#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ check_vasp.py -- health check of a (running or finished) VASP job. VASP 作业健康检查：收敛情况、受力、磁矩、耗时与常见报错诊断。 Part of DFT-Tutorials (https://github.com/zenggs05/DFT-Tutorials) - MIT License What it reports --------------- * VASP version, task type (IBRION/ISIF/NSW), number of ionic steps and the number of electronic steps of every ionic step (SCF steps hitting NELM are flagged as unconverged); * 'reached required accuracy' / NSW exhausted / job finished or still running; * last energies (TOTEN, energy(sigma->0)), T*S per atom (SIGMA check), E-fermi, external pressure (+ Pulay stress) for ISIF >= 3; * maximum force, taking Selective dynamics flags of POSCAR/CONTCAR into account (the force is projected onto the allowed displacement directions, exact also for non-orthogonal cells) and compared with EDIFFG; * total and largest site magnetic moments (collinear and non-collinear); * elapsed time, average time per ionic step, memory; * WARNING / ADVICE boxes of VASP 6 and known error strings in OUTCAR and in the stdout files (vasp.out, slurm-*.out, *.o, ...), each with a fix hint in Chinese (ZBRENT, BRMIX, EDDDAV, Sub-Space-Matrix, IBZKPT, Tetrahedron, PRICEL, ZPOTRF, time limit, out-of-memory, ...). Usage examples -------------- python check_vasp.py # current directory python check_vasp.py relax/ scf/ band/ # several jobs python check_vasp.py */ --brief # one line per directory python check_vasp.py neb/ # 00..NN image directories are detected python check_vasp.py -o vasp.out --json report.json python check_vasp.py --strict && echo converged # exit code 1 if not converged Exit status: 0 = OK, 1 = errors found (or not converged with --strict), 2 = no VASP output found. Dependencies: python >= 3.9, standard library only (numpy is used if present). """ from __future__ import annotations import argparse import glob import json import math import os import re import sys import time from pathlib import Path __version__ = "1.0.0" # --------------------------------------------------------------------------- # Known error / warning strings (VASP 5.4 - 6.6 and batch systems) -> advice # (key, severity, [substrings], Chinese advice) # --------------------------------------------------------------------------- PATTERNS = [ ("ZBRENT", "error", ["ZBRENT: fatal error in bracketing", "ZBRENT: fatal internal in", "ZBRENT: fatal error: bracketing interval incorrect"], "离子优化线搜索失败：cp CONTCAR POSCAR 后续算；减小 POTIM（如 0.2）；EDIFF 收紧到 1E-6 或更小；" "改用 IBRION=1，或 VTST 的 FIRE（IBRION=3, POTIM=0, IOPT=7）。"), ("BRMIX", "error", ["BRMIX: very serious problems"], "电荷混合严重失败：删除旧 CHGCAR/WAVECAR 重算；检查对称性（ISYM=0 或 SYMPREC=1E-4~1E-6）；" "必要时减小 AMIX/BMIX。"), ("EDDDAV", "error", ["Error EDDDAV: Call to ZHEGV failed", "EDDDAV: Call to ZHEGV failed"], "Davidson 对角化失败：先检查原子是否重叠；换 ALGO=Fast 或 All；ISTART=0 不读 WAVECAR；减小 POTIM。"), ("SUBSPACE", "warning", ["Sub-Space-Matrix is not hermitian in DAV"], "Davidson 子空间矩阵非厄米：偶尔出现可忽略；频繁出现时换 ALGO=Fast/All 并检查结构与步长。"), ("EDDRMM", "error", ["WARNING in EDDRMM: call to ZHEGV failed"], "RMM-DIIS 失败：改用 ALGO=Normal（Davidson），或减小 POTIM、检查结构。"), ("ZPOTRF", "error", ["LAPACK: Routine ZPOTRF failed", "Routine ZPOTRF ZTRTRI"], "波函数正交化失败：多因原子过近或上一步结构异常；检查 CONTCAR，ISTART=0 重启，减小 POTIM，ALGO=Normal。"), ("PSSYEVX", "error", ["ERROR in subspace rotation PSSYEVX", "ERROR in subspace rotation PDSYEVX"], "ScaLAPACK 子空间旋转失败：尝试 ALGO=Normal，或在 INCAR 中设 LSCALAPACK = .FALSE.。"), ("EDDIAG", "error", ["ERROR EDDIAG: Call to routine ZHEEV failed", "ERROR in EDDIAG: call to ZHEEV/ZHEEVX/DSYEV/DSYEVX failed"], "子空间对角化失败：换 ALGO（Normal 与 Fast 互换），检查结构。"), ("EDWAV", "error", ["EDWAV: internal error, the gradient is not orthogonal"], "梯度不正交：试 ALGO=Normal 或 Fast；反复出现时检查编译（降低优化级别）与 MPI/数学库。"), ("IBZKPT", "error", ["internal error in subroutine IBZKPT", "Reciprocal lattice and k-lattice belong to different class of lattices"], "k 网格与晶格对称性不相容（常见于六角/三方体系的偶数 Monkhorst-Pack 网格）：改用 Gamma 中心网格或修改网格。"), ("TETRAHEDRON", "error", ["Tetrahedron method fails", "tetrahedron method fails", "Routine TETIRR needs special values"], "ISMEAR=-5 需要至少 4 个 k 点的 Gamma 中心网格：分子、Gamma 点和 slab 计算改用 ISMEAR=0, SIGMA=0.05。"), ("DENTET", "warning", ["DENTET"], "四面体法无法精确确定费米能级：增加 k 点，或改用 ISMEAR=0。"), ("PRICEL", "error", ["internal error in subroutine PRICEL", "PRICELV: current lattice and primitive lattice are incommensurate"], "结构对称性有噪声：用 spglib/vaspkit 规整结构，调 SYMPREC（1E-4 或 1E-6），必要时 ISYM=0。"), ("ROTMAT", "error", ["Found some non-integer element in rotation matrix", "SGRCON"], "对称操作矩阵非整数（结构对称性噪声）：规整结构，调 SYMPREC，或 ISYM=0。"), ("INV_ROT", "error", ["rotation matrix was not found (increase SYMPREC)"], "找不到逆对称操作：增大 SYMPREC（如 1E-4），或 ISYM=0。"), ("SYMPREC_NOISE", "warning", ["determination of the symmetry of your systems shows a strong"], "对称性判断对 SYMPREC 很敏感：规整坐标或调整 SYMPREC。"), ("POSMAP", "error", ["POSMAP"], "对称等价原子映射失败：多为原子过近或 SYMPREC 不当；检查结构，调整 SYMPREC 或 ISYM=0。"), ("RHOSYG", "error", ["RHOSYG"], "电荷密度对称化失败：ISYM=0 或调整 SYMPREC。"), ("POINT_GROUP", "error", ["group operation missing"], "点群操作缺失：规整结构或 ISYM=0。"), ("BRAVAIS", "error", ["Inconsistent Bravais lattice"], "晶胞与识别出的布拉维格子不一致：用标准化晶胞或调 SYMPREC。"), ("TOO_FEW_BANDS", "error", ["TOO FEW BANDS", "number of bands is not sufficient"], "能带数不足：增大 NBANDS。"), ("HIGH_BAND_OCC", "warning", ["Your highest band is occupied at some k-points"], "最高能带仍有占据：增大 NBANDS（光学、金属、高温 MD 尤其注意）。"), ("NBANDS_CHANGED", "info", ["The number of bands has been changed"], "VASP 为适配并行自动增加了 NBANDS；LOBSTER/LOPTICS/MLFF 等需要一致 NBANDS 时请手动设置为并行组数的整数倍。"), ("READ_ERROR", "error", ["Error reading item"], "INCAR 某标签读取失败：检查拼写与数值；dos2unix 去掉 Windows 换行，不要用 Tab；" "MAGMOM 个数需等于原子数（非共线为 3×NIONS），可用 N*value 简写。"), ("TRIPLE_PRODUCT", "error", ["ERROR: the triple product of the basis vectors"], "晶格矢量构成左手系（三重积为负）：交换两个晶格矢量以及对应的坐标分量。"), ("REAL_OPTLAY", "error", ["REAL_OPTLAY: internal error", "REAL_OPT: internal ERROR"], "实空间投影出错：设 LREAL=.FALSE.（或保持 LREAL=Auto 但检查 ENCUT/PREC）。"), ("RSPHER", "error", ["ERROR RSPHER"], "实空间投影出错：设 LREAL=.FALSE.。"), ("BRIONS", "error", ["BRIONS problems: POTIM should be increased"], "增大 POTIM，或改用 IBRION=1/2。"), ("AMIN", "warning", ["One of the lattice vectors is very long (>50 A), but AMIN"], "存在很长的晶格矢量（厚真空层）：设置 AMIN = 0.01。"), ("SBESSEL", "error", ["SBESSELITER : nicht konvergent"], "球贝塞尔函数迭代不收敛：通常与 LREAL 有关，设 LREAL=.FALSE.。"), ("ELF_KPAR", "error", ["ELF: KPAR>1 not implemented"], "ELF 计算不支持 KPAR>1：设 KPAR=1。"), ("ELF_NCL", "error", ["ELF not implemented for non collinear case"], "ELF 不支持非共线计算：改用共线计算。"), ("DFPT_NCORE", "error", ["PEAD routines do not work for NCORE", "remove the tag NPAR from the INCAR file"], "DFPT/LCALCEPS 不支持 NCORE/NPAR>1：删除 INCAR 中的 NCORE 和 NPAR。"), ("HNFORM", "error", ["HNFORM: k-point generating"], "k 点生成失败：改变 k 网格，或使用标准化晶胞。"), ("KMESH_SYMM", "error", ["Fatal error detecting k-mesh", "Fatal error: unable to match k-point"], "k 网格与对称性不匹配：改用 Gamma 中心网格，或 ISYM=0。"), ("KSHIFT", "error", ["Could not get correct shifts"], "无法得到正确的 k 点平移：改用 Gamma 中心网格。"), ("WAVECAR", "error", ["while reading WAVECAR", "while reading plane"], "WAVECAR 与当前设置（ENCUT/晶胞/k 点）不匹配：删除 WAVECAR 或设 ISTART=0。"), ("SET_CORE_WF", "error", ["internal error in SET_CORE_WF"], "POTCAR 有问题（版本不兼容或文件损坏）：重新生成 POTCAR。"), ("XC_TABLE", "error", ["supplied exchange-correlation table"], "交换关联表与 POTCAR 不匹配：检查 POTCAR 的 LEXCH 与 INCAR 中 GGA/METAGGA。"), ("ALGO_TET", "warning", ["ALGO=A and IALGO=5X tend to fail"], "ALGO=All 与四面体法（ISMEAR=-5）组合不稳定：换 ALGO=Normal 或 ISMEAR=0。"), ("SCF_NOT_CONVERGED", "warning", ["electronic self-consistency was not achieved"], "有离子步的电子步未收敛：增大 NELM，换 ALGO=All/Normal，减小 AMIX，或增加 NBANDS。"), ("IONS_CLOSE", "error", ["distance between some ions is very small"], "原子间距过小：检查结构（重复原子、坐标单位或 Direct/Cartesian 写错）。"), ("VERY_BAD_NEWS", "error", ["VERY BAD NEWS"], "VASP 内部错误：看该行上下文给出的子程序名。"), ("MPI_ABORT", "error", ["BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES", "APPLICATION TERMINATED WITH THE EXIT STRING", "MPI_ABORT was invoked"], "MPI 进程异常退出：看前面的报错；常见原因是内存不足或段错误（作业脚本中 ulimit -s unlimited）。"), ("SEGFAULT", "error", ["forrtl: severe (174)", "SIGSEGV", "Segmentation fault"], "段错误：ulimit -s unlimited；检查内存、编译器与数学库版本是否匹配。"), ("OOM", "error", ["oom-kill", "Out Of Memory", "out-of-memory", "OOM Killed"], "内存不足：减少每节点 MPI 进程数、合理设置 NCORE/KPAR、LREAL=Auto，或申请更多内存/节点。"), ("TIME_LIMIT", "error", ["DUE TO TIME LIMIT", "exceeded limit", "TIME LIMIT EXCEEDED"], "作业超时被杀：cp CONTCAR POSCAR（MD 还要保留速度）后续算，或延长 walltime。"), ] ADVICE_BANNERS = ("W W AA RRRRR", "EEEEEEE RRRRRR", "ADVICE TO THIS USER RUNNING VASP") ART_RE = re.compile(r"^[WANRIG!EO#\s]*$") FLOAT = r"[-+]?(?:\d+\.?\d*|\.\d+)(?:[EeDd][-+]?\d+)?" def fnum(s): try: return float(s.replace("D", "E").replace("d", "e")) except (ValueError, AttributeError): return None # --------------------------------------------------------------------------- # POSCAR (for selective dynamics) # --------------------------------------------------------------------------- def read_poscar_constraints(path: Path): """Return (lattice 3x3 list, flags list[N][3] or None, natoms) or None.""" try: lines = path.read_text(errors="replace").splitlines() scale = [float(x) for x in lines[1].split()[:3]] lat = [[float(x) for x in lines[i].split()[:3]] for i in (2, 3, 4)] i = 5 tok = lines[i].split() if not all(t.isdigit() for t in tok): i += 1 counts = [int(x) for x in lines[i].split()] i += 1 sel = lines[i].strip()[:1] in ("s", "S") if sel: i += 1 i += 1 nat = sum(counts) if len(scale) == 1 and scale[0] > 0: lat = [[x * scale[0] for x in v] for v in lat] elif len(scale) == 3: lat = [[v[k] * scale[k] for k in range(3)] for v in lat] flags = None if sel: flags = [] for j in range(nat): t = lines[i + j].split() flags.append([x.upper().startswith("T") for x in t[3:6]] if len(t) >= 6 else [True] * 3) return lat, flags, nat except (OSError, IndexError, ValueError): return None def project_force(force, lat, flag): """Project a Cartesian force onto the subspace of allowed displacements spanned by the lattice vectors whose selective-dynamics flag is T.""" free = [lat[k] for k in range(3) if flag[k]] if len(free) == 3: return force if not free: return [0.0, 0.0, 0.0] # Gram-Schmidt orthonormal basis of the free directions basis = [] for v in free: w = list(v) for b in basis: d = sum(w[i] * b[i] for i in range(3)) w = [w[i] - d * b[i] for i in range(3)] n = math.sqrt(sum(x * x for x in w)) if n > 1e-10: basis.append([x / n for x in w]) out = [0.0, 0.0, 0.0] for b in basis: d = sum(force[i] * b[i] for i in range(3)) out = [out[i] + d * b[i] for i in range(3)] return out # --------------------------------------------------------------------------- # OSZICAR # --------------------------------------------------------------------------- ELEC_RE = re.compile(r"^\s*([A-Za-z]{2,4})\s*:\s*(\d+)\s+(" + FLOAT + ")") IONIC_RE = re.compile(r"^\s*(\d+)\s+(F=|T=)") def parse_oszicar(path: Path) -> dict: res = {"ionic": [], "nelec": [], "algos": {}, "current_elec": 0} n = 0 with open(path, errors="replace") as fh: for line in fh: m = ELEC_RE.match(line) if m: n += 1 res["algos"][m.group(1)] = res["algos"].get(m.group(1), 0) + 1 continue m = IONIC_RE.match(line) if m: d = {"step": int(m.group(1))} for key, pat in (("F", r"\bF=\s*(" + FLOAT + ")"), ("E0", r"\bE0=\s*(" + FLOAT + ")"), ("T", r"\bT=\s*(" + FLOAT + ")"), ("E", r"\bE=\s*(" + FLOAT + ")"), ("dE", r"d E\s*=\s*(" + FLOAT + ")")): mm = re.search(pat, line) if mm: d[key] = fnum(mm.group(1)) mm = re.search(r"mag=\s*(.*)$", line) if mm: vals = [fnum(x) for x in re.findall(FLOAT, mm.group(1))] vals = [v for v in vals if v is not None] d["mag"] = vals[0] if len(vals) == 1 else vals res["ionic"].append(d) res["nelec"].append(n) n = 0 res["current_elec"] = n return res # --------------------------------------------------------------------------- # OUTCAR (streamed: works for multi-GB MD files) # --------------------------------------------------------------------------- PARAM_RES = { "NIONS": re.compile(r"NIONS\s*=\s*(\d+)"), "NKPTS": re.compile(r"NKPTS\s*=\s*(\d+)"), "NBANDS": re.compile(r"NBANDS\s*=\s*(\d+)"), "NELM": re.compile(r"\bNELM\s*=\s*(\d+)"), "EDIFF": re.compile(r"\bEDIFF\s*=\s*(" + FLOAT + ")"), "EDIFFG": re.compile(r"\bEDIFFG\s*=\s*(" + FLOAT + ")"), "NSW": re.compile(r"\bNSW\s*=\s*(\d+)"), "IBRION": re.compile(r"\bIBRION\s*=\s*(-?\d+)"), "ISIF": re.compile(r"\bISIF\s*=\s*(\d+)"), "ISPIN": re.compile(r"\bISPIN\s*=\s*(\d+)"), "ISMEAR": re.compile(r"\bISMEAR\s*=\s*(-?\d+)"), "SIGMA": re.compile(r"\bSIGMA\s*=\s*(" + FLOAT + ")"), "ENCUT": re.compile(r"\bENCUT\s*=\s*(" + FLOAT + ")"), "IMAGES": re.compile(r"\bIMAGES\s*=\s*(\d+)"), "LSORBIT": re.compile(r"\bLSORBIT\s*=\s*([TF])"), } def scan_patterns(line: str, found: dict, src: str) -> None: for key, sev, subs, _adv in PATTERNS: for s in subs: if s in line: ent = found.setdefault(key, {"count": 0, "line": line.strip()[:160], "sources": set()}) ent["count"] += 1 ent["sources"].add(src) break def _skip_line(line: str) -> bool: st = line.lstrip() return not st or st[0].isdigit() or st[0] in "-.+" def parse_outcar(path: Path) -> dict: r = {"params": {}, "toten": [], "e0": [], "found": {}, "boxes": {}, "loop_real": [], "finished": False, "reached": False, "iterations": 0, "ionic_iter": 0, "forces": None, "magmom_tot": None, "site_mag": None, "efermi": None, "pressure": None, "pulay": None, "ts_last": None, "version": None} header = True nions = None with open(path, errors="replace") as fh: it = iter(fh) for lineno, line in enumerate(it): if r["version"] is None and lineno < 20: m = re.search(r"(vasp\.\d+\.\d+\.\d+)", line) if m: r["version"] = m.group(1) if header: if "Iteration" in line and "(" in line: header = False else: for k, rx in PARAM_RES.items(): m = rx.search(line) if m: r["params"][k] = m.group(1) if "NIONS" in line and nions is None and "NIONS" in r["params"]: nions = int(r["params"]["NIONS"]) if "Iteration" in line and "(" in line: m = re.search(r"Iteration\s*(\d+)\s*\(\s*(\d+)\)", line) if m: r["ionic_iter"] = int(m.group(1)) r["iterations"] += 1 continue if "EENTRO" in line: m = re.search(r"EENTRO\s*=\s*(" + FLOAT + ")", line) if m: r["ts_last"] = fnum(m.group(1)) continue if "FREE ENERGIE OF THE ION-ELECTRON SYSTEM" in line: for _ in range(6): try: nxt = next(it) except StopIteration: break m = re.search(r"TOTEN\s*=\s*(" + FLOAT + ")", nxt) if m: r["toten"].append(fnum(m.group(1))) m = re.search(r"energy\(sigma->0\)\s*=\s*(" + FLOAT + ")", nxt) if m: r["e0"].append(fnum(m.group(1))) break continue if "TOTAL-FORCE" in line and "POSITION" in line: if nions is None: continue try: next(it) # dashes fr = [] for _ in range(nions): t = next(it).split() fr.append([fnum(x) for x in t[3:6]]) if all(None not in f and len(f) == 3 for f in fr): r["forces"] = fr except StopIteration: pass continue if "magnetization (x)" in line and nions: try: rows = [] for _ in range(3): next(it) for _ in range(nions): rows.append(fnum(next(it).split()[-1])) r["site_mag"] = {"x": rows} except (StopIteration, IndexError): pass continue if ("magnetization (y)" in line or "magnetization (z)" in line) and nions and r["site_mag"]: comp = "y" if "(y)" in line else "z" try: rows = [] for _ in range(3): next(it) for _ in range(nions): rows.append(fnum(next(it).split()[-1])) r["site_mag"][comp] = rows except (StopIteration, IndexError): pass continue if "number of electron" in line and "magnetization" in line: parts = line.split() try: idx = parts.index("magnetization") + 1 vals = [v for v in (fnum(x) for x in parts[idx:]) if v is not None] if vals: r["magmom_tot"] = vals[0] if len(vals) == 1 else vals except ValueError: pass continue if "E-fermi" in line: m = re.search(r"E-fermi\s*:\s*(" + FLOAT + ")", line) if m: r["efermi"] = fnum(m.group(1)) continue if "external pressure" in line: m = re.search(r"external pressure\s*=\s*(" + FLOAT + ")", line) p = re.search(r"Pullay stress\s*=\s*(" + FLOAT + ")", line) if m: r["pressure"] = fnum(m.group(1)) if p: r["pulay"] = fnum(p.group(1)) continue if "reached required accuracy" in line: r["reached"] = True continue if "LOOP+" in line: m = re.search(r"real time\s*(" + FLOAT + ")", line) if m: r["loop_real"].append(fnum(m.group(1))) continue if "Elapsed time (sec)" in line: r["elapsed"] = fnum(line.split(":")[-1].strip()) continue if "Total CPU time used (sec)" in line: r["cpu"] = fnum(line.split(":")[-1].strip()) continue if "Maximum memory used (kb)" in line: r["maxmem_kb"] = fnum(line.split(":")[-1].strip().rstrip(".")) continue if "General timing and accounting" in line: r["finished"] = True continue if any(b in line for b in ADVICE_BANNERS): kind = "ADVICE" if "ADVICE" in line else ("ERROR" if "EEEEEEE" in line else "WARNING") text = [] try: while True: nxt = next(it) if nxt.lstrip().startswith("-----") or not nxt.strip().startswith("|"): break inner = nxt.strip().strip("|").strip() if not inner or ART_RE.match(inner): continue text.append(inner) scan_patterns(inner, r["found"], "OUTCAR") except StopIteration: pass msg = " ".join(text[:2])[:150] if text else "(empty box)" key = (kind, msg) r["boxes"][key] = r["boxes"].get(key, 0) + 1 continue if line.lstrip().startswith("WARNING"): key = ("WARNING", line.strip()[:150]) r["boxes"][key] = r["boxes"].get(key, 0) + 1 if not _skip_line(line): scan_patterns(line, r["found"], "OUTCAR") return r def scan_stdout(paths, found: dict) -> None: for p in paths: try: with open(p, errors="replace") as fh: for line in fh: if not _skip_line(line) or "|" in line: scan_patterns(line, found, p.name) except OSError: continue def stdout_candidates(d: Path, explicit): if explicit: return [Path(x) for x in explicit if Path(x).is_file()] names = ["vasp.out", "vasp.log", "vasp_out", "stdout", "out", "OUTPUT", "log", "job.out", "run.log"] c = [d / n for n in names if (d / n).is_file()] for pat in ("slurm-*.out", "*.o[0-9]*", "*.out", "*.log"): for f in sorted(glob.glob(str(d / pat)), key=os.path.getmtime)[-3:]: fp = Path(f) if fp not in c and fp.name not in ("OUTCAR",) and fp.is_file(): c.append(fp) return c # --------------------------------------------------------------------------- # analysis of one directory # --------------------------------------------------------------------------- def analyse(d: Path, args) -> dict: rep = {"dir": str(d), "status": "unknown", "problems": [], "notes": []} outcar = d / "OUTCAR" oszicar = d / "OSZICAR" if not outcar.is_file() and not oszicar.is_file(): rep["status"] = "no-output" return rep osz = parse_oszicar(oszicar) if oszicar.is_file() else None out = parse_outcar(outcar) if outcar.is_file() else None found = out["found"] if out else {} scan_stdout(stdout_candidates(d, args.stdout), found) p = out["params"] if out else {} def ip(k, default=None): try: return int(p[k]) except (KeyError, ValueError): return default def fp(k, default=None): v = fnum(p.get(k)) if k in p else None return default if v is None else v nions, nelm, nsw, ibrion, isif = ip("NIONS"), ip("NELM", 60), ip("NSW", 0), ip("IBRION", -1), ip("ISIF", 2) ediffg = fp("EDIFFG") rep.update(version=out["version"] if out else None, nions=nions, nsw=nsw, ibrion=ibrion, isif=isif, nelm=nelm, ediffg=ediffg, ismear=ip("ISMEAR"), sigma=fp("SIGMA"), encut=fp("ENCUT"), nkpts=ip("NKPTS"), nbands=ip("NBANDS"), ispin=ip("ISPIN")) is_md = ibrion == 0 relax = ibrion in (1, 2, 3) and nsw > 0 # ionic / electronic steps if osz: nion = len(osz["ionic"]) rep["n_ionic"] = nion rep["nelec"] = osz["nelec"] rep["algos"] = osz["algos"] unconv = [i + 1 for i, n in enumerate(osz["nelec"]) if nelm and n >= nelm] rep["scf_unconverged_steps"] = unconv if unconv: rep["problems"].append(f"{len(unconv)} ionic step(s) hit NELM={nelm} (SCF not converged): " f"{unconv[:10]}{' ...' if len(unconv) > 10 else ''}") if osz["ionic"]: last = osz["ionic"][-1] rep["E0_oszicar"] = last.get("E0") rep["F_oszicar"] = last.get("F") if "mag" in last: rep["mag_oszicar"] = last["mag"] if is_md: temps = [x.get("T") for x in osz["ionic"] if x.get("T") is not None] if temps: rep["md_T_last"] = temps[-1] tail = temps[len(temps) // 2:] rep["md_T_mean_2nd_half"] = sum(tail) / len(tail) if osz["current_elec"]: rep["notes"].append(f"{osz['current_elec']} electronic steps after the last ionic step " "(SCF running or interrupted)") else: rep["n_ionic"] = out["ionic_iter"] if out else 0 if out: rep["finished"] = out["finished"] rep["reached"] = out["reached"] if out["e0"]: rep["E0"] = out["e0"][-1] if out["toten"]: rep["TOTEN"] = out["toten"][-1] rep["efermi"] = out["efermi"] if out["ts_last"] is not None and nions: rep["TS_per_atom_meV"] = 1000.0 * out["ts_last"] / nions if abs(rep["TS_per_atom_meV"]) > 1.0 and rep.get("ismear") not in (-5,): rep["problems"].append(f"|T*S| = {abs(rep['TS_per_atom_meV']):.2f} meV/atom > 1 meV: " "SIGMA too large, reduce it or compare energy(sigma->0)") if out["pressure"] is not None: rep["pressure_kB"] = out["pressure"] rep["pulay_kB"] = out["pulay"] if isif >= 3 and relax and abs(out["pressure"]) > 5: rep["notes"].append(f"external pressure {out['pressure']:.2f} kB after a variable-cell run: " "relax again from CONTCAR / raise ENCUT (Pulay stress)") if out.get("elapsed") is not None: rep["elapsed_s"] = out["elapsed"] if out.get("maxmem_kb"): rep["maxmem_MB"] = out["maxmem_kb"] / 1024.0 lr = [x for x in out["loop_real"] if x is not None] if lr: rep["time_per_ionic_s"] = sum(lr) / len(lr) rep["time_ionic_total_s"] = sum(lr) # magnetism if out["magmom_tot"] is not None and (ip("ISPIN", 1) == 2 or p.get("LSORBIT") == "T"): rep["magmom_total"] = out["magmom_tot"] sm = out["site_mag"] if sm: if "y" in sm and "z" in sm: mags = [math.sqrt(x * x + y * y + z * z) for x, y, z in zip(sm["x"], sm["y"], sm["z"])] else: mags = sm["x"] order = sorted(range(len(mags)), key=lambda k: -abs(mags[k])) rep["site_mag_top"] = [(k + 1, round(mags[k], 3)) for k in order[:args.top]] # forces with selective dynamics if out["forces"]: cons = None for nm in ("CONTCAR", "POSCAR"): f = d / nm if f.is_file() and f.stat().st_size > 0: cons = read_poscar_constraints(f) if cons: break forces = out["forces"] if cons and cons[1] is not None and cons[2] == len(forces): lat, flags, _ = cons eff = [project_force(fv, lat, fl) for fv, fl in zip(forces, flags)] rep["n_fixed_dof"] = sum(3 - sum(fl) for fl in flags) else: eff = forces norms = [math.sqrt(sum(x * x for x in v)) for v in eff] k = max(range(len(norms)), key=lambda i: norms[i]) rep["fmax"] = norms[k] rep["fmax_atom"] = k + 1 rep["frms"] = math.sqrt(sum(x * x for x in norms) / len(norms)) if relax and ediffg is not None and ediffg < 0 and rep["fmax"] > abs(ediffg): rep["notes"].append(f"max force {rep['fmax']:.4f} eV/A (atom {k + 1}) > |EDIFFG| = {abs(ediffg):g}") # boxes rep["boxes"] = [{"kind": k[0], "text": k[1], "count": c} for k, c in out["boxes"].items()] try: age = time.time() - outcar.stat().st_mtime rep["outcar_age_s"] = age except OSError: age = None # pattern hits hits = [] for key, sev, _subs, adv in PATTERNS: if key in found: e = found[key] hits.append({"key": key, "severity": sev, "count": e["count"], "line": e["line"], "sources": sorted(e["sources"]), "advice": adv}) rep["hits"] = hits # status errors = [h for h in hits if h["severity"] == "error"] finished = rep.get("finished", False) if relax: if rep.get("reached"): rep["status"] = "converged" elif finished and rep.get("n_ionic", 0) >= nsw: rep["status"] = "NSW-exhausted" rep["problems"].append(f"NSW={nsw} ionic steps used without reaching EDIFFG: " "cp CONTCAR POSCAR and continue") elif finished: rep["status"] = "stopped" else: rep["status"] = "running/killed" elif is_md: rep["status"] = "finished" if finished else "running/killed" else: if finished: rep["status"] = "finished" if not rep.get("scf_unconverged_steps") else "SCF-unconverged" else: rep["status"] = "running/killed" if rep["status"] == "running/killed" and out is not None: age = rep.get("outcar_age_s") if age is not None and age < args.running_age: rep["status"] = "running" else: rep["status"] = "incomplete" if errors: rep["status"] += "+errors" return rep # --------------------------------------------------------------------------- # reporting # --------------------------------------------------------------------------- def fmt_time(s): if s is None: return "-" s = float(s) if s < 120: return f"{s:.1f} s" if s < 7200: return f"{s / 60:.1f} min" return f"{s / 3600:.2f} h" def print_report(rep: dict, args) -> None: print("=" * 78) print(f"Directory / 目录: {rep['dir']}") if rep["status"] == "no-output": print(" no OUTCAR/OSZICAR found / 未找到 OUTCAR 或 OSZICAR") return print(f" VASP version : {rep.get('version') or '?'}") task = "MD" if rep.get("ibrion") == 0 else ("relaxation" if rep.get("ibrion") in (1, 2, 3) and rep.get("nsw", 0) > 0 else f"IBRION={rep.get('ibrion')}") print(f" task : {task} (NSW={rep.get('nsw')}, ISIF={rep.get('isif')}, NIONS={rep.get('nions')}, " f"NKPTS={rep.get('nkpts')}, NBANDS={rep.get('nbands')}, ENCUT={rep.get('encut')})") print(f" status / 状态 : {rep['status']}") nel = rep.get("nelec") or [] print(f" ionic steps : {rep.get('n_ionic', 0)}" + (f" electronic steps: last={nel[-1]}, max={max(nel)}, " f"total={sum(nel)} (NELM={rep.get('nelm')})" if nel else "")) if nel and args.verbose: print(" SCF steps per ionic step: " + " ".join(str(x) for x in nel)) if rep.get("algos"): print(" SCF algorithm : " + ", ".join(f"{k}x{v}" for k, v in rep["algos"].items())) if "E0" in rep: print(f" energy(sigma->0): {rep['E0']:.8f} eV" + (f" TOTEN: {rep['TOTEN']:.8f} eV" if "TOTEN" in rep else "") + (f" per atom: {rep['E0'] / rep['nions']:.6f} eV" if rep.get("nions") else "")) elif rep.get("E0_oszicar") is not None: print(f" E0 (OSZICAR) : {rep['E0_oszicar']:.8f} eV") if "TS_per_atom_meV" in rep: print(f" T*S per atom : {rep['TS_per_atom_meV']:.3f} meV (ISMEAR={rep.get('ismear')}, SIGMA={rep.get('sigma')})") if rep.get("efermi") is not None: print(f" E-fermi : {rep['efermi']:.4f} eV") if "fmax" in rep: extra = f", {rep['n_fixed_dof']} fixed degrees of freedom excluded" if rep.get("n_fixed_dof") else "" print(f" max force : {rep['fmax']:.4f} eV/A on atom {rep['fmax_atom']} (RMS {rep['frms']:.4f}{extra})" + (f" EDIFFG={rep['ediffg']:g}" if rep.get("ediffg") is not None else "")) if "pressure_kB" in rep: print(f" pressure : {rep['pressure_kB']:.2f} kB (Pulay stress {rep.get('pulay_kB') or 0:.2f} kB)") if "magmom_total" in rep: m = rep["magmom_total"] ms = f"{m:.4f}" if isinstance(m, float) else " ".join(f"{x:.4f}" for x in m) print(f" total magmom : {ms} mu_B") elif "mag_oszicar" in rep: m = rep["mag_oszicar"] print(f" total magmom : {m if isinstance(m, float) else m} mu_B (OSZICAR)") if rep.get("site_mag_top"): print(" largest |m_i| : " + ", ".join(f"#{i}: {m}" for i, m in rep["site_mag_top"])) if "md_T_last" in rep: print(f" MD temperature : last {rep['md_T_last']:.1f} K, mean of 2nd half {rep['md_T_mean_2nd_half']:.1f} K") t = [] if "elapsed_s" in rep: t.append(f"elapsed {fmt_time(rep['elapsed_s'])}") if "time_per_ionic_s" in rep: t.append(f"{fmt_time(rep['time_per_ionic_s'])}/ionic step") if "maxmem_MB" in rep: t.append(f"max memory {rep['maxmem_MB']:.0f} MB (root rank)") if "outcar_age_s" in rep and not rep.get("finished"): t.append(f"OUTCAR last modified {fmt_time(rep['outcar_age_s'])} ago") if t: print(" timing : " + "; ".join(t)) for n in rep["notes"]: print(f" note : {n}") for pmsg in rep["problems"]: print(f" PROBLEM : {pmsg}") boxes = rep.get("boxes") or [] if boxes: print(f" VASP WARNING/ADVICE messages ({len(boxes)} distinct):") for b in boxes[: args.max_boxes]: print(f" [{b['kind']} x{b['count']}] {b['text']}") if len(boxes) > args.max_boxes: print(f" ... {len(boxes) - args.max_boxes} more (use --max-boxes)") hits = rep.get("hits") or [] if hits: print(" Known messages / 已知报错与建议:") for h in hits: print(f" [{h['severity'].upper()}] {h['key']} x{h['count']} in {', '.join(h['sources'])}") print(f" > {h['line']}") print(f" 建议: {h['advice']}") elif rep["status"] != "no-output": print(" no known error strings found / 未发现已知报错") def brief_line(rep: dict) -> str: e0 = rep.get("E0", rep.get("E0_oszicar")) e0s = f"{e0:.6f}" if isinstance(e0, float) else "-" fm = f"{rep['fmax']:.4f}" if "fmax" in rep else "-" mag = rep.get("magmom_total", rep.get("mag_oszicar")) if isinstance(mag, list): mag = math.sqrt(sum(x * x for x in mag)) ms = f"{mag:.3f}" if isinstance(mag, float) else "-" errs = ",".join(h["key"] for h in rep.get("hits", []) if h["severity"] == "error") or "-" return (f"{rep['dir'][:28]:<28} {rep['status'][:22]:<22} {rep.get('n_ionic', 0):>5} " f"{e0s:>16} {fm:>9} {ms:>8} {fmt_time(rep.get('elapsed_s')):>10} {errs}") def expand_dirs(dirs): out = [] for d in dirs: p = Path(d) if p.is_file(): p = p.parent if not p.is_dir(): print(f"WARNING: '{d}' is not a directory, skipped", file=sys.stderr) continue if not (p / "OUTCAR").is_file() and not (p / "OSZICAR").is_file(): imgs = sorted(x for x in p.iterdir() if x.is_dir() and re.fullmatch(r"\d{2}", x.name)) if imgs: out += imgs continue out.append(p) return out def build_parser(): ap = argparse.ArgumentParser( prog="check_vasp.py", formatter_class=argparse.RawDescriptionHelpFormatter, description=("Health check of VASP jobs: SCF/ionic convergence, forces (selective dynamics aware), " "magnetisation, timing and known error messages with fixes.\n" "VASP 作业健康检查：收敛、受力（考虑 Selective dynamics）、磁矩、耗时与常见报错的中文修复建议。"), epilog=("Examples / 示例:\n python check_vasp.py\n python check_vasp.py */ --brief\n" " python check_vasp.py relax -o relax/vasp.out --json r.json\n" "Part of DFT-Tutorials (https://github.com/zenggs05/DFT-Tutorials) - MIT License")) ap.add_argument("dirs", nargs="*", default=["."], help="job directories (default .); NEB 00..NN detected") ap.add_argument("-o", "--stdout", action="append", help="VASP stdout file(s) to scan (default: auto-detect)") ap.add_argument("--brief", action="store_true", help="one line per directory / 每个目录一行") ap.add_argument("--json", help="write the full report to this JSON file") ap.add_argument("--strict", action="store_true", help="exit 1 unless every job converged/finished cleanly") ap.add_argument("--top", type=int, default=5, help="number of largest site moments to show (default 5)") ap.add_argument("--max-boxes", type=int, default=8, help="max distinct WARNING/ADVICE boxes shown") ap.add_argument("--running-age", type=float, default=900, help="OUTCAR modified within this many seconds => 'running' (default 900)") ap.add_argument("-v", "--verbose", action="store_true", help="print SCF step count of every ionic step") ap.add_argument("--version", action="version", version=f"%(prog)s {__version__}") return ap def main(argv=None) -> int: try: sys.stdout.reconfigure(errors="replace") except AttributeError: pass args = build_parser().parse_args(argv) dirs = expand_dirs(args.dirs) if not dirs: print("ERROR: no valid directory given", file=sys.stderr) return 2 reports = [] for d in dirs: try: reports.append(analyse(d, args)) except OSError as exc: reports.append({"dir": str(d), "status": "read-error", "problems": [str(exc)], "notes": []}) if args.brief: print(f"{'directory':<28} {'status':<22} {'ionic':>5} {'E0 (eV)':>16} {'Fmax':>9} {'mag':>8} {'elapsed':>10} errors") for r in reports: print(brief_line(r)) else: for r in reports: print_report(r, args) if args.json: def conv(o): if isinstance(o, set): return sorted(o) return str(o) with open(args.json, "w", encoding="utf-8") as fh: json.dump(reports, fh, indent=1, ensure_ascii=False, default=conv) print(f"JSON report written to {args.json}") if all(r["status"] == "no-output" for r in reports): return 2 bad = any("+errors" in r["status"] for r in reports) if args.strict: good = ("converged", "finished") bad = bad or any(r["status"] not in good for r in reports) return 1 if bad else 0 if __name__ == "__main__": sys.exit(main())