reorganize

.gitignore

@@ -3,6 +3,7 @@ __pycache__/
 *.py[cod]
 *$py.class
 .pyenv
+.vscode/settings.json
 
 # C extensions
 *.so

setup.py

@@ -24,5 +24,6 @@
         "seaborn",
         "scipy",
         "statsmodels",
+        "itis-dakota",
     ],
 )

src/mmux_utils/funs_evaluate.py

@@ -0,0 +1,36 @@
+### Allows to evaluate in different modes - batch, single set, ...
+### also in OSPARC or local deployment
+from typing import Callable, List
+from pathlib import Path
+import dakota.environment as dakenv
+
+
+### Otherwise could create a class that I can instantiate and give a "model" at initialization time,
+# which is given to "run dakota" as input parameter
+def batch_evaluator(model: Callable, batch_input: List[dict]):
+    return map(model, batch_input)  # FIXME not sure this will work
+
+
+def single_evaluator(model: Callable, input: dict):
+    return model(input)
+
+
+def run_dakota(dakota_conf_path: Path, batch_mode: bool = True):
+    print("Starting dakota")
+    dakota_conf = dakota_conf_path.read_text()
+    callbacks = (
+        {"batch_evaluator": batch_evaluator}
+        if batch_mode
+        else {"evaluator": single_evaluator}  # not sure this will work
+    )
+    study = dakenv.study(
+        callbacks=callbacks,
+        input_string=dakota_conf,
+    )
+    study.execute()
+    ## TODO access documentation of dakenv.study -- cannot, also cannot find in https://github.com/snl-dakota/dakota/tree/devel/packages
+    # would need to ask Werner
+
+
+if __name__ == "__main__":
+    print("DONE")