1. 21 Mar, 2018 5 commits
  2. 20 Mar, 2018 3 commits
  3. 15 Mar, 2018 3 commits
  4. 14 Mar, 2018 1 commit
  5. 08 Mar, 2018 2 commits
  6. 07 Mar, 2018 5 commits
  7. 06 Mar, 2018 3 commits
  8. 05 Mar, 2018 4 commits
  9. 04 Mar, 2018 4 commits
    • Kevin Jacques's avatar
    • Kevin Jacques's avatar
      Merge branch 'ExtrapolatingPolynomial' · 8cd435b5
      Kevin Jacques authored
      ***NEW: Build Lagrange Interpolating Polynomial to predict current solution instead of starting simply from previous Time Solution.
      Polynomial degree:
      degree 0 ==> initialization at the previous Time Solution
      degree 1 ==> linear extrapolation from the 2 previous Time Solutions
      degree 2 ==> quadratic extrapolation from the 3 previous Time Solutions (this by default)
      NB: The more data points that are used in the interpolation, the higher the degree of the resulting polynomial, thus the greater oscillation it will exhibit between the data points, and therefore the poorest the prediction may be ...
      
      ***NEW: New Linesearch for the relaxation factor with SolveJacAdapt when CheckAll=2
      find the maximum relaxation factor that decreases the residual:
      	- start with the relaxation factor from the previous time step or from the previous iteration
      	- the relaxation factor is multiplied by a ratio as long as the residual decreases
      	- the relaxation factor is decreased by a ratio until a decreasing residual is found
      
      ***NEW: Current.SolveJacAdaptFailed to detect when the Linesearch failed in SolveJacAdapt
      8cd435b5
    • Kevin Jacques's avatar
    • Kevin Jacques's avatar
      recompute bc at the begin of Vector_b_Vinch_K, seems recommended with... · 6fb4ff8a
      Kevin Jacques authored
      recompute bc at the begin of Vector_b_Vinch_K, seems recommended with ExtrapolatePolynomial, seems better to use {h} instead of {h}[1] in h_Vinch_K arguments in .pro file (see team32) ?
      6fb4ff8a
  10. 03 Mar, 2018 4 commits
  11. 02 Mar, 2018 6 commits