This blog contains exploratory reports that I regularly make in my research. They are only made available for whoever wants to check and investigate the details of my analysis. For more constructed (but still detailed) reports, see my research reports.
Emapse
Emapse is a framework allowing the estimation of a random variable that results from several additions and maximum performed on a set of random variables.
Simulation code and project.
Modeling storage systems
Experiments and analysis:
- 2012-01-17: PVFS exploration with one server and one client
- 2012-02-28: PVFS analysis with at most two servers and two clients
- 2012-03-01: PVFS analysis with at most two servers and two clients 2
- 2012-04-01: HDFS exploration with one datanode and one client
- 2012-04-05: HDFS exploration with one datanode and one client with bug fixes
- 2012-04-06: HDFS exploration with one datanode and one client without stressing the system
- 2012-04-13: HDFS exploration with one datanode and one client with uniform sampling
- 2012-05-02: Analysis of a bioinformatic application resource requirements with 5 8-core nodes
- 2012-05-18: IP packet lengths with HDFS operations
- 2012-07-05: Analysis of a bioinformatic application resource requirements with 58 8-core nodes
Scheduling on desktop grids with uncertain availabilities
Analysis:
Reducing distributing elements with an associative operation
Experiments and analysis:
- 2012-03-02: Empirical approximation ratios of binomial and Fibonacci strategies
- 2013-03-26: Probabilistic simulation
- 2013-03-27: Probabilistic simulation with computation costs
- 2013-04-18: Probabilistic simulation plan
- 2013-05-30: Probabilistic simulation plan – confirmation
- 2013-05-30: Probabilistic simulation plan with more realistic distribution
- 2013-06-03: Final simulation for Heteropar article
- 2013-12-28: Initial attempt to determine optimal solutions for array reductions
- 2013-12-30: Optimal solutions for array reductions with any segment order
- 2014-01-24: Extension of Heteropar simulations with new distributions
- 2014-02-10: Merge of results with several distributions
- 2014-04-29: Data generation for CCPE
- 2014-04-29: Figure generation for CCPE
- 2014-04-30: Figure illustrating the distributions
- 2014-07-21: Data analysis for CCPE
Cost matrix generation
Experiments and analysis:
- 2013-04-10: Generation of cost matrices with parameterized correlations
- 2014-03-14: Generation of cost matrix with specific correlations and distributions
- 2014-03-30: Empirical Study of Siegel’s Method
- 2014-04-11: Aggregation method
- 2014-04-16: LPT with simplified unrelated instances: Siegel and corr
- 2014-05-06: Input distributions to avoid negative costs
- 2014-06-18: Additional algorithms for the unrelated case
- 2014-07-01: Correlation parameters in the literature
- 2014-09-11: Study heterogeneity quadrant
- 2014-09-14: Generating matrices of the heterogeneity quadrant
- 2014-10-01: Analysing analytically Siegel’s methods
- 2014-11-07: Summarizing usage of Siegel’s methods in the literature
- 2014-11-20: Summarizing Siegel’s methods with new measures
- 2014-11-21: Finishing Siegel’s methods analysis
- 2014-12-17: Analyse noise-based method
- 2015-01-10: Improving shuffling method and noise-based method analysis
- 2015-02-07: Relevance of asymptotic assumption
- 2015-03-12: Correlation analysis of existing methods
- 2015-03-12: Implementation and study of TMA
- 2015-03-18: Coefficient of variation of the singular/eigen values
- 2015-03-19: More robust generation methods with given correlation
- 2015-03-25: Generating cost matrices with given singular values
- 2015-03-26: Summarizing range-based and CVB methods with correlation
- 2015-04-01: Investigating singular value distribution effect
- 2015-05-19: Equivalence with polynomial problem
- 2015-07-10: Fixing noise-based method
- 2015-07-16: Large measure study
- 2015-07-23: Ideas for investigating and improving the TMA
- 2015-07-28: Large generation study
- 2015-11-05: Study of correlation conjecture
- 2015-12-15: Smoother adaptation of noise-based method for correlation
- 2015-12-21: Convergence of Markov process method
- 2016-01-01: Heterogeneity properties of correlation-aware generation methods
- 2016-02-10: Relation between the TMA and correlations in cost matrices
- 2016-05-12: Contingency table based MCMC
- 2016-05-31: Fixing the heat bath method
- 2017-06-19: Designing different matrix initialization methods
- 2017-06-20: Detecting convergence of MCMC
- 2017-06-27: Adaptative method to detect convergence
- 2017-12-04: Initial matrices with bounds
- 2018-02-12: Fixing initial matrix generation methods
- 2018-02-19: Showing limits of shuffling method
- 2018-03-02: Distribution of values in uniform vectors
- 2018-03-10: New methods to generate initial matrices
Simulation code and here.
Reproducibility of MPI_Reduce
- 2016-03-15: Open MPI versions study
- 2016-03-17: Open MPI parameters study
- 2016-03-19: Open MPI algorithm selection study
- 2016-03-23: MVAPICH2 default and tuned performance study
- 2016-04-11: Reproducing venkata2013a
- 2016-05-24: MVAPICH2 versions study
- 2016-06-30: Reproducing ma2012a
- 2016-07-21: Reproducing ma2012a on Grid5000
MVAPICH2 versions
The official website does not keep the previous versions of MVAPICH2. Here are some of them that were retrieved using RPM archives.
- mvapich2-0.9.8.tar.gz
- mvapich2-1.0.3.tar.gz
- mvapich2-1.2p1.tgz
- mvapich2-1.4.tgz
- mvapich2-1.6.tgz
- mvapich2-1.8a2.tgz
- mvapich2-1.9a.tgz
- mvapich2-2.0b.tgz
- mvapich2-2.0.tar.gz
- mvapich2-2.1.tar.gz
- mvapich2-2.2a.tar.gz
- mvapich2-2.2b.tar.gz
(Pm,Pk)||Cmax
- 2016-11-23: Preliminary empirical evaluation for (Pm,Pk)||Cmax
- 2016-12-01: Comparison with heuristics for R||Cmax
- 2017-01-07: Validation of reimplementation of balancing-based greedy algorithm
- 2017-01-10: Comparison with competing algorithms from the literature
- 2017-01-23: Euro-Par simulations
- 2017-02-14: Improving heuristics
- 2017-03-17: Testing heuristics with 3 types of processors
- 2017-03-28: Designing an algorithm improving the makespan by moving tasks in cycle
- 2017-03-30: Limiting the number of iterations
- 2017-05-25: Fixing bug in HeteroPrio implementation
- 2017-09-07: Cleaning BalancedEstimate
- 2017-09-14: Extending the Experimental Results
- 2018-01-19: Generating random graphs
- 2018-02-01: Denser random graphs
- 2018-02-06: Cholesky instances
- 2018-02-07: Cholesky with varying proc
- 2018-02-08: Using Kasahara instances
- 2018-02-14: Revisit random, Cholesky and Kasahara instances
Stochastic optimization in cloud context
- 2017-06-01: Testing evaluation methods for the makespan of a dynamic scheduling
- 2017-06-22: Testing methods for evaluating the makespan of a dynamic approach with heterogeneous costs
- 2018-04-18: Dynamic Programming with parallel machines
Chiroptères
- Données acoustiques et météorologiques à Laissey (août à octobre 2018):
- 2019-08-27: Étude données météorologiques à Besançon d’août à octobre 2018
- 2019-08-27: Étude données acoustiques chauves-souris à Laissey du 22/8/2018 au 30/10/2018
- 2019-11-03: Comparaison des données météorologiques à Besançon et à Montbéliard
- 2019-11-10: Heures d’activité des chauves-souris et vérification des données météorologiques à Laissey
- 2019-12-09: Lien entre activité et météo à Laissey du 22/8/2018 au 30/10/2018
- 2019-11-01: Exploration d’une route de vol à Gendrey