Save local changes before rebase

Author: JunnHuo

applications/application_configs/interface/ase.yaml renamed to applications/application_configs/calculator/ase.yaml

@@ -7,8 +7,8 @@ defaults:
   - _self_
   - run: default
   - model: load_model
-  - task: md
-  - interface: ase
+  - task: md                              # opt, md
+  - calculator: ase
   - system: load_system                   # load_system, ase_create
   - logger: default
   - override hydra/job_logging: disabled

applications/application_configs/md_ase.yaml

@@ -7,8 +7,8 @@ defaults:
   - _self_
   - run: default
   - model: load_model
-  - task: opt
-  - interface: ase
+  - task: opt                            # opt, md
+  - calculator: ase
   - system: ase_create                   # load_system, ase_create
   - logger: default
   - override hydra/job_logging: disabled

applications/application_configs/optimizer_ase.yaml

@@ -7,8 +7,8 @@
 from hydra.utils import instantiate 
 from omegaconf import DictConfig, OmegaConf
 
-from ppmat.predict import PPMatPredictor
-from ppmat.predict.structures import prepare_structures
+from ppmat.predictor import PPMatPredictor
+from ppmat.predictor.structures import build_init_structures
 from ppmat.utils import logger
 
 
@@ -51,7 +51,7 @@ def main(cfg: DictConfig):
         predictor.load_inference_model( interface_type=None )
 
     # Load structures
-    files, structures = prepare_structures(cfg, predictor)
+    files, structures = build_init_structures(cfg, predictor)
 
     if cfg.get("task") is not None:
         # Initialize the task

applications/main.py

@@ -28,8 +28,8 @@
 from ppmat.utils import save_load
 from ppmat.utils import logger
 
-class PPMatPredictor:
-    """PPMaterial predictor.
+class Predictor:
+    """
 
     This class provides an interface for predicting properties of crystalline
     structures using pre-trained deep learning models.

ppmat/interface/ase.py renamed to ppmat/calculator/ase.py

@@ -0,0 +1,59 @@
+import os
+import os.path as osp
+import sys
+
+__dir__ = os.path.dirname(os.path.abspath(__file__))  # ruff: noqa
+sys.path.insert(0, os.path.abspath(os.path.join(__dir__, "..")))  # ruff: noqa
+
+from ppmat.utils import logger
+from ppmat.datasets import build_dataloader
+from ppmat.datasets import set_signal_handlers
+
+
+if __name__ == "__main__":
+    
+    # init logger
+    logger_path = osp.join(sys.path[0], "test/create_dataset.log")
+    logger.init_logger(log_file=logger_path)
+    logger.info(f"Logger saved to {logger_path}")
+    logger.info(f"Test create new dataset.")
+
+    set_signal_handlers()
+    data_cfg = {
+        "dataset": {
+            "__class_name__": "JarvisDataset",
+            "__init_params__": {
+                "path": osp.join(sys.path[0], "data/jarvis"),
+                "jarvis_data_name": "dft_2d",
+                "property_names": "formation_energy_peratom",  # 这里改成你实际需要的标签名
+                "build_structure_cfg": {
+                    "format": "jarvis",
+                    "num_cpus": 10,
+                },
+                "build_graph_cfg": {
+                    "__class_name__": "FindPointsInSpheres",
+                    "__init_params__": {
+                        "cutoff": 4.0,
+                        "num_cpus": 10,
+                    },
+                },
+                "cache_path": osp.join(sys.path[0], "data/jarvis"),
+                # "overwrite": True,
+            },
+            "num_workers": 4,
+            "use_shared_memory": False,
+        },
+        "sampler": {
+            "__class_name__": "BatchSampler",
+            "__init_params__": {
+                "shuffle": True,
+                "drop_last": True,
+                "batch_size": 10,
+            },
+        },
+    }
+    logger.info("Train data config:\n" + 
+                "\n".join( f"{k}: {v}" for k, v in data_cfg.items() )
+            )
+
+    train_loader = build_dataloader(data_cfg)

ppmat/predict/__init__.py renamed to ppmat/predictor/__init__.py

@@ -0,0 +1,60 @@
+import os
+import os.path as osp
+import sys
+
+__dir__ = os.path.dirname(os.path.abspath(__file__))  # ruff: noqa
+sys.path.insert(0, os.path.abspath(os.path.join(__dir__, "..")))  # ruff: noqa
+
+from ppmat.utils import logger
+from ppmat.datasets import build_dataloader
+from ppmat.datasets import set_signal_handlers
+
+
+if __name__ == "__main__":
+    
+    # init logger
+    logger_path = osp.join(sys.path[0], "test/create_dataset.log")
+    logger.init_logger(log_file=logger_path)
+    logger.info(f"Logger saved to {logger_path}")
+    logger.info(f"Test create new dataset.")
+
+    set_signal_handlers()
+    train_data_cfg = {
+        "dataset": {
+            "__class_name__": "MP2024Dataset",
+            # "__class_name__": "MP2018Dataset",
+            "__init_params__": {
+                "path": osp.join(sys.path[0], "data/mp2024_train_130k/mp2024_train.txt"),
+                # "path": osp.join(sys.path[0], "data/mp2018_train_60k/mp.2018.6.1_val.json"),
+                "property_names": ["formation_energy_per_atom"],  # 这里改成你实际需要的标签名
+                "build_structure_cfg": {
+                    "format": "dict",
+                    "num_cpus": 10,
+                },
+                "build_graph_cfg": {
+                    "__class_name__": "FindPointsInSpheres",
+                    "__init_params__": {
+                        "cutoff": 4.0,
+                        "num_cpus": 10,
+                    },
+                },
+                "cache_path": osp.join(sys.path[0], "data/mp2024_train_130k_cache_find_points_in_spheres_cutoff_4/mp2024_train"),
+                # "cache_path": osp.join(sys.path[0], "data/mp2018_train_60k_cache_find_points_in_spheres_cutoff_4/mp.2018.6.1_val"),
+            },
+            "num_workers": 4,
+            "use_shared_memory": False,
+        },
+        "sampler": {
+            "__class_name__": "BatchSampler",
+            "__init_params__": {
+                "shuffle": True,
+                "drop_last": True,
+                "batch_size": 10,
+            },
+        },
+    }
+    logger.info("Train data config:\n" + 
+                "\n".join( f"{k}: {v}" for k, v in train_data_cfg.items() )
+            )
+
+    train_loader = build_dataloader(train_data_cfg)