Mario Giacobini

Hawks and Doves on small-world networks.

We explore the Hawk-Dove game on networks with topologies ranging from regular lattices to random graphs with small-world networks in between. This is done by means of computer simulations using several update rules for the population evolutionary dynamics. We find the overall result that cooperation is sometimes inhibited and sometimes enhanced in those network structures, with respect to the mixing population case. The differences are due to different update rules and depend on the gain-to-cost ratio. We analyze and qualitatively explain this behavior by using local topological arguments.

Selection intensity in cellular evolutionary algorithms for regular lattices

In this paper, we present quantitative models for the selection pressure of cellular evolutionary algorithms on regular one- and two-dimensional (2-D) lattices. We derive models based on probabilistic difference equations for synchronous and several asynchronous cell update policies. The models are validated using two customary selection methods: binary tournament and linear ranking. Theoretical results are in agreement with experimental values, showing that the selection intensity can be controlled by using different update methods. It is also seen that the usual logistic approximation breaks down for low-dimensional lattices and should be replaced by a polynomial approximation. The dependence of the models on the neighborhood radius is studied for both topologies. We also derive results for 2-D lattices with variable grid axes ratio.

Takeover time curves in random and small-world structured populations

We present discrete stochastic mathematical models for the growth curves of synchronous and synchronous evolutionary algorithms with populations structured ccording to a random graph. We show that, to good approximation, randomly structured and panmictic populations have the some growth behavior. Furthermore, we show that global selection intensity depends on the update policy. The validity of the models is confirmed by comparison with experimental results of simulations. We also present experimental results on small-world nd scale-free population graph topologies. We show that they lead to qualitatively similar results. However, the different nature of the nodes can be exploited to obtain more varied evolutionary behavior.

Effects of scale-free and small-world topologies on binary coded self-adaptive CEA

In this paper we investigate the properties of CEAs with populations structured as Watts–Strogatz small-world graphs and Albert–Barabasi scale-free graphs as problem solvers, using several standard discrete optimization problems as a benchmark. The EA variants employed include self-adaptation of mutation rates. Results are compared with the corresponding classical panmictic EA showing that topology together with self-adaptation drastically influences the search.

Selection intensity in asynchronous cellular evolutionary algorithms

This paper presents a theoretical study of the selection pressure in asynchronous cellular evolutionary algorithms (cEAs). This work is motivated by the search for a general model for asynchronous update of the individuals in a cellular EA, and by the necessity of better accuracy beyond what existing models of selection intensity can provide. Therefore, we investigate the differences between the expected and actual values of the selection pressure induced by several asynchronous update policies, and formally characterize the update dynamics of each variant of the algorithm. New models for these two issues are proposed, and are shown to be more accurate (lower fit error) than previous ones.

Modeling Selection Intensity for Toroidal Cellular Evolutionary Algorithms

We present quantitative models for the selection pressure on cellular evolutionary algorithms structured as a ring of cells. We obtain results for synchronous and asynchronous cell update policies. Theoretical results are in agreement with experimental values and show that the selection intensity can be controlled by using different update methods.

Comparing Synchronous and Asynchronous Cellular Genetic Algorithms

This paper presents a comparative study of several asynchronous policies for updating the population in a cellular genetic algorithm (cGA). Cellular GAs are regular GAs with the important exception that individuals are placed in a given geographical distribution (usually a 2-d grid). Operators are applied locally on a set made of each individual and the surrounding neighbors, thus promoting intra-neighborhood exploitation and inter-neighborhood exploration of the search space. Here, we analyze the respective advantages and drawbacks of dealing with this decentralized population in the traditional synchronous manner or in several possible asynchronous update policies. Asynchronous behavior has proven to be better in many domains such as cellular automata and distributed GAs, which, in turn, is also the main conclusion of this work. We will undergo a structured analysis on a set of problems with different features in order to get well grounded conclusions.

Spatio-temporal patterns of distribution of West Nile virus vectors in eastern Piedmont Region, Italy

BackgroundWest Nile Virus (WNV) transmission in Italy was first reported in 1998 as an equine outbreak near the swamps of Padule di Fucecchio, Tuscany. No other cases were identified during the following decade until 2008, when horse and human outbreaks were reported in Emilia Romagna, North Italy. Since then, WNV outbreaks have occurred annually, spreading from their initial northern foci throughout the country. Following the outbreak in 1998 the Italian public health authority defined a surveillance plan to detect WNV circulation in birds, horses and mosquitoes. By applying spatial statistical analysis (spatial point pattern analysis) and models (Bayesian GLMM models) to a longitudinal dataset on the abundance of the three putative WNV vectors [Ochlerotatus caspius (Pallas 1771), Culex pipiens (Linnaeus 1758) and Culex modestus (Ficalbi 1890)] in eastern Piedmont, we quantified their abundance and distribution in space and time and generated prediction maps outlining the areas with the highest vector productivity and potential for WNV introduction and amplification.ResultsThe highest abundance and significant spatial clusters of Oc. caspius and Cx. modestus were in proximity to rice fields, and for Cx. pipiens, in proximity to highly populated urban areas. The GLMM model showed the importance of weather conditions and environmental factors in predicting mosquito abundance. Distance from the preferential breeding sites and elevation were negatively associated with the number of collected mosquitoes. The Normalized Difference Vegetation Index (NDVI) was positively correlated with mosquito abundance in rice fields (Oc. caspius and Cx. modestus). Based on the best models, we developed prediction maps for the year 2010 outlining the areas where high abundance of vectors could favour the introduction and amplification of WNV.ConclusionsOur findings provide useful information for surveillance activities aiming to identify locations where the potential for WNV introduction and local transmission are highest. Such information can be used by vector control offices to stratify control interventions in areas prone to the invasion of WNV and other mosquito-transmitted pathogens.

Early warning of West Nile virus mosquito vector: climate and land use models successfully explain phenology and abundance of Culex pipiens mosquitoes in North-Western Italy

BackgroundWest Nile Virus (WNV) is an emerging global health threat. Transmission risk is strongly related to the abundance of mosquito vectors, typically Culex pipiens in Europe. Early-warning predictors of mosquito population dynamics would therefore help guide entomological surveillance and thereby facilitate early warnings of transmission risk.MethodsWe analysed an 11-year time series (2001 to 2011) of Cx. pipiens mosquito captures from the Piedmont region of north-western Italy to determine the principal drivers of mosquito population dynamics. Linear mixed models were implemented to examine the relationship between Cx. pipiens population dynamics and environmental predictors including temperature, precipitation, Normalized Difference Water Index (NDWI) and the proximity of mosquito traps to urban areas and rice fields.ResultsWarm temperatures early in the year were associated with an earlier start to the mosquito season and increased season length, and later in the year, with decreased abundance. Early precipitation delayed the start and shortened the length of the mosquito season, but increased total abundance. Conversely, precipitation later in the year was associated with a longer season. Finally, higher NDWI early in the year was associated with an earlier start to the season and increased season length, but was not associated with abundance. Proximity to rice fields predicted higher total abundance when included in some models, but was not a significant predictor of phenology. Proximity to urban areas was not a significant predictor in any of our models. Predicted variations in start of the season and season length ranged from one to three weeks, across the measured range of variables. Predicted mosquito abundance was highly variable, with numbers in excess of 1000 per trap per year when late season temperatures were low (average 21°C) to only 150 when late season temperatures were high (average 30°C).ConclusionsClimate data collected early in the year, in conjunction with local land use, can be used to provide early warning of both the timing and magnitude of mosquito outbreaks. This potentially allows targeted mosquito control measures to be implemented, with implications for prevention and control of West Nile Virus and other mosquito borne diseases.

Limiting the Number of Fitness Cases in Genetic Programming Using Statistics

Fitness evaluation is often a time consuming activity in genetic programming applications and it is thus of interest to find criteria that can help in reducing the time without compromising the quality of the results. We use well-known results in statistics and information theory to limit the number of fitness cases that are needed for reliable function reconstruction in genetic programming. By using two numerical examples, we show that the results agree with our theoretical predictions. Since our approach is problem-independent, it can be used together with techniques for choosing an efficient set of fitness cases.