General Strategy


  • Monte Carlo generation of data replicas→ no need for linear propagation of errors
    → possibility to test for non Gaussian behaviour in fitted PDFs
  • Neural Networks parametrization of PDFs
  • → 7 independent PDFs, 259 parameters
    → unbiased parametrization
  • Evolution using DGLAP equations
  • Genetic Algorithm’s training of neural networks parameters
  • Analysis of χ2 distributions