CUDA BFGSTS

 Beginning of optimization cycle 1.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in      15 steps
 EF step uphill of magnitude 0.00452566 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        -0.0385806        -8.52468     -0.00452566            0.22      0.00342022 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.004743056

 Beginning of optimization cycle 2.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       2 steps
 EF step uphill of magnitude 0.000114918 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1      -0.000973671        -8.47275    -0.000114918           0.242     7.12773e-05 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001837434

 Beginning of optimization cycle 3.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       2 steps
 EF step uphill of magnitude 9.507911e-06 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       7.79454e-05        -8.19796     9.50791e-06          0.2662     0.000118634 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.003502528

 Beginning of optimization cycle 4.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 3.480705e-06 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       2.84184e-05        -8.16456      3.4807e-06         0.29282     0.000116878 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001289905

 Beginning of optimization cycle 5.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 4.474655e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000367927        -8.22247     4.47465e-05        0.322102     0.000130051 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001367141

 Beginning of optimization cycle 6.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 9.62146e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000790875        -8.21991     9.62146e-05        0.354312     0.000142002 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001747454

 Beginning of optimization cycle 7.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 0.0001679348 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00138319         -8.2365     0.000167935        0.389743     0.000167765 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001662639

 Beginning of optimization cycle 8.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 0.0002680119 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00219895        -8.20465     0.000268012        0.428718     0.000198606 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.002789698

 Beginning of optimization cycle 9.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 0.0002226631 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00182972        -8.21742     0.000222663         0.47159     0.000184768 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001321944

 Beginning of optimization cycle 10.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 0.000162106 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00133296        -8.22275     0.000162106             0.5     0.000162715 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001258587

 Beginning of optimization cycle 11.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 0.0002107001 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00173259        -8.22301       0.0002107             0.5     0.000169886 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001019004

 Beginning of optimization cycle 12.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 0.0002029928 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00166783        -8.21619     0.000202993             0.5     0.000177471 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0005911983

 Beginning of optimization cycle 13.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 0.0001383822 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1        0.00113725         -8.2182     0.000138382             0.5     0.000158372 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0008701646

 Beginning of optimization cycle 14.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 9.5175e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000782071        -8.21719      9.5175e-05             0.5      0.00014072 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =     0.001714746

 Beginning of optimization cycle 15.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 0.0001117429 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000918056        -8.21579     0.000111743             0.5     0.000146129 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0004952231

 Beginning of optimization cycle 16.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 8.882253e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000729844        -8.21688     8.88225e-05             0.5     0.000140733 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0004575121

 Beginning of optimization cycle 17.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 9.715289e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000798148        -8.21538     9.71529e-05             0.5      0.00014244 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0005238992

 Beginning of optimization cycle 18.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       3 steps
 EF step uphill of magnitude 8.085511e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000664679        -8.22062     8.08551e-05             0.5     0.000140477 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0004975653

 Beginning of optimization cycle 19.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 5.619363e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000461771         -8.2175     5.61936e-05             0.5     0.000132718 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0003589817

 Beginning of optimization cycle 20.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 5.585261e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000458995        -8.21796     5.58526e-05             0.5     0.000132777 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0003909304

 Beginning of optimization cycle 21.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 3.320125e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000272704        -8.21368     3.32013e-05             0.5     0.000124838 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0002828283

 Beginning of optimization cycle 22.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 2.249429e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000184768        -8.21399     2.24943e-05             0.5      0.00012111 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0002256908

 Beginning of optimization cycle 23.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.435382e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000117922        -8.21539     1.43538e-05             0.5     0.000118678 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0002852152

 Beginning of optimization cycle 24.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.269645e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000104318        -8.21632     1.26964e-05             0.5     0.000118189 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0002953888

 Beginning of optimization cycle 25.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 9.482311e-06 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       7.78987e-05        -8.21516     9.48231e-06             0.5     0.000117339 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0004091412

 Beginning of optimization cycle 26.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.137353e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       9.34519e-05        -8.21661     1.13735e-05             0.5     0.000118014 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0001900548

 Beginning of optimization cycle 27.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.283199e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       0.000105439        -8.21685      1.2832e-05             0.5     0.000118508 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0001955091

 Beginning of optimization cycle 28.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.028572e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       8.45074e-05        -8.21599     1.02857e-05             0.5      0.00011851 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0001568647

 Beginning of optimization cycle 29.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 1.198785e-05 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       9.84852e-05        -8.21542     1.19878e-05             0.5     0.000118103 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    0.0001169435

 Beginning of optimization cycle 30.
 -------------------------------
 Lowest eigenvector search (Rayleigh-Ritz):
 Smallest eigenvalue converged in       0 steps
 EF step uphill of magnitude 9.660014e-06 is taken

 ----------------------------------------------------------------------------------------
  Vector        Gradient          Secder            Step        Max step     Trust ratio 
 ----------------------------------------------------------------------------------------
     1       7.93684e-05        -8.21617     9.66001e-06             0.5     0.000117407 
 ----------------------------------------------------------------------------------------

 Minimization perpendicular to lowest eigenvector:
 True RMS grad =    9.849929e-05

 Reason for termination: Converged to a transition state of Hessian index 1

