File size: 6,299 Bytes
f614769
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
###########################################################################################
# Script for evaluating configurations contained in an xyz file with a trained model for
# energy and force MAE
# This program is distributed under the MIT License (see MIT.md)
###########################################################################################

import argparse

# import ase.data
import ase.io
import torch
from mace import data
from mace.tools import torch_geometric, torch_tools, utils

import matplotlib.pyplot as plt
from sklearn.metrics import r2_score

""" python /home/civil/phd/cez218288/scratch/mace_v_0.3.5/md_simulation/mace/eval_mae.py --configs "/home/civil/phd/cez218288/Benchmarking/MDBENCHGNN/example/lips_1/data/test/botnet.xyz" --model "/scratch/scai/phd/aiz238703/MDBENCHGNN/Repulsive/OutputZBL1/MACE_model_500_lips_ZBL1_swa.model"  --device cuda"""


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser()
    parser.add_argument("--configs", help="path to XYZ configurations", required=True)
    parser.add_argument("--model", help="path to model", required=True)
    # parser.add_argument("--output",help="output path",required=True)
    parser.add_argument(
        "--device",
        help="select device",
        type=str,
        choices=["cpu", "cuda"],
        default="cpu",
    )
    parser.add_argument(
        "--default_dtype",
        help="set default dtype",
        type=str,
        choices=["float32", "float64"],
        default="float64",
    )
    parser.add_argument("--batch_size", help="batch size", type=int, default=1)
    parser.add_argument(
        "--info_prefix",
        help="prefix for energy, forces and stress keys",
        type=str,
        default="MACE_",
    )

    return parser.parse_args()


def plot_r2_score(actual, pred, title="Title"):
    save_dir = "./"  # Ensure this path ends with a slash

    # Calculate R² score
    r2 = r2_score(actual, pred)

    # Create the scatter plot
    plt.scatter(actual, pred)
    plt.xlabel("Actual", fontsize=20, fontweight="bold")
    plt.ylabel("Predicted", fontsize=20, fontweight="bold")
    plt.xticks(fontsize=20, fontweight="bold")
    plt.yticks(fontsize=20, fontweight="bold")

    # Plot the 45-degree line
    min_val = min(min(actual), min(pred))
    max_val = max(max(actual), max(pred))
    plt.plot([min_val, max_val], [min_val, max_val], color="red", linestyle="--")
    plt.title(title, fontsize=20, fontweight="bold")

    # Annotate the R² score on the plot
    plt.text(
        0.05,
        0.95,
        f"R² = {r2:.5f}",
        transform=plt.gca().transAxes,
        fontsize=12,
        verticalalignment="top",
    )

    # Save the plot to the specified location with the title in the filename
    filename = f"{title.replace(' ', '_')}.png"
    plt.savefig(f"{save_dir}{filename}")

    # Show the plot
    plt.show()
    # Clear the current figure to avoid overlap
    plt.clf()


def main():
    args = parse_args()
    torch_tools.set_default_dtype(args.default_dtype)
    device = torch_tools.init_device(args.device)

    # Load model
    model = torch.load(f=args.model, map_location=args.device).to(device)
    model = model.double()

    # Load data and prepare input
    atoms_list = ase.io.read(args.configs, index=":")
    configs = [data.config_from_atoms(atoms) for atoms in atoms_list]

    z_table = utils.AtomicNumberTable([int(z) for z in model.atomic_numbers])

    data_loader = torch_geometric.dataloader.DataLoader(
        dataset=[
            data.AtomicData.from_config(
                config, z_table=z_table, cutoff=float(model.r_max)
            )
            for config in configs
        ],
        batch_size=args.batch_size,
        shuffle=False,
        drop_last=False,
    )

    # Collect data

    # Create counter variables
    counter = 0
    e_mae = 0
    f_mae = 0
    e_rmse = 0
    f_rmse = 0

    Predictions_Fx = []
    Actuals_Fx = []

    Predictions_Fy = []
    Actuals_Fy = []

    Predictions_Fz = []
    Actuals_Fz = []
    for batch in data_loader:
        counter += 1
        batch = batch.to(device)
        output = model(batch.to_dict())
        # temp_e1 = abs(batch["energy"]).mean()
        # temp_f1 = abs(batch["forces"]).mean()

        temp_e = (abs(batch["energy"] - output["energy"])).mean()
        temp_f = (abs(batch["forces"] - output["forces"])).mean()
        temp_re = torch.sqrt(((batch["energy"] - output["energy"]) ** 2).mean())

        temp_rf = torch.sqrt(((batch["forces"] - output["forces"]) ** 2).mean())
        Pred_Forces = output["forces"]
        Actual_Forces = batch["forces"]

        Predictions_Fx += Pred_Forces[:, 0].reshape(-1).detach().cpu().numpy().tolist()
        Actuals_Fx += Actual_Forces[:, 0].reshape(-1).detach().cpu().numpy().tolist()

        Predictions_Fy += Pred_Forces[:, 1].reshape(-1).detach().cpu().numpy().tolist()
        Actuals_Fy += Actual_Forces[:, 1].reshape(-1).detach().cpu().numpy().tolist()

        Predictions_Fz += Pred_Forces[:, 2].reshape(-1).detach().cpu().numpy().tolist()
        Actuals_Fz += Actual_Forces[:, 2].reshape(-1).detach().cpu().numpy().tolist()

        counter += 1
        if counter > 500:
            break

        # print("Batch: ",counter,"\te_mae: ",round((temp_e-temp_e1).item(),3),"\tf_mae: ",round((temp_f-temp_f1).item(),3))
        print(
            "Batch_old: ",
            counter,
            "\te_mae: ",
            round((temp_e).item(), 3),
            "\tf_mae: ",
            round((temp_f).item(), 3),
        )

        e_mae += temp_e  # -temp_e1
        f_mae += temp_f  # -temp_f1

        e_rmse += temp_re  # -temp_e1
        f_rmse += temp_rf  # -temp_f1

    print("||Final Results:||")
    print(
        "E_MAE: ",
        round((e_mae / counter).item(), 3),
        "\t F_MAE: ",
        round((f_mae / (counter)).item(), 3),
    )
    print(
        "E_RMSE: ",
        round((e_rmse / counter).item(), 3),
        "\t F_RMSE: ",
        round((f_rmse / (counter)).item(), 3),
    )

    plot_r2_score(Actuals_Fx, Predictions_Fx, "UpstreamMacelips_Fx")
    plot_r2_score(Actuals_Fy, Predictions_Fy, "UpstreamMacelips_Fy")
    plot_r2_score(Actuals_Fz, Predictions_Fz, "UpstreamMacelips_Fz")


if __name__ == "__main__":
    main()