GitLab

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

Build/Test Cases Summary
Enabled Packages: Belos, Domi, Intrepid2, MueLu, Phalanx, Pike, PikeBlackBox, PikeImplicit, ROL, Sacado, ShyLUBDDC, Stokhos, Xpetra
Disabled Packages: PyTrilinos,Pliris,Claps,TriKota
Enabled all Forward Packages
0) MPI_DEBUG => Test case MPI_DEBUG was not run! => Does not affect push readiness! (-1.00 min)
1) SERIAL_RELEASE => Test case SERIAL_RELEASE was not run! => Does not affect push readiness! (-1.00 min)
2) MPI_RELEASE_DEBUG_SHARED_ST => passed: passed=1823,notpassed=0 (372.70 min)
Other local commits for this build/test group: dab529ca, 70511c33, 8f491b6c, 006799fd, 255616c4, daf5ef39, f4d67f75, b6dd0b36, 4eacf051, 42bf667a, 487c1574

These newlines got added by the script that removes redundant TriBITS includes
(see trilinos/Trilinos#489).

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

This removes clutter and speeds up configures.  See trilinos/Trilinos#489.

These newlines got added by the script that removes redundant TriBITS includes
(see trilinos/Trilinos#489).

These newlines got added by the script that removes redundant TriBITS includes
(see trilinos/Trilinos#489).

These newlines got added by the script that removes redundant TriBITS includes
(see trilinos/Trilinos#489).

This got added by the script that was removing redundant TriBITS includes (see
Trilinose #489).

of Chebyshev1Kusuoka.

the integral, we use GaussLegendreQuadrature.

-- New virtual functions.

-- Better trace estimation.

-- Fixed a bug where the random vector used in trace estimation
   was not zero-mean.

  high order evaluation is a bit tricky as it uses jacobi polynomial
  and jacobi functions are evaluated for a set of points, which means
  that the point loop goes inside of dof loop traversal. This is not
  good as we want the minimal compute routine (serial) is for a single
  point. To amortize the function call overhead and satisfy the need
  of Matt's PIC interface, I put a parameter that allows a tuning parameter
  for the size of points evaluated in a single jacobi function call.
  For this moment, it is set as 1; later, find a general tune values.

Forgot to commit earlier

This is in support of computing D2fDx2, D2fDxDp, D2fDpdx, D2fDp2.
This didn't have any internal support in the response library.

This is tested against a mass matrix

Now we have D2gDx2, D2gDp2, D2gDxDp, D2gDpDx derivatives

Residual is next

Now response is nonlinear in both the state and the control variables

This is only with respect to "x" (in terms of a model evaluator). The
distributed variables "p" are next

   Allowing use of both external camd and camd in Trilinos in the
   examples.