UQ Training Course

Hartree Centre, Daresbury Laboratory, Warrington, UK

4th - 6th May 2016

In collaboration with the Institute for Risk and Uncertainty and Hartree Centre, we are offering a 3 days training course on Uncertainty Quantification using COSSAN Software on High Performance Computing.

Structure of the training programme

Each day focuses on a specific topic. This allows the partecipation to attend a specific training day. The first day is dedicated to an introduction of general concepts of stochastic and probabilistic analysis as well as an introduction to High Performance Computing with practical exercises. The second day is dedicated to COSSAN-X software while the last day will concentrate on the OpenCossan software.

Aims and Learning outcomes

You will learn the main techniques available for dealing with Risk and Uncertainty and how to use High Performance Computing to speed up the analysis.

Main Concepts and techniques

  • Random Variables and Random Variables Sets
  • Monte Carlo simulation and advanced simulation techiniques (Subset simulation, Line Sampling, Importance Sampling, Latin Hypercube Sampling)
  • Global and Local Sensitivity analysis
  • Global optimization techniques
  • Surrogate Models (Artificial Neural Networks, Response surface, Kriging)
  • Reliability based and robust desing
  • Programme

    Day 1 - 4th May 2016

    Time Activity Type Description
    10:00 Intruduction Welcome, aim and structure of the course. Getting started: basic linux commands and remote connection
    10:30 Lecture Introduction to Uncertainty Quantification and modelling of the uncertainties (random variables, random variable sets, stochastic processes and Gaussian mixture distributions)
    11:00 Lecture UQ tools and Monte Carlo simulation and advanced Monte Carlo methods (Importance sampling, Line Sampling, Subset Simulation) for UQ and reliability analysis.
    11:30 Lecture Meta-models (Response surface, Artificial Neural Networks, Polyharmonic lines), Optimisation (Gradient based methods and stochastic optimisations) and Robust/Reliability Design optimisation
    12:00 Visit Demonstration at Virtual Engineering of Engineering Cloud, Visualization techniques and virtual prototypes.
    12:30 Lunch break
    13:30 Lecture Introduction to High Performance Computing (Hartree Centre)
    14:30 Pracical session Practical session and exercises on HPC
    16:30 Wrapping up Summary of the day

    Day 2 - 5th May 2016

    Time Activity Type Description
    10:00 Lecture Introduction to COSSAN-X (Main features and capabilities, toolboxes and wizards)
    10:30 Demonstration Getting started and familiarisation with the software. Define Random Variables and construct a stochastic model
    11:00 Practcal session Basic Tutorial: UQ and Sensitivity analysis of a simple Cantilever Beam
    11:30 Practical session Applications and Exercises
    12:30 Lunch break
    13:30 Practical session Connect COSSAN-X with 3rd party solvers (ABAQUS, NASTRAN, LS_DYNA, etc) and job submission to cluster
    14:00 Practical session Meta-models, Robust Optimization and Reliability Based Optimization
    15:30 Pracical session User defined model and advanced simulations methods
    16:30 Wrapping up Summary of the day

    Day 3 - 6th May 2016

    Time Activity Type Description
    10:00 Lecture Introduction to OpenCossan (Main features and capabilities, toolboxes and wizards)
    10:30 Lecture Basic Matlab concepts and Object Orienting Programming
    11:00 Practical session Code and use simple objects in Matlab
    11:30 Demonstration OpenCossan structure, Tutorials and Unit Test
    11:45 Practical session Uncertainty Quantification of simple models and advanced simulation methods
    12:30 Lunch break
    13:30 Demonstration Interaction with 3rd party solver and job submission to cluster
    14:00 Practical session Advanced Monte Carlo methods
    15:30 Pracical session User defined solution, loops and advanced methods
    16:00 Lecture OpenCossan 2.0 - upcoming features and collaborative development
    16:30 Wrapping up Summary of the day