set gen_specs.user.lb and ub from vocs variables bounds

docs/platforms/frontier.rst

@@ -64,10 +64,9 @@ Now grab an interactive session on one node::
 
 Then in the session run::
 
-    python run_libe_forces.py --nworkers 9
+    python run_libe_forces.py --nworkers 8
 
-This places the generator on the first worker and runs simulations on the
-others (each simulation using one GPU).
+The workers will each run simulations with one GPU each.
 
 To see GPU usage, ssh into the node you are on in another window and run::
 

docs/platforms/perlmutter.rst

@@ -96,10 +96,9 @@ Now grab an interactive session on one node::
 Then in the session run::
 
     export LIBE_PLATFORM="perlmutter_g"
-    python run_libe_forces.py -n 5
+    python run_libe_forces.py -n 4
 
-This places the generator on the first worker and runs simulations on the
-others (each simulation using one GPU).
+The workers will each run simualations with one GPU each.
 
 To see GPU usage, ssh into the node you are on in another window and run::
 

docs/tutorials/xopt_bayesian_gen.rst

@@ -52,7 +52,7 @@ Define the VOCS specification and set up the generator.
 
 .. code-block:: python
 
-    libE_specs = LibeSpecs(gen_on_manager=True, nworkers=4)
+    libE_specs = LibeSpecs(nworkers=4)
 
     vocs = VOCS(
         variables={"x1": [0, 1.0], "x2": [0, 10.0]},

examples/tutorials/gpcam_surrogate_model/gpcam.ipynb

@@ -285,9 +285,9 @@
     "\n",
     "nworkers = 4\n",
     "\n",
-    "# When using gen_on_manager, nworkers is number of concurrent sims.\n",
+    "# nworkers is number of concurrent sims.\n",
     "# final_gen_send means the last evaluated points are returned to the generator to update the model.\n",
-    "libE_specs = LibeSpecs(nworkers=nworkers, gen_on_manager=True, final_gen_send=True)\n",
+    "libE_specs = LibeSpecs(nworkers=nworkers, final_gen_send=True)\n",
     "\n",
     "n = 2  # Input dimensions\n",
     "batch_size = 4\n",
@@ -400,7 +400,7 @@
     "import matplotlib\n",
     "import matplotlib.pyplot as plt\n",
     "\n",
-    "# Get \"mean_squared_error\" from generators return (worker 0 as we ran gen_on_manager)\n",
+    "# Get \"mean_squared_error\" from generators return\n",
     "mse = persis_info[0][\"mean_squared_error\"]\n",
     "niter = len(mse)\n",
     "num_sims = list(range(batch_size, (niter * batch_size) + 1, batch_size))\n",

examples/tutorials/xopt_bayesian_gen/xopt_EI_example.ipynb

@@ -98,7 +98,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "libE_specs = LibeSpecs(gen_on_manager=True, nworkers=4)\n",
+    "libE_specs = LibeSpecs(nworkers=4)\n",
     "\n",
     "vocs = VOCS(\n",
     "    variables={\"x1\": [0, 1.0], \"x2\": [0, 10.0]},\n",

libensemble/specs.py

@@ -2,6 +2,7 @@
 import warnings
 from pathlib import Path
 
+import numpy as np
 import pydantic
 from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
 
@@ -354,6 +355,24 @@ def set_fields_from_vocs(self):
             if "_id" not in self.persis_in:
                 self.persis_in.append("_id")
 
+        # Set user["lb"]/["ub"] from VOCS continuous variables (for legacy generators
+        # that read bounds from gen_specs["user"]). Skip variables without a ``.domain``
+        # attribute (e.g., DiscreteVariable). Do not overwrite user-provided values.
+        if self.user is None:
+            self.user = {}
+        if "lb" not in self.user or "ub" not in self.user:
+            lbs, ubs = [], []
+            for _name, var in (getattr(self.vocs, "variables", None) or {}).items():
+                domain = getattr(var, "domain", None)
+                if domain is not None and len(domain) == 2:
+                    lbs.append(domain[0])
+                    ubs.append(domain[1])
+            if lbs:
+                if "lb" not in self.user:
+                    self.user["lb"] = np.array(lbs, dtype=float)
+                if "ub" not in self.user:
+                    self.user["ub"] = np.array(ubs, dtype=float)
+
         return self
 
     @model_validator(mode="after")

libensemble/tests/functionality_tests/test_1d_super_simple.py

@@ -14,6 +14,7 @@
 # TESTSUITE_NPROCS: 2 4
 
 import numpy as np
+from gest_api.vocs import VOCS
 
 from libensemble.alloc_funcs.give_sim_work_first import give_sim_work_first
 from libensemble.gen_funcs.sampling import latin_hypercube_sample as gen_f
@@ -38,14 +39,13 @@ def sim_f(In):
         "out": [("f", float)],
     }
 
+    vocs = VOCS(variables={"x0": [-3, 3]}, objectives={"f": "MINIMIZE"})
+
     gen_specs = {
         "gen_f": gen_f,
         "out": [("x", float, (1,))],
         "batch_size": 500,
-        "user": {
-            "lb": np.array([-3]),
-            "ub": np.array([3]),
-        },
+        "vocs": vocs,
     }
 
     exit_criteria = {"gen_max": 501}

libensemble/tests/functionality_tests/test_asktell_sampling.py

@@ -45,10 +45,6 @@ def sim_f(In):
         "out": [("x", float, (2,))],
         "initial_batch_size": 20,
         "batch_size": 10,
-        "user": {
-            "lb": np.array([-3, -2]),
-            "ub": np.array([3, 2]),
-        },
     }
 
     variables = {"x0": [-3, 3], "x1": [-2, 2]}

libensemble/tests/functionality_tests/test_calc_exception.py

@@ -11,7 +11,7 @@
 # TESTSUITE_COMMS: mpi local tcp
 # TESTSUITE_NPROCS: 2 4
 
-import numpy as np
+from gest_api.vocs import VOCS
 
 from libensemble.alloc_funcs.give_sim_work_first import give_sim_work_first
 from libensemble.gen_funcs.sampling import uniform_random_sample as gen_f
@@ -35,15 +35,14 @@ def six_hump_camel_err(H, persis_info, sim_specs, _):
         "out": [("f", float)],
     }
 
+    vocs = VOCS(variables={"x0": [-3, 3], "x1": [-2, 2]}, objectives={"f": "MINIMIZE"})
+
     gen_specs = {
         "gen_f": gen_f,
         "in": ["sim_id"],
         "out": [("x", float, 2)],
         "batch_size": 10,
-        "user": {
-            "lb": np.array([-3, -2]),
-            "ub": np.array([3, 2]),
-        },
+        "vocs": vocs,
     }
 
     alloc_specs = {

libensemble/tests/functionality_tests/test_cancel_in_alloc.py

@@ -18,6 +18,7 @@
 # TESTSUITE_NPROCS: 4
 
 import numpy as np
+from gest_api.vocs import VOCS
 
 from libensemble.alloc_funcs.give_sim_work_first import give_sim_work_first
 from libensemble.gen_funcs.sampling import uniform_random_sample as gen_f
@@ -39,16 +40,15 @@
         "user": {"uniform_random_pause_ub": 10},
     }
 
+    vocs = VOCS(variables={"x0": [-3, 3], "x1": [-2, 2]}, objectives={"f": "MINIMIZE"})
+
     gen_specs = {
         "gen_f": gen_f,
         "in": ["sim_id"],
         "out": [("x", float, (2,))],
         "batch_size": 5,
         "num_active_gens": 1,
-        "user": {
-            "lb": np.array([-3, -2]),
-            "ub": np.array([3, 2]),
-        },
+        "vocs": vocs,
     }
 
     alloc_specs = {
@@ -62,7 +62,13 @@
     exit_criteria = {"sim_max": 10, "wallclock_max": 300}
 
     # Perform the run
-    H, persis_info, flag = libE(sim_specs, gen_specs, exit_criteria, libE_specs=libE_specs, alloc_specs=alloc_specs)
+    H, persis_info, flag = libE(
+        sim_specs,
+        gen_specs,
+        exit_criteria,
+        libE_specs=libE_specs,
+        alloc_specs=alloc_specs,
+    )
 
     if is_manager:
         test = np.any(H["cancel_requested"]) and np.any(H["kill_sent"])

libensemble/tests/functionality_tests/test_comms.py

@@ -15,6 +15,7 @@
 # TESTSUITE_NPROCS: 2 4
 
 import numpy as np
+from gest_api.vocs import VOCS
 
 from libensemble.alloc_funcs.give_sim_work_first import give_sim_work_first
 from libensemble.executors.mpi_executor import MPIExecutor  # Only used to get workerID in float_x1000
@@ -41,15 +42,14 @@
         "out": [("arr_vals", float, array_size), ("scal_val", float)],
     }
 
+    vocs = VOCS(variables={"x0": [-3, 3], "x1": [-2, 2]}, objectives={"f": "MINIMIZE"})
+
     gen_specs = {
         "gen_f": gen_f,
         "in": ["sim_id"],
         "out": [("x", float, (2,))],
         "batch_size": sim_max,
-        "user": {
-            "lb": np.array([-3, -2]),
-            "ub": np.array([3, 2]),
-        },
+        "vocs": vocs,
     }
 
     exit_criteria = {"sim_max": sim_max, "wallclock_max": 300}

libensemble/tests/functionality_tests/test_elapsed_time_abort.py

@@ -13,7 +13,7 @@
 # TESTSUITE_COMMS: mpi local tcp
 # TESTSUITE_NPROCS: 2 4
 
-import numpy as np
+from gest_api.vocs import VOCS
 
 from libensemble.alloc_funcs.give_sim_work_first import give_sim_work_first
 from libensemble.gen_funcs.sampling import uniform_random_sample as gen_f
@@ -34,16 +34,15 @@
         "user": {"pause_time": 2},
     }
 
+    vocs = VOCS(variables={"x0": [-3, 3], "x1": [-2, 2]}, objectives={"f": "MINIMIZE"})
+
     gen_specs = {
         "gen_f": gen_f,
         "in": ["sim_id"],
         "out": [("x", float, (2,))],
         "batch_size": 5,
         "num_active_gens": 2,
-        "user": {
-            "lb": np.array([-3, -2]),
-            "ub": np.array([3, 2]),
-        },
+        "vocs": vocs,
     }
 
     alloc_specs = {

libensemble/tests/functionality_tests/test_evaluate_existing_plus_gen.py

@@ -15,6 +15,7 @@
 # TESTSUITE_NPROCS: 2 4
 
 import numpy as np
+from gest_api.vocs import VOCS
 
 # Import libEnsemble items for this test
 from libensemble import Ensemble
@@ -24,11 +25,8 @@
 from libensemble.specs import AllocSpecs, ExitCriteria, GenSpecs, SimSpecs
 
 
-def create_H0(gen_specs, H0_size):
+def create_H0(lb, ub, H0_size):
     """Create an H0 for give_pregenerated_sim_work"""
-    # Manually creating H0
-    ub = gen_specs["user"]["ub"]
-    lb = gen_specs["user"]["lb"]
     n = len(lb)
     b = H0_size
 
@@ -45,18 +43,19 @@ def create_H0(gen_specs, H0_size):
     sampling = Ensemble(parse_args=True)
     sampling.sim_specs = SimSpecs(sim_f=sim_f, inputs=["x"], out=[("f", float)])
 
-    gen_specs = {
-        "gen_f": gen_f,
-        "outputs": [("x", float, (2,))],
-        "batch_size": 50,
-        "user": {
-            "lb": np.array([-3, -3]),
-            "ub": np.array([3, 3]),
-        },
-    }
-    sampling.gen_specs = GenSpecs(**gen_specs)
+    vocs = VOCS(variables={"x0": [-3, 3], "x1": [-3, 3]}, objectives={"f": "MINIMIZE"})
+    lb = np.array([-3, -3])
+    ub = np.array([3, 3])
+
+    sampling.gen_specs = GenSpecs(
+        gen_f=gen_f,
+        persis_in=["f"],
+        outputs=[("x", float, (2,))],
+        batch_size=50,
+        vocs=vocs,
+    )
     sampling.exit_criteria = ExitCriteria(sim_max=100)
-    sampling.H0 = create_H0(gen_specs, 50)
+    sampling.H0 = create_H0(lb, ub, 50)
     sampling.alloc_specs = AllocSpecs(alloc_f=give_sim_work_first)
     sampling.run()
 

libensemble/tests/functionality_tests/test_executor_forces_tutorial.py

@@ -1,8 +1,8 @@
 import os
 import sys
 
-import numpy as np
 from forces_simf import run_forces  # Sim func from current dir
+from gest_api.vocs import VOCS
 
 from libensemble import Ensemble
 from libensemble.executors import MPIExecutor
@@ -37,17 +37,16 @@
         outputs=[("energy", float)],
     )
 
+    vocs = VOCS(variables={"nparticles": [1000, 3000]}, objectives={"energy": "MINIMIZE"})
+
     ensemble.gen_specs = GenSpecs(
         gen_f=gen_f,
         inputs=[],  # No input when starting persistent generator
         persis_in=["sim_id"],  # Return sim_ids of evaluated points to generator
         outputs=[("x", float, (1,))],
         initial_batch_size=nsim_workers,
         async_return=False,
-        user={
-            "lb": np.array([1000]),  # min particles
-            "ub": np.array([3000]),  # max particles
-        },
+        vocs=vocs,
     )  # gen_specs_end_tag
 
     # Starts one persistent generator. Simulated values are returned in batch.

libensemble/tests/functionality_tests/test_executor_forces_tutorial_2.py

@@ -1,8 +1,8 @@
 import os
 import sys
 
-import numpy as np
 from forces_simf import run_forces  # Sim func from current dir
+from gest_api.vocs import VOCS
 
 from libensemble import Ensemble, logger
 from libensemble.executors import MPIExecutor
@@ -39,17 +39,16 @@
         outputs=[("energy", float)],
     )
 
+    vocs = VOCS(variables={"nparticles": [1000, 3000]}, objectives={"energy": "MINIMIZE"})
+
     ensemble.gen_specs = GenSpecs(
         gen_f=gen_f,
         inputs=[],  # No input when starting persistent generator
         persis_in=["sim_id"],  # Return sim_ids of evaluated points to generator
         outputs=[("x", float, (1,))],
         initial_batch_size=nsim_workers,
         async_return=True,
-        user={
-            "lb": np.array([1000]),  # min particles
-            "ub": np.array([3000]),  # max particles
-        },
+        vocs=vocs,
     )  # gen_specs_end_tag
 
     # Starts one persistent generator. Simulated values are returned in batch.