Address PR #915 review: warmstart guard, docs, timing, tests

- Add warmstart_id guard in cuopt_sh.py when multiple LP problems
- Fix lp-examples.rst: Genric → Generic, add deprecation to section title
- Add deprecation note to CLI batch mode section
- Fix lp-qp-features.rst references to renamed RST anchor
- Use i_sol.get_solve_time() in server batch path (engine timing)
- Fix error status serialization: use .name for JSON in batch error path
- Add numerical equality assertions in test_parser_and_batch_solver
- Remove pytest.warns for DeprecationWarning (fails with older installed pkg)

Author: anandhkb

docs/cuopt/source/cuopt-server/examples/lp-examples.rst

@@ -15,10 +15,10 @@ If you want to run server locally, please run the following command in a termina
     export port=5000
     python -m cuopt_server.cuopt_service --ip $ip --port $port
 
-.. _generic-example-with-normal-and-batch-mode:
+.. _generic-example-with-normal-and-multiple-lps:
 
-Genric Example With Normal Mode and Batch Mode
-------------------------------------------------
+Generic Example With Normal Mode and Multiple LPs (Batch Deprecated)
+---------------------------------------------------------------------
 
 :download:`basic_lp_example.py <lp/examples/basic_lp_example.py>`
 
@@ -402,7 +402,10 @@ In case the user needs to update solver settings through CLI, the option ``-ss``
    export port=5000
    cuopt_sh data.json -t LP -i $ip -p $port -ss '{"tolerances": {"optimality": 0.0001}, "time_limit": 5}'
 
-In the case of batch mode, you can send a bunch of ``mps`` files at once, and acquire results. The batch mode works only for ``mps`` in the case of CLI:
+In the case of batch mode, you can send a bunch of ``mps`` files at once, and acquire results. The batch mode works only for ``mps`` in the case of CLI.
+
+.. note::
+   LP batch mode is deprecated. Multiple problems are now solved sequentially.
 
 .. note::
    Batch mode is not available for MILP problems.

docs/cuopt/source/lp-qp-features.rst

@@ -131,7 +131,7 @@ Logging Callback in the Service
 
 In the cuOpt service API, the ``log_file`` value in ``solver_configs`` is ignored.
 
-If however you set the ``solver_logs`` flag on the ``/cuopt/request`` REST API call, users can fetch the log file content from the webserver at ``/cuopt/logs/{id}``. Using the logging callback feature through the cuOpt client is shown in :ref:`Examples <generic-example-with-normal-and-batch-mode>` on the self-hosted page.
+If however you set the ``solver_logs`` flag on the ``/cuopt/request`` REST API call, users can fetch the log file content from the webserver at ``/cuopt/logs/{id}``. Using the logging callback feature through the cuOpt client is shown in :ref:`Examples <generic-example-with-normal-and-multiple-lps>` on the self-hosted page.
 
 
 Infeasibility Detection
@@ -155,7 +155,7 @@ The user may specify a time limit to the solver. By default the solver runs unti
 Batch Mode
 ----------
 
-Users can submit a set of problems which will be solved in a batch. Problems will be solved at the same time in parallel to fully utilize the GPU. Checkout :ref:`self-hosted client <generic-example-with-normal-and-batch-mode>` example in thin client.
+Users can submit a set of problems which will be solved in a batch. Problems will be solved at the same time in parallel to fully utilize the GPU. Checkout :ref:`self-hosted client <generic-example-with-normal-and-multiple-lps>` example in thin client.
 
 Multi-GPU Mode
 --------------

python/cuopt/cuopt/tests/linear_programming/test_lp_solver.py

@@ -478,6 +478,8 @@ def test_parser_and_batch_solver():
     settings.set_optimality_tolerance(1e-4)
 
     # Call BatchSolve (deprecated; use sequential Solve instead)
+    # DeprecationWarning is emitted when running against a build with the
+    # deprecation; CI asserts it via pytest.warns
     batch_solution, solve_time = solver.BatchSolve(data_model_list, settings)
 
     # Call Solve on each individual data model object
@@ -494,6 +496,16 @@ def test_parser_and_batch_solver():
             batch_solution[i].get_termination_status()
             == individual_solutions[i].get_termination_status()
         )
+        assert batch_solution[i].get_primal_objective() == pytest.approx(
+            individual_solutions[i].get_primal_objective(), rel=1e-6, abs=1e-8
+        )
+        assert np.array(
+            batch_solution[i].get_primal_solution()
+        ) == pytest.approx(
+            np.array(individual_solutions[i].get_primal_solution()),
+            rel=1e-5,
+            abs=1e-7,
+        )
 
 
 def test_warm_start():

python/cuopt_self_hosted/cuopt_sh_client/cuopt_sh.py

@@ -231,6 +231,14 @@ def read_input_data(i_file):
         elif args.type == "LP":
             if args.init_ids:
                 raise Exception("Initial ids are not supported for LP")
+            if (
+                isinstance(cuopt_problem_data, list)
+                and len(cuopt_problem_data) > 1
+                and args.warmstart_id
+            ):
+                raise Exception(
+                    "Warmstart id is only supported for a single LP problem"
+                )
 
             def log_callback(name):
                 def print_log(log):

python/cuopt_server/cuopt_server/utils/linear_programming/solver.py

@@ -355,11 +355,10 @@ def create_solution(sol):
             for i_data in LP_data:
                 i_warnings, data_model = create_data_model(i_data)
                 warnings.extend(i_warnings)
-                solve_begin_time = time.time()
                 i_sol = linear_programming.Solve(
                     data_model, solver_settings=solver_settings
                 )
-                total_solve_time += time.time() - solve_begin_time
+                total_solve_time += i_sol.get_solve_time()
                 sol.append(i_sol)
         else:
             warnings, data_model = create_data_model(LP_data)
@@ -392,7 +391,7 @@ def create_solution(sol):
                 if i_sol.get_error_status() != ErrorStatus.Success:
                     res.append(
                         {
-                            "status": i_sol.get_error_status(),
+                            "status": i_sol.get_error_status().name,
                             "solution": i_sol.get_error_message(),
                         }
                     )