add simplify jdata docs; fix and check example

doc/simplify/simplify-jdata.rst

@@ -0,0 +1,6 @@
+dpgen simplify parameters
+=========================
+
+.. dargs::
+   :module: dpgen.simplify.arginfo
+   :func: simplify_jdata_arginfo

dpgen/simplify/arginfo.py

@@ -1,10 +1,113 @@
-from dargs import Argument
+from typing import List
+from dargs import Argument, Variant
 
 from dpgen.arginfo import general_mdata_arginfo
+from dpgen.generator.arginfo import (
+    basic_args,
+    data_args,
+    training_args,
+    fp_style_vasp_args,
+    fp_style_gaussian_args,
+)
+
+
+def general_simplify_arginfo() -> Argument:
+    """General simplify arginfo.
+
+    Returns
+    -------
+    Argument
+        arginfo
+    """
+    doc_labeled = "If true, the initial data is labeled."
+    doc_pick_data = "Path to the directory with the pick data with the deepmd/npy format. Systems are detected recursively."
+    doc_init_pick_number = "The number of initial pick data."
+    doc_iter_pick_number = "The number of pick data in each iteration."
+    doc_model_devi_f_trust_lo = "The lower bound of forces for the selection for the model deviation."
+    doc_model_devi_f_trust_hi = "The higher bound of forces for the selection for the model deviation."
+
+    return [
+        Argument("labeled", bool, optional=False, default=False, doc=doc_labeled),
+        Argument("pick_data", str, doc=doc_pick_data),
+        Argument("init_pick_number", int, doc=doc_init_pick_number),
+        Argument("iter_pick_number", int, doc=doc_iter_pick_number),
+        Argument("model_devi_f_trust_lo", float, optional=False, doc=doc_model_devi_f_trust_lo),
+        Argument("model_devi_f_trust_hi", float, optional=False, doc=doc_model_devi_f_trust_hi),
+    ]
+
+
+def fp_style_variant_type_args() -> Variant:
+    """Generate variant for fp style variant type.
+    
+    Returns
+    -------
+    Variant
+        variant for fp style
+    """
+    doc_fp_style = 'Software for First Principles, if `labeled` is false. Options include “vasp”, “gaussian” up to now.'
+    doc_fp_style_none = 'No fp.'
+    doc_fp_style_vasp = 'VASP.'
+    doc_fp_style_gaussian = 'Gaussian. The command should be set as `g16 < input`.'
+
+    return Variant("fp_style", [
+        Argument("none", dict, doc=doc_fp_style_none),
+        # simplify use the same fp method as run
+        Argument("vasp", dict, fp_style_vasp_args(), doc=doc_fp_style_vasp),
+        Argument("gaussian", dict, fp_style_gaussian_args(),
+                 doc=doc_fp_style_gaussian),
+    ],
+        optional=True,
+        default_tag="none",
+        doc=doc_fp_style)
+
+
+def fp_args() -> List[Argument]:
+    """Generate arginfo for fp.
+
+    Returns
+    -------
+    List[Argument]
+        arginfo
+    """
+    doc_fp_task_max = 'Maximum of structures to be calculated in 02.fp of each iteration.'
+    doc_fp_task_min = 'Minimum of structures to be calculated in 02.fp of each iteration.'
+
+    return [
+        Argument("fp_task_max", int, optional=True, doc=doc_fp_task_max),
+        Argument("fp_task_min", int, optional=True, doc=doc_fp_task_min),
+    ]
+
+
+def simplify_jdata_arginfo() -> Argument:
+    """Generate arginfo for dpgen simplify jdata.
+
+    Returns
+    -------
+    Argument
+        arginfo
+    """
+    doc_run_jdata = "Parameters for simplify.json, the first argument of `dpgen simplify`."
+    return Argument("simplify_jdata",
+                    dict,
+                    sub_fields=[
+                        *basic_args(),
+                        # TODO: we may remove sys_configs; it is required in train method
+                        *data_args(),
+                        *general_simplify_arginfo(),
+                        # simplify use the same training method as run
+                        *training_args(),
+                        *fp_args(),
+                    ],
+                    sub_variants=[
+                        fp_style_variant_type_args(),
+                    ],
+                    doc=doc_run_jdata,
+                    )
+
 
 def simplify_mdata_arginfo() -> Argument:
     """Generate arginfo for dpgen simplify mdata.
-    
+
     Returns
     -------
     Argument

examples/simplify/qm7.json

@@ -20,7 +20,6 @@
         "auto"
     ],
     "numb_models": 4,
-    "train_param": "input.json",
     "default_training_param": {
         "model": {
             "type_map": [
@@ -92,11 +91,8 @@
     },
     "use_clusters": true,
     "fp_style": "gaussian",
-    "shuffle_poscar": false,
     "fp_task_max": 1000,
     "fp_task_min": 10,
-    "fp_pp_path": "/home/jzzeng/",
-    "fp_pp_files": [],
     "fp_params": {
         "keywords": "mn15/6-31g** force nosymm scf(maxcyc=512)",
         "nproc": 28,
@@ -105,9 +101,7 @@
     },
     "init_pick_number":100,
     "iter_pick_number":100,
-    "e_trust_lo":1e10,
-    "e_trust_hi":1e10,
-    "f_trust_lo":0.25,
-    "f_trust_hi":0.45,
+    "model_devi_f_trust_lo":0.10,
+    "model_devi_f_trust_hi":0.30,
     "_comment": " that's all "
 }

tests/test_check_examples.py

@@ -9,8 +9,12 @@
 from dpgen.data.arginfo import (
     init_reaction_jdata_arginfo,
 )
+from dpgen.simplify.arginfo import (
+    simplify_jdata_arginfo,
+)
 
 init_reaction_jdata = init_reaction_jdata_arginfo()
+simplify_jdata = simplify_jdata_arginfo()
 
 # directory of examples
 p_examples = Path(__file__).parent.parent / "examples"
@@ -19,6 +23,7 @@
 #   tuple of example list
 input_files = (
     (init_reaction_jdata, p_examples / "init" / "reaction.json"),
+    (simplify_jdata, p_examples / "simplify" / "qm7.json"),
 )