Convergence of classical parallel iterations is detected by performing a reduction operation at each iteration in order to compute a residual error relative to a potential solution vector. To efficiently run asynchronous iterations, blocking communication requests are avoided, which makes it hard to isolate and handle any global vector. While some termination protocols were proposed for asynchronous iterations, only very few of them are based on global residual computation and guarantee effective convergence. But the most effective and efficient existing solutions feature two reduction operations, which constitutes an important factor of termination delay. In this paper, we present new, non-intrusive, protocols to compute a residual error under asynchronous iterations, requiring only one reduction operation. Various communication models show that some heuristics can even be introduced and formally evaluated. Extensive experiments with up to 5600 processor cores confirm the practical effectiveness and efficiency of our approach.

No abstract

The gut-brain axis may play a central role in the pathogenesis of neurological disorders. Dozens of case-control studies have been carried out to identify bacterial markers by the use of targeted metagenomics. Alterations of several taxonomic profiles have been confirmed across several populations, however no consensus has been made regarding alpha-diversity. A recent publication has described and validated a novel method based on richness and evenness measures of the gut microbiome in order to reduce the complexity and multiplicity of alpha-diversity indices. We used these recently described richness and evenness composite measures to investigate the potential link between gut microbiome alpha-diversity and neurological disorders and to determine to what extent it could be used as a marker to diagnose neurological disorders from stool samples. We performed an exhaustive review of the literature to identify original published clinical studies including 16S rRNA gene sequencing on Parkinson’s disease, multiple Sclerosis and Alzheimer’s disease. Richness and evenness factors loadings were quantified from sequencing files in addition with the Shannon diversity index. For each disease, we performed a meta-analysis comparing the indices between patients and healthy controls. Seven studies were meta-analyzed for Parkinson’s disease, corresponding to 1067 subjects (631 Parkinson’s Disease/436 healthy controls). Five studies were meta-analyzed for multiple sclerosis, corresponding to 303 subjects (164 Multiple Sclerosis/139 healthy controls). For Alzheimer’s disease, the meta-analysis was not done as only two studies matched our criteria. Neither richness nor evenness was significantly altered in Parkinson’s disease and multiple sclerosis patients in comparison to healthy controls (p-value > 0.05). Shannon index was neither associated to neurological disorders (p-value > 0.05). After adjusting for age and sex, none of the alpha-diversity measures were associated with Parkinson’s Disease. This is the first report investigating systematically alpha-diversity and its potential link to neurological disorders. Our study has demonstrated that unlike in other gastro-intestinal, immune and metabolic disorders, loss of bacterial diversity is not associated with Parkinson’s disease and multiple sclerosis.