Scientific Computing Lab


The student should achieve a basic understanding of
  • time series data with stationary noise and signals
  • statistical concepts (moments, median, higher moments)
  • template placement, mismatch, ROC curve
  • sensitivity computational cost analysis
  • parallelization with a cluster using condor

suggested program

  1. find the right parameter in data with noise
  2. noise realization, statistical properties of noise, plot different realizations of noise and properties there-of (moments, rms)
  3. properties of simple signal without noise, plot different sample systems, i.e. $s(t) = A \cos(2 \pi f t - \varphi)$
  4. work with different files, store these in SCM
  5. adding signal + noise, plot various realizations
  6. learn to use prober, our small black box program computing a scalar result based upon input time series and parameter point $(f,\varphi)$
  7. start to learn how prober results vary w.r.t. varying input parameters of injected signal while probing the same parameter point
  8. plot/document result findings, e.g. how large may the mismatch between injection and probe parameter may become, what about weak signals?
  9. develop strategy to find signal with prober if signal parameters $(A,f,\varphi)$ are unknown
  10. with this strategy perform multiple injections to arrive at mismatch graphs (with and without signals) and ROC curve to evaluate template bank properties
  11. repeat with coarser and finer template bank

plot always means a graphical output with proper labels, title


The student should have at least basic knowledge of
  • C programming language
  • scripting language (Python is particularly useful)
  • SCM (software configuration management) tool (git)
  • basic statistical working knowledge (mean. variance, higher moments), time series analysis

programming skills

  • using debugging, profiling and performance monitoring tools
  • construct a Makefile
  • writing of programs to solve particular problems
  • creating data type precision problems
  • discovering and handle IO problems
  • program optimization
  • software configuration management
  • download and compile the LAL software
  • parallel programming skills, such as GPU programming, MPI and OpenMP

condor skills

  • the usage of condor to manage jobs
  • generating DAG-man jobs

scientific skills

  • generate a random and a stochastic template bank
  • generate a mismatch statistic of a random, a stochastic and an optimized template bank, what is the meaning of it
  • generate a roc curve, explain the consequences in terms of data analysis resources and sensitivity
  • find an injected signal
  • find correlation between different signals

  • Percival, Walden: "Spectral Analysis for Physical Applications : Multitaper and Conventional Univariate Techniques"
  • Porat "A Course in Digital Signal Processing"
  • E.T. Jaynes, "Probability. The logic of Science".

-- HenningFehrmann - 26 Jun 2015
Topic revision: r8 - 21 Jul 2015, HenningFehrmann
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback