Updating my feature branch

adslab_parse_stats.py

@@ -0,0 +1,120 @@
+import re
+import json
+from collections import defaultdict
+from pathlib import Path
+
+import matplotlib.cm as cmx
+import matplotlib.colors as colors
+import matplotlib.pyplot as plt
+import numpy as np
+from minydra import resolved_args
+from tqdm import tqdm
+
+if __name__ == "__main__":
+
+    args = resolved_args()
+
+    assert args.file is not None
+    assert Path(args.file).exists()
+    assert Path(args.file).is_file()
+
+    with open(args.file, "r") as f:
+        lines = f.read()
+
+    samples = [
+        s
+        for s in lines.split(
+            "------------------------------\n------------------------------"
+        )
+        if "Actions to Data" in s and "ABORTING" not in s
+    ]
+
+    times = defaultdict(list)
+    metadatas = []
+    time_regex = re.compile(r"(.*) \| Done! \((.*)s\)")
+    total_adsorbed_regex = re.compile(r"Total adsorbed_surfaces: (\d+)")
+    non_reasonable_regex = re.compile(r"Non reasonable configs: (\d+)/(\d+)")
+
+    metadata_regexs = {
+        "adsorbate_id": re.compile(
+            r"args(?:\.actions|)\.adsorbate_id is None, choosing (\d+)"
+        ),
+        "adsorbate_desc": re.compile(r"# Selected adsorbate: (.+)"),
+        "bulk_id": re.compile(r"args\.actions\.bulk_id is None, choosing (\d+)"),
+        "bulk_desc": re.compile(r"# Selected bulk: (.+)"),
+        "surface_id": re.compile(r"args\.actions\.surface_id is None, choosing (\d+)"),
+        "surface_desc": re.compile(r"# Selected surface: (.+)"),
+        "bond_indices": re.compile(r"bond_indices: (.+)"),
+    }
+
+    keys = []
+    time_keys = []
+    for s, sample in tqdm(enumerate(samples), total=len(samples)):
+        metadatas.append({})
+        matches = time_regex.findall(sample)
+        if not time_keys:
+            time_keys = set([k.strip() for k, _ in matches] + ["Actions to Data"])
+        matches += [
+            (
+                "Total adsorbed_surfaces",
+                int(total_adsorbed_regex.findall(sample)[0]),
+            ),
+            (
+                "Proportion of non reasonable adsorbed_surfaces",
+                float(non_reasonable_regex.findall(sample)[0][0])
+                / float(non_reasonable_regex.findall(sample)[0][1]),
+            ),
+        ]
+
+        for k, v in matches:
+            k = k.strip()
+            if "Actions to Data" in k:
+                k = "Actions to Data"
+            times[k].append(float(v))
+            metadatas[-1][k] = float(v)
+            if s == 0:
+                keys.append(k)
+        for name, reg in metadata_regexs.items():
+            meta = reg.findall(sample)[0]
+            if "id" in name:
+                meta = int(meta)
+            metadatas[-1][name] = meta
+
+    means = {k: m for k, v in times.items() if ((m := np.mean(v)) > 0.1)}
+    stds = {k: np.std(v) for k, v in times.items() if k in means}
+    keys = [k for k in keys if k in means]
+
+    cmap = plt.get_cmap("viridis")
+    cnorm = colors.Normalize(vmin=0, vmax=len(samples))
+    scalar_map = cmx.ScalarMappable(norm=cnorm, cmap=cmap)
+
+    n_plots = len(means.keys())
+
+    ncols = args.plot_ncols or 3
+    nrows = n_plots // ncols
+    if n_plots % ncols != 0:
+        nrows += 1
+
+    fig, axs = plt.subplots(nrows, ncols, figsize=(ncols * 5, nrows * 4))
+
+    for i, k in tqdm(enumerate(keys), total=len(keys)):
+        ax = axs.flat[i]
+        bars = ax.bar(range(len(times[k])), times[k])
+        for b, bar in enumerate(bars):
+            bar.set_color(scalar_map.to_rgba(b))
+
+        title = k
+        if k in time_keys:
+            title += f" ({means[k]:.2f}s +/- {stds[k]:.2f}s)"
+        else:
+            title += " (count)"
+
+        ax.set_title(title, fontsize=8)
+        ax.xaxis.set_tick_params(labelsize=6)
+        ax.yaxis.set_tick_params(labelsize=6)
+
+    plt.suptitle(f"Time (s) for operations or printed counts ({len(samples)} samples)")
+
+    plt.savefig(args.out_png or f"{Path(args.file).stem}.png", dpi=150)
+    with open(args.out_json or f"{Path(args.file).stem}.json", "w") as f:
+        json.dump(metadatas, f)

configs/sample/defaults.yaml

@@ -0,0 +1,115 @@
+# --------------------------------------------------------------
+# -----  minydra default args values for sample_adslab.py  -----
+# --------------------------------------------------------------
+
+# ------------------
+# -----  Data  -----
+# ------------------
+paths:
+  # path to the bulk_db_flat pickle file
+  bulk_db_flat: /network/projects/_groups/ocp/oc20/dataset-creation/bulk_db_flat_2021sep20.pkl
+
+  # path to the adsorbate_db pickle file
+  adsorbate_db: /network/projects/_groups/ocp/oc20/dataset-creation/adsorbate_db_2021apr28.pkl
+
+  # path to the precomputed_structures pickle file with all surfaces
+  precomputed_structures: /network/projects/_groups/ocp/oc20/dataset-creation/precomputed_surfaces_2021Sep20
+# ------------------------------------
+# -----  Adslab parametrization  -----
+# ------------------------------------
+
+# random seed
+seed: 123
+
+# number of runs
+nruns: 1
+
+actions:
+  # adsorbate smiles representation (see end of this file for reference)
+  # null -> sample uniformly
+  adsorbate_smiles: null # H2O
+
+  # index of the bulk in bulk_db_flat.
+  # null -> sample uniformly
+  bulk_id: null
+
+  # index of the surface to select for a given bulk
+  # null -> sample uniformly
+  surface_id: null
+
+  # index of the adsorption site to select for a given surface
+  # can be -1 (=all), a list of ints or a single int
+  binding_site_index: -1
+
+
+# whether or not to use pre-computed surfaces.
+# if not they will be computed on the fly but it takes
+use_precomputed_surfaces: true
+
+# Loader animation
+animate: false
+# Ignore loader prints
+no_loader: false
+
+# prints
+verbose: 0
+
+# avaliable adsorbates smiles (82):
+# {chemical_formula: smiles}
+# {  'O': ['*O'],
+#    'H': ['*H'],
+#    'HO': ['*OH'],
+#    'H2O': ['*OH2'],
+#    'C': ['*C'],
+#    'CO': ['*CO'],
+#    'CH': ['*CH'],
+#    'CHO': ['*CHO', '*COH'],
+#    'CH2': ['*CH2'],
+#    'CH2O': ['*CH2*O', '*CHOH'],
+#    'CH3': ['*CH3'],
+#    'CH3O': ['*OCH3', '*CH2OH'],
+#    'CH4': ['*CH4'],
+#    'CH4O': ['*OHCH3'],
+#    'C2': ['*C*C'],
+#    'C2O': ['*CCO'],
+#    'C2H': ['*CCH'],
+#    'C2HO': ['*CHCO', '*CCHO'],
+#    'C2HO2': ['*COCHO'],
+#    'C2H2O': ['*CCHOH', 'CH2*CO', '*CHCHO', 'CH*COH'],
+#    'C2H2': ['*CCH2', '*CH*CH'],
+#    'C2H2O2': ['*COCH2O', '*CHO*CHO', '*COHCHO', '*COHCOH'],
+#    'C2H3': ['*CCH3', '*CHCH2'],
+#    'C2H3O': ['*COCH3', '*OCHCH2', '*COHCH2', '*CHCHOH', '*CCH2OH'],
+#    'C2H3O2': ['*CHOCHOH', '*COCH2OH', '*COHCHOH'],
+#    'C2H4': ['*CH2*CH2'],
+#    'C2H4O': ['*OCHCH3', '*COHCH3', '*CHOHCH2', '*CHCH2OH'],
+#    'C2H4O2': ['*OCH2CHOH', '*CHOCH2OH', '*COHCH2OH', '*CHOHCHOH'],
+#    'C2H5': ['*CH2CH3'],
+#    'C2H5O': ['*OCH2CH3', '*CHOHCH3', '*CH2CH2OH'],
+#    'C2H5O2': ['*CHOHCH2OH'],
+#    'C2H6O': ['*OHCH2CH3'],
+#    'C2H8N2': ['*NH2N(CH3)2'],
+#    'C2H6N2O': ['*ONN(CH3)2'],
+#    'CH4N2O': ['*OHNNCH3'],
+#    'CH3N2': ['*NNCH3'],
+#    'HNO': ['*ONH'],
+#    'H2N2': ['*NHNH'],
+#    'H4N2': ['*NHN2'],
+#    'HN2': ['*N*NH'],
+#    'N2O3': ['*ONNO2'],
+#    'N2O4': ['*NO2NO2'],
+#    'N2O': ['*N*NO'],
+#    'N2': ['*N2'],
+#    'H2N2O': ['*ONNH2'],
+#    'H2N': ['*NH2'],
+#    'H3N': ['*NH3'],
+#    'HN2O': ['*NONH'],
+#    'HN': ['*NH'],
+#    'NO2': ['*NO2'],
+#    'NO': ['*NO'],
+#    'N': ['*N'],
+#    'NO3': ['*NO3'],
+#    'H3NO': ['*OHNH2'],
+#    'HNO2': ['*ONOH'],
+#    'CN': ['*CN']
+# }