| import json, os, re, sys, numpy as np, pandas as pd |
| W=sys.argv[1] |
| def legacy(i): |
| p,_,b=i.rpartition("-") |
| if not re.fullmatch(r"[a-z]+",b): return i |
| n=0 |
| for c in b: n=n*26+ord(c)-97 |
| return f"{p or 'mp'}-{n}" |
| core={d["material_id"]:d for d in json.load(open(W+"/core_lemat.json"))} |
| tasks={t["task_id"]:t for t in json.load(open(W+"/tasks_meta.json"))} |
| m=pd.read_parquet(W+"/mp_vs_thermo.parquet") |
| |
| alias=json.load(open(W+"/core_alias.json")) if os.path.exists(W+"/core_alias.json") else {} |
| thermo={d["material_id"]:d for d in json.load(open(W+"/thermo_gga.json"))} |
| for i,parent in alias.items(): |
| if i in core: continue |
| core[i]=core[parent] |
| th=thermo.get(parent) |
| if th and th.get("origins"): |
| m.loc[m.immutable_id==i,"thermo_task"]=th["origins"][0]["task_id"] |
| rows=[] |
| for r in m.itertuples(index=False): |
| d=core.get(r.immutable_id); rec={"immutable_id":r.immutable_id} |
| if d is None: rec["status"]="material_not_found"; rows.append(rec); continue |
| dep={legacy(t) for t in d.get("deprecated_tasks") or []} |
| rec["material_deprecated"]=d["deprecated"] |
| cands=[] |
| for k,ct in (d.get("calc_types") or {}).items(): |
| t=tasks.get(legacy(k)) |
| if t is None or (t.get("output") or {}).get("energy") is None: continue |
| cands.append((legacy(k),t)) |
| rec["n_cand"]=len(cands) |
| best=None |
| for tid,t in cands: |
| dE=t["output"]["energy"]-r.energy |
| if best is None or abs(dE)<abs(best[1]): best=(tid,dE,t) |
| if best is not None: |
| tid,dE,t=best |
| rec.update(lemat_task=tid, lemat_task_dE=dE, lemat_calc_type=t["calc_type"], lemat_batch=t.get("batch_id"), |
| lemat_task_deprecated=tid in dep, lemat_task_vol=t.get("volume"), lemat_task_nsites=t.get("nsites"), |
| lemat_task_hubbards=json.dumps(t["input"].get("hubbards")) if t.get("input") else None, |
| lemat_potcars=",".join(p["titel"] for p in (t.get("input") or {}).get("potcar_spec") or [])) |
| th=r.thermo_task |
| if isinstance(th,str): |
| th=legacy(th); t=tasks.get(th) |
| rec.update(mp_task=th, mp_task_found=t is not None) |
| if t is not None: |
| rec.update(mp_calc_type=t["calc_type"], mp_energy=t["output"]["energy"], mp_nsites=t["nsites"], mp_task_vol=t.get("volume"), |
| mp_batch=t.get("batch_id"), mp_task_deprecated=th in dep, |
| mp_potcars=",".join(p["titel"] for p in (t.get("input") or {}).get("potcar_spec") or []), |
| mp_hubbards=json.dumps(t["input"].get("hubbards")) if t.get("input") else None) |
| rows.append(rec) |
| f=pd.DataFrame(rows) |
| if "status" not in f: f["status"]=None |
| f=m[["immutable_id","energy","nsites","vol","epa","d_med","d_min","d_max","dft_hull","elements","chemical_formula_reduced","fmax","p_kbar","last_modified","uncorrected_energy_per_atom"]].merge(f,on="immutable_id") |
| f["identified"]=f.lemat_task_dE.abs()<1e-3 |
| f["mp_epa"]=f.mp_energy/f.mp_nsites |
| f["dE_mp"]=f.epa-f.mp_epa |
| f["same_task"]=f.lemat_task==f.mp_task |
| f["lemat_run"]=f.lemat_calc_type.str.split(" ").str[0]; f["mp_run"]=f.mp_calc_type.str.split(" ").str[0] |
| f["dvol_mp"]=(f.vol/f.nsites)/(f.mp_task_vol/f.mp_nsites)-1 |
| f["lemat_task_deprecated"]=f.lemat_task_deprecated.astype("boolean").fillna(False).astype(bool) |
| f["mlip_epa"]=f.epa-f.d_med |
| f["mp_vs_mlip"]=f.mp_epa-f.mlip_epa |
| rm=(f.lemat_run!=f.mp_run)&f.mp_run.notna() |
| f["grp"]=np.select([f.status.eq("material_not_found").values,(~f.identified).values,f.same_task.values, |
| (f.lemat_task_deprecated&rm).values,f.lemat_task_deprecated.values,rm.values], |
| ["not_found","unidentified","same_task","dep_runmismatch","deprecated","runmismatch"],"live_other_task") |
| f.to_parquet(W+"/identify.parquet") |
| pd.set_option("display.width",250) |
| print("identified:",f.identified.sum(),"of",len(f)); print(f.status.value_counts(dropna=False).to_dict() if "status" in f else "") |
| print(pd.crosstab(f.lemat_calc_type,f.lemat_task_deprecated,margins=True)) |
| print(pd.crosstab([f.lemat_run,f.lemat_task_deprecated],f.mp_run,margins=True)) |
| print("same task:",f.same_task.sum()) |
|
|
