add new function for warmstart (#688)

* add new function for warmstart

* change function namel; remive pnnl ci tests

* remove push mirror to PNNL

* Update spack_build.yml to remove E4S cache and use newer OS (#689)

* Update spack_build.yml to remove E4S cache and use newer OS

* Build ^libevent~openssl

* add back PNNL CI

---------

Co-authored-by: Cameron Rutherford <robert.rutherford@pnnl.gov>

.github/push_mirror.yml

@@ -0,0 +1,24 @@
+#name: Push Mirror
+
+#on: [push, delete]
+
+#jobs:
+#  PNNL_GitLab:
+#    runs-on: ubuntu-22.04
+#    steps:
+#      - uses: actions/checkout@v1
+#      - uses: spyoungtech/mirror-action@master
+#        with:
+#          REMOTE: ${{ secrets.GIT_REPO_URL }}
+#          GIT_USERNAME: ${{ secrets.GIT_USER }}
+#          GIT_PASSWORD: ${{ secrets.GIT_PASSWORD }}
+#          GIT_PUSH_ARGS: --push-option=ci.skip --tags --force --prune
+#      - uses: nelonoel/branch-name@v1.0.1
+#      - name: Trigger Pipeline
+#        run: |
+#          response=$(curl -X POST -F token=${{ secrets.PNNL_PIPELINE_TRIGGER_TOKEN }} -F ref=${BRANCH_NAME} https://gitlab.pnnl.gov/api/v4/projects/769/trigger/pipeline)
+#          exit_code=$?
+#          sudo apt install jq
+#          pipeline_id=$(echo $response | jq '.id' | sed 's/"//g')
+#          echo "PIPELINE_ID=${pipeline_id}" >> $GITHUB_ENV
+#          exit $exit_code

.github/workflows/spack_build.yml

@@ -4,15 +4,14 @@ on: [push]
 
 jobs:
   hiop_spack_builds:
-    # 20.04 is a version shared by E4S cache and Spack binaries for x86_64
-    runs-on: ubuntu-20.04
-    container: spack/ubuntu-focal:latest
+    runs-on: ubuntu-22.04
+    container: spack/ubuntu-jammy:latest
     strategy:
       matrix:
         spack_spec:
-          - hiop@develop+mpi~raja~shared~kron~sparse ^openmpi
-          - hiop@develop~mpi~raja~shared~kron~sparse
-          - hiop@develop~mpi+raja~shared~kron~sparse
+          - hiop@develop+mpi~raja~shared~kron~sparse ^openmpi ^libevent~openssl
+          - hiop@develop~mpi~raja~shared~kron~sparse ^libevent~openssl
+          - hiop@develop~mpi+raja~shared~kron~sparse ^libevent~openssl
 
             # We will need coinhsl for this, but what are the rules for using
             # a coinhsl tarball?
@@ -41,13 +40,7 @@ jobs:
           spack env activate ./spack-env
           spack add $SPACK_SPEC target=x86_64
           spack develop --path $(pwd) --no-clone hiop@develop
-          # Add E4S mirror - likely relying on spack cache but nice backup
-          # https://oaciss.uoregon.edu/e4s/inventory.html
-          # Need to add build cache before concretizing to re-use dependencies
-          # Using custom e4s cache due to known bug
-          # TODO: Update cache link after bug is resolved
-          spack mirror add E4S https://cache.e4s.io/23.02
-          spack buildcache keys --install --trust
+          spack external find --all --exclude python
           spack concretize --reuse
           git config --global --add safe.directory $(pwd)
           spack --stacktrace install --fail-fast

src/Interface/hiopInterface.hpp

@@ -465,7 +465,38 @@ class hiopInterfaceBase
   {
     return true;
   }
-
+
+  /** 
+   * This method is used to provide an user all the hiop iterate
+   * procedure. @see solution_callback() for an explanation of the parameters.
+   * 
+   * @param[in] x array of (local) entries of the primal variables (managed by Umpire, see note below)
+   * @param[in] z_L array of (local) entries of the dual variables for lower bounds (managed by Umpire, see note below)
+   * @param[in] z_U array of (local) entries of the dual variables for upper bounds (managed by Umpire, see note below)
+   * @param[in] yc array of (local) entries of the dual variables for equality constraints (managed by Umpire, see note below)
+   * @param[in] yd array of (local) entries of the dual variables for inequality constraints (managed by Umpire, see note below)
+   * @param[in] s array of the slacks added to transfer inequalities to equalities (managed by Umpire, see note below)
+   * @param[in] v_L array of (local) entries of the dual variables for constraint lower bounds (managed by Umpire, see note below)
+   * @param[in] v_U array of (local) entries of the dual variables for constraint upper bounds (managed by Umpire, see note below)
+   * 
+   * @note HiOp's option `callback_mem_space` can be used to change the memory location of array parameters managaged by Umpire. 
+   * More specifically, when `callback_mem_space` is set to `host` (and `mem_space` is `device`), HiOp transfers the 
+   * arrays from device to host first, and then passes/returns pointers on host for the arrays managed by Umpire. These pointers
+   * can be then used in host memory space (without the need to rely on or use Umpire).
+   * 
+   */
+  virtual bool iterate_full_callback(const double* x,
+                                     const double* z_L,
+                                     const double* z_U,
+                                     const double* yc,
+                                     const double* yd,
+                                     const double* s,
+                                     const double* v_L,
+                                     const double* v_U)
+  {
+    return true;
+  }
+
   /**
    * A wildcard function used to change the primal variables.
    *

src/Optimization/hiopNlpFormulation.cpp

@@ -1213,7 +1213,7 @@ bool hiopNlpFormulation::user_callback_iterate(int iter,
     hiopVectorPar x_host(n_vars_, vec_distrib_, comm_);
     x.copy_to_vectorpar(x_host);
 
-    hiopVectorPar s_host(n_vars_, vec_distrib_, comm_);
+    hiopVectorPar s_host(n_cons_ineq_, vec_distrib_, comm_);
     s.copy_to_vectorpar(s_host);
 
     hiopVectorPar zl_host(n_vars_, vec_distrib_, comm_);
@@ -1271,6 +1271,73 @@ bool hiopNlpFormulation::user_callback_iterate(int iter,
   return bret; 
 }
 
+bool hiopNlpFormulation::user_callback_full_iterate(hiopVector& x,
+                                                    hiopVector& z_L,
+                                                    hiopVector& z_U,
+                                                    hiopVector& y_c,
+                                                    hiopVector& y_d,
+                                                    hiopVector& s,
+                                                    hiopVector& v_L,
+                                                    hiopVector& v_U)
+{
+  assert(x.get_size()==n_vars_);
+  assert(y_c.get_size() == n_cons_eq_);
+  assert(y_d.get_size() == n_cons_ineq_);
+
+  bool bret{false};
+
+  if(options->GetString("callback_mem_space")=="host" && options->GetString("mem_space")=="device") {
+
+#if !defined(HIOP_USE_MPI)
+    int* vec_distrib_ = nullptr;
+    MPI_Comm comm_ = MPI_COMM_SELF;
+#endif  
+    hiopVectorPar x_host(n_vars_, vec_distrib_, comm_);
+    x.copy_to_vectorpar(x_host);
+
+    hiopVectorPar zl_host(n_vars_, vec_distrib_, comm_);
+    z_L.copy_to_vectorpar(zl_host);
+
+    hiopVectorPar zu_host(n_vars_, vec_distrib_, comm_);
+    z_U.copy_to_vectorpar(zu_host);
+
+    hiopVectorPar yc_host(n_cons_eq_, vec_distrib_, comm_);
+    y_c.copy_to_vectorpar(yc_host);
+
+    hiopVectorPar yd_host(n_cons_ineq_, vec_distrib_, comm_);
+    y_d.copy_to_vectorpar(yd_host);
+
+    hiopVectorPar s_host(n_cons_ineq_, vec_distrib_, comm_);
+    s.copy_to_vectorpar(s_host);
+
+    hiopVectorPar vl_host(n_cons_ineq_, vec_distrib_, comm_);
+    v_L.copy_to_vectorpar(zl_host);
+
+    hiopVectorPar vu_host(n_cons_ineq_, vec_distrib_, comm_);
+    v_U.copy_to_vectorpar(zu_host);    
+
+    bret = interface_base.iterate_full_callback(x_host.local_data_const(),
+                                                zl_host.local_data_const(),
+                                                zu_host.local_data_const(),
+                                                yc_host.local_data_const(),
+                                                yd_host.local_data_const(),
+                                                s_host.local_data_const(),
+                                                vl_host.local_data_const(),
+                                                vu_host.local_data_const());
+  } else {
+    bret = interface_base.iterate_full_callback(x.local_data_const(),
+                                                z_L.local_data_const(),
+                                                z_U.local_data_const(),
+                                                y_c.local_data_const(),
+                                                y_d.local_data_const(),
+                                                s.local_data_const(),
+                                                v_L.local_data_const(),
+                                                v_U.local_data_const());
+  }   
+  return bret; 
+}
+
+
 bool hiopNlpFormulation::user_force_update(int iter,
                                            double& obj_value,
                                            hiopVector& x,

src/Optimization/hiopNlpFormulation.hpp

@@ -192,6 +192,16 @@ class hiopNlpFormulation
                              double alpha_pr,
                              int ls_trials);
 
+  virtual
+  bool user_callback_full_iterate(hiopVector& x,
+                                  hiopVector& z_L,
+                                  hiopVector& z_U,
+                                  hiopVector& y_c,
+                                  hiopVector& y_d,
+                                  hiopVector& s,
+                                  hiopVector& v_L,
+                                  hiopVector& v_U);
+
   virtual
   bool user_force_update(int iter,
                          double& obj_value,