues. Compared to other works of the literature, our solution is not limited to a specific use case and benefits from the versatility and scalability of SimGrid.

ABSTRACT. This paper explores the improvement of non-iterative co-simulation master. A simple hybrid system is depicted and analyzed. Based on this analysis guidelines for calculating the calling sequence are introduced. Guidelines allow the implementation of a new constraint programming algorithm. This algorithm allows better calling sequence selection based solely on information about connecting the co-simulation network. The algorithm is conrmed by the example of co-simulation of a hybrid electric vehicle. In this example, the constraint programming algorithm found a subjectively good calling sequence without any involvement of the model developer.

ABSTRACT. Reaction selection is a major and time consuming step of stochastic simulation algorithms. Current approaches focus on constant sets of reactions. However, in the case of multiple agents whose behaviors are governed by diverse reactions at multiple levels, with the number of agents the number of reactions varies during simulation. Therefore, we equip different variants of stochastic simulation algorithms with strategies to handle dynamic sets of reactions. We implement the next reaction method with a heap and the direct reaction method with two tree-based selection strategies, compare their performance, and discuss open questions for future research.

ABSTRACT. tbd

ABSTRACT. Many runs of a computer simulation are needed to model uncertainty and evaluate alternate design choices. Such an ensemble of runs often contains many commonalities among the different individual runs. Simulation cloning is a technique that capitalizes on this fact to reduce the amount of computation required by the ensemble. Granular cloning is proposed that allows the sharing of state and computations at the scale of simulation objects as small as individual variables, offering savings in computation and memory, increased parallelism and improved tractability of sample path patterns across multiple runs. The ensemble produces results that are identical to separately executed runs. Whenever simulation objects interact, granular cloning will resolve their association to subsets of runs though binary operations on tags. Algorithms and computational techniques required to efficiently implement granular cloning are presented. Results from an experimental study using a cellular automata-based transportation simulation model and a coupled transportation and land use model are presented providing evidence the approach can yield significant speed ups relative to brute force replicated runs.

ABSTRACT. Parallel Discrete Event Simulations (PDES) running at large scales involve the coordination of billions of very fine grain events distributed across a large number of processes. At such large scales optimistic synchronization protocols, such as TimeWarp, allow for a high degree of parallelism between processes, but with the additional complexity of managing event rollback and cancellation. This can become especially problematic in models that exhibit imbalance resulting in low event efficiency, which increases the total amount of work required to run a simulation to completion. Managing this complexity becomes key to achieving a high degree of performance across a wide range of models. In this paper, we address this issue by analyzing the relationship between synchronization cost and event efficiency. We first look at how these two characteristics are coupled via the computation of Global Virtual Time (GVT). We then introduce dynamic load balancing, and show how, when combined with low overhead GVT computation, we can achieve higher efficiency with less synchronization cost. In doing so, we achieve up to 2× better performance on a variety of benchmarks and models of practical importance.

ABSTRACT. We present a fine-grained load migration protocol intended for parallel discrete event simulation of spatially extended models. Typical models have domains that are fine-grained discretizations of some volume, e.g., a cell, using an irregular three-dimensional mesh, where most events span several subvolumes. Phenomena of interest in e.g., cellular biology, are often non-homogeneous and migrate over the simulated domain, making load-balancing a crucial part of a successful PDES. Our load migration protocol is local in the sense that it involves only those processors that exchange workload, and does not affect the running parallel simulation. We present a detailed description of the protocol and a thorough proof for its correctness. We combine our protocol with a strategy for deciding when and what load to migrate, which optimizes both for load balancing and inter-processor communication using tunable parameters. Our evaluation shows that the overhead of the load migration protocol is negligible, and that it significantly reduces the number of rollbacks caused by load imbalance.