labplan 0.1.0__py3-none-any.whl

labplan/__init__.py

@@ -0,0 +1,50 @@
+"""labplan: plan the measurement, fit the model, keep the record --
+for any lab model.
+
+Nine research packages in this organization ended up needing the same
+loop: predict the error bars of a planned measurement before spending
+the instrument time, choose the most informative settings, fit the
+model to the data with honest uncertainties, and refuse -- with an
+explanation -- whenever the design cannot identify the parameters.
+labplan is that loop extracted into one dependency-light package that
+works with ANY forward model you can write as a Python function: a
+spin resonance versus field, a critical current versus temperature, a
+transmittance versus wavelength, a sensor reading versus anything.
+
+The promise that makes planning meaningful: the planner's error bars
+and the fitter's error bars are the SAME matrix, (J^T W J)^-1 -- so
+what is promised before the measurement is what is reported after,
+exactly, whenever the model describes the data. On top of the
+model-based statistics, split conformal prediction supplies
+distribution-free intervals with an exact finite-sample guarantee,
+for the day the model is wrong in ways nobody modeled. And every fit
+can be turned into an audit record that states what was fitted, to
+exactly which data (sha256), by which software versions, when.
+
+Statistics sources, stated once: weighted least squares and Fisher
+information are textbook (e.g. Cox & Hinkley, Theoretical Statistics
+(1974)); D-optimal design follows F. Pukelsheim, Optimal Design of
+Experiments, SIAM (2006); split conformal prediction follows Vovk,
+Gammerman & Shafer (2005), Lei et al., J. Am. Stat. Assoc. 113, 1094
+(2018) and Angelopoulos & Bates, arXiv:2107.07511. Every statistical
+claim in the test suite is pinned to a closed form, an exact
+identity, or seeded simulation against an exact formula -- never a
+stored number.
+"""
+from .model import Model
+from .fit import FitResult, fit
+from .plan import design, information, repeats_for
+from .conformal import (conformal_interval, conformal_quantile,
+                        coverage_exact)
+from .report import audit_record, report_text
+from .records import load_measurements_csv, save_measurements_csv
+
+__version__ = "0.1.0"
+__all__ = [
+    "Model", "FitResult", "fit",
+    "information", "design", "repeats_for",
+    "conformal_quantile", "conformal_interval", "coverage_exact",
+    "audit_record", "report_text",
+    "save_measurements_csv", "load_measurements_csv",
+    "__version__",
+]

labplan/conformal.py

@@ -0,0 +1,91 @@
+"""Distribution-free prediction intervals with an exact guarantee.
+
+The model-based error bars elsewhere in this package assume the model
+is right and the noise is Gaussian. Split conformal prediction makes
+a different, weaker promise that holds with NO such assumptions: hold
+out calibration measurements, score how wrong the model was on each
+(for instance |measured - predicted|), and take
+
+    q = the ceil((n + 1)(1 - alpha))-th smallest of the n scores.
+
+Then a new measurement, exchangeable with the calibration set, falls
+within q of its prediction with probability at least 1 - alpha --
+exactly, at finite n, whatever the model and whatever the noise (V.
+Vovk, A. Gammerman and G. Shafer, Algorithmic Learning in a Random
+World, Springer (2005); J. Lei et al., J. Am. Stat. Assoc. 113, 1094
+(2018); A. N. Angelopoulos and S. Bates, arXiv:2107.07511). For
+continuous scores the coverage is also at most 1 - alpha + 1/(n + 1),
+and its exact value is ceil((n + 1)(1 - alpha)) / (n + 1) -- a rank
+statement the tests verify by seeded simulation against the closed
+form.
+
+Honest limits, stated plainly: the guarantee is marginal (on average
+over calibration sets and test points, not conditional on a
+particular x), and it needs exchangeability -- calibration data from
+last month's instrument state do not certify next month's drift.
+"""
+from __future__ import annotations
+
+import numpy as np
+
+__all__ = ["conformal_quantile", "conformal_interval",
+           "coverage_exact"]
+
+
+def conformal_quantile(scores, alpha=0.1):
+    """The split-conformal quantile of held-out error scores.
+
+    scores : (n,) nonnegative error scores of held-out calibration
+        measurements (e.g. |measured - predicted|).
+    alpha : miscoverage level (0.1 = 90% intervals).
+
+    Returns q such that |new error| <= q with probability >= 1-alpha.
+    Refuses when n is too small for the level: the rank
+    ceil((n+1)(1-alpha)) must exist among n scores, which needs
+    n >= (1-alpha)/alpha.
+    """
+    s = np.asarray(scores, dtype=float).ravel()
+    if s.size < 1 or not np.all(np.isfinite(s)) or np.any(s < 0.0):
+        raise ValueError("scores must be finite and >= 0 (use "
+                         "absolute errors)")
+    a = float(alpha)
+    if not (0.0 < a < 1.0):
+        raise ValueError("alpha must lie in (0, 1)")
+    n = s.size
+    k = int(np.ceil((n + 1) * (1.0 - a)))
+    if k > n:
+        need = int(np.ceil((1.0 - a) / a))
+        raise ValueError(
+            f"{n} calibration scores cannot certify level "
+            f"{1 - a:.3g}: the required rank {k} exceeds n. Collect "
+            f"at least {need} calibration measurements, or lower the "
+            "confidence")
+    return float(np.sort(s)[k - 1])
+
+
+def conformal_interval(prediction, q):
+    """The interval the guarantee applies to: prediction +/- q.
+
+    Returns (lo, hi) arrays matching the prediction's shape.
+    """
+    pred = np.asarray(prediction, dtype=float)
+    qv = float(q)
+    if not (np.isfinite(qv) and qv >= 0.0):
+        raise ValueError("q must be finite and >= 0")
+    return pred - qv, pred + qv
+
+
+def coverage_exact(n, alpha):
+    """The exact marginal coverage for continuous scores:
+    ceil((n+1)(1-alpha)) / (n+1). It always lies in
+    [1-alpha, 1-alpha + 1/(n+1)] -- the two-sided guarantee of the
+    references, recovered as arithmetic."""
+    n = int(n)
+    a = float(alpha)
+    if n < 1 or not (0.0 < a < 1.0):
+        raise ValueError("need n >= 1 and alpha in (0, 1)")
+    k = int(np.ceil((n + 1) * (1.0 - a)))
+    if k > n:
+        raise ValueError("level not certifiable at this n (see "
+                         "conformal_quantile)")
+    return k / (n + 1.0)

labplan/fit.py

@@ -0,0 +1,167 @@
+"""Fit the model to measured data, with error bars that mean it.
+
+Levenberg-Marquardt weighted least squares in pure NumPy: no
+dependency beyond the one this package already has. With measurement
+errors supplied, the parameter covariance is the exact known-noise
+result (J^T W J)^-1 and a chi-squared consistency check is reported;
+without them, the error bars are scaled from the residual scatter,
+which needs at least one spare measurement -- stated, not hidden.
+
+A design that cannot tell the parameters apart is refused with an
+explanation, never silently pseudo-inverted; a fit that does not
+converge raises instead of returning the last iterate with a
+well-formatted covariance.
+"""
+from __future__ import annotations
+
+import dataclasses
+import hashlib
+
+import numpy as np
+
+from .model import Model, _settings
+from .stats import SINGULAR_MSG, check_sigmas, invert_information
+
+__all__ = ["FitResult", "fit"]
+
+
+@dataclasses.dataclass
+class FitResult:
+    """Result of `fit`.
+
+    values : fitted value of each parameter, by name.
+    sigma : 1-sigma uncertainty of each parameter, by name.
+    theta, cov : the same as arrays, in `param_names` order.
+    chi2, chi2_dof : goodness of fit (chi2 is None when no
+        measurement errors were given).
+    condition_number : of the unit-free information matrix.
+    data_digest : sha256 of the (x, y, sigma) arrays -- the exact
+        data this result answers for, recorded for the audit trail.
+    model_name, reference : carried from the model.
+    """
+
+    values: dict
+    sigma: dict
+    theta: np.ndarray
+    cov: np.ndarray
+    chi2: float
+    chi2_dof: int
+    n_points: int
+    condition_number: float
+    n_iter: int
+    model_name: str
+    reference: str
+    data_digest: str
+
+
+def data_digest(x, y=None, sigmas=None):
+    """sha256 over the byte content of the data arrays, so a report
+    can say exactly which data it answers for."""
+    h = hashlib.sha256()
+    for a in (x, y, sigmas):
+        if a is None:
+            h.update(b"\x00none\x00")
+        else:
+            arr = np.ascontiguousarray(np.asarray(a, dtype=float))
+            h.update(str(arr.shape).encode())
+            h.update(arr.tobytes())
+    return h.hexdigest()
+
+
+def fit(model: Model, x, y, theta0, sigmas=None, max_iter=200,
+        tol=1e-12):
+    """Weighted least-squares fit of `model` to measured data.
+
+    model : a `Model`.
+    x : (n, d) measurement settings (or (n,) for one setting).
+    y : (n,) measured readings.
+    theta0 : starting parameter vector, in `param_names` order.
+    sigmas : optional 1-sigma measurement errors (scalar or (n,)).
+
+    Returns a `FitResult`. Refuses non-identifiable designs, too few
+    points, and non-convergence, each with an explanation.
+    """
+    x = _settings(x)
+    y = np.asarray(y, dtype=float).ravel()
+    n, p = x.shape[0], model.n_params
+    if y.size != n:
+        raise ValueError("need one measured reading per settings row")
+    if not np.all(np.isfinite(y)):
+        raise ValueError("measured readings must be finite")
+    sig = check_sigmas(sigmas, n)
+    if sig is None and n < p + 1:
+        raise ValueError(f"{n} measurements cannot determine {p} "
+                         "parameters and an error scale; add points "
+                         "or supply sigmas")
+    if n < p:
+        raise ValueError(f"{n} measurements cannot determine {p} "
+                         "parameters")
+    w = np.ones(n) if sig is None else 1.0 / sig
+    th = np.asarray(theta0, dtype=float).ravel()
+    if th.size != p or not np.all(np.isfinite(th)):
+        raise ValueError(f"theta0 must be {p} finite starting values")
+
+    def cost(t):
+        r = (model.predict(t, x) - y) * w
+        return r, float(r @ r)
+
+    r, c = cost(th)
+    mu = 1e-3
+    converged = False
+    it = 0
+    c_old = c
+    for it in range(1, max_iter + 1):
+        jac = model.jacobian(th, x) * w[:, None]
+        jtj = jac.T @ jac
+        g = jac.T @ r
+        stepped = False
+        for _ in range(60):
+            try:
+                dth = np.linalg.solve(
+                    jtj + mu * np.diag(np.maximum(np.diag(jtj),
+                                                  1e-30)), -g)
+            except np.linalg.LinAlgError:
+                mu *= 10.0
+                continue
+            try:
+                r_new, c_new = cost(th + dth)
+            except ValueError:
+                mu *= 10.0
+                continue
+            if c_new <= c:
+                th, r, c_old, c = th + dth, r_new, c, c_new
+                mu = max(mu / 10.0, 1e-15)
+                stepped = True
+                break
+            mu *= 10.0
+        if not stepped:
+            converged = True
+            break
+        if abs(c_old - c) <= tol * (1.0 + c) and \
+                float(np.max(np.abs(dth))) \
+                <= 1e-10 * (1.0 + float(np.max(np.abs(th)))):
+            converged = True
+            break
+    if not converged:
+        raise RuntimeError("fit did not converge; check the starting "
+                           "values and that the readings actually "
+                           "vary with the parameters")
+
+    jac = model.jacobian(th, x) * w[:, None]
+    fisher = jac.T @ jac
+    ok, cond, cov, _ = invert_information(fisher, model.param_names)
+    if not ok:
+        raise ValueError(SINGULAR_MSG)
+    if sig is None:
+        cov = cov * (c / (n - p))
+        chi2, chi2_dof = None, None
+    else:
+        chi2, chi2_dof = c, n - p
+    err = np.sqrt(np.diag(cov))
+    return FitResult(
+        values={k: float(v) for k, v in zip(model.param_names, th)},
+        sigma={k: float(s) for k, s in zip(model.param_names, err)},
+        theta=th.copy(), cov=cov, chi2=chi2, chi2_dof=chi2_dof,
+        n_points=n, condition_number=cond, n_iter=it,
+        model_name=model.name, reference=model.reference,
+        data_digest=data_digest(x, y, sig))

labplan/model.py

@@ -0,0 +1,115 @@
+"""The one thing every calibration needs: a forward model with a name,
+named parameters, and a source.
+
+A `Model` wraps YOUR prediction function -- any Python callable that
+maps a parameter vector and measurement settings to predicted values.
+That is the whole requirement: if you can predict what your instrument
+would read, labplan can plan the measurement, fit the parameters, and
+keep the record. The model carries its provenance (`reference`) the
+same way every physical constant in this organization does: it is
+required, on purpose, because a fit whose model nobody can trace is
+not a calibration.
+
+Conventions, stated once: `theta` is a 1-D parameter vector in the
+order of `param_names`; `x` is an (n, d) array of measurement settings
+(one row per planned or performed measurement -- a field value, a
+frequency, a temperature, anything, in your units); the forward
+function returns the (n,) predicted readings in your measurement
+units. Nothing here assumes any physics: the statistics downstream are
+exact for any model that is differentiable in its parameters.
+"""
+from __future__ import annotations
+
+import dataclasses
+
+import numpy as np
+
+__all__ = ["Model"]
+
+
+def _check_ref(reference):
+    if not isinstance(reference, str) or len(reference.strip()) < 8:
+        raise ValueError(
+            "a real `reference` string is required: a model without a "
+            "traceable source (a paper, a manual, your own derivation "
+            "note) cannot anchor a calibration record")
+
+
+@dataclasses.dataclass(frozen=True)
+class Model:
+    """A named forward model.
+
+    name : short name of the model (appears in every report).
+    f : callable f(theta, x) -> (n,) predictions, with theta a 1-D
+        parameter vector and x an (n, d) settings array.
+    param_names : one name per parameter, in theta order.
+    reference : where the model comes from. Required, on purpose.
+    units : optional free-text note on the units of settings and
+        readings, carried into reports verbatim.
+    """
+
+    name: str
+    f: object
+    param_names: tuple
+    reference: str
+    units: str = ""
+
+    def __post_init__(self):
+        if not isinstance(self.name, str) or not self.name.strip():
+            raise ValueError("the model needs a non-empty name")
+        if not callable(self.f):
+            raise ValueError("f must be a callable f(theta, x) -> "
+                             "predictions")
+        names = tuple(str(n) for n in self.param_names)
+        if len(names) < 1 or len(set(names)) != len(names):
+            raise ValueError("param_names must be non-empty and "
+                             "unique")
+        object.__setattr__(self, "param_names", names)
+        _check_ref(self.reference)
+
+    @property
+    def n_params(self):
+        return len(self.param_names)
+
+    def predict(self, theta, x):
+        """Evaluate the forward model with shape checking."""
+        theta = np.asarray(theta, dtype=float).ravel()
+        if theta.size != self.n_params:
+            raise ValueError(f"theta must have {self.n_params} "
+                             "entries, one per parameter name")
+        x = _settings(x)
+        y = np.asarray(self.f(theta, x), dtype=float).ravel()
+        if y.size != x.shape[0]:
+            raise ValueError("the forward function must return one "
+                             "prediction per settings row")
+        if not np.all(np.isfinite(y)):
+            raise ValueError("the forward function returned non-"
+                             "finite predictions; check theta and x")
+        return y
+
+    def jacobian(self, theta, x, rel_step=1e-6):
+        """d(prediction)/d(parameter) by central differences, one
+        column per parameter."""
+        theta = np.asarray(theta, dtype=float).ravel()
+        x = _settings(x)
+        jac = np.empty((x.shape[0], self.n_params))
+        for j in range(self.n_params):
+            h = rel_step * max(abs(theta[j]), 1e-3)
+            tp, tm = theta.copy(), theta.copy()
+            tp[j] += h
+            tm[j] -= h
+            jac[:, j] = (self.predict(tp, x)
+                         - self.predict(tm, x)) / (2.0 * h)
+        return jac
+
+
+def _settings(x):
+    x = np.asarray(x, dtype=float)
+    if x.ndim == 1:
+        x = x[:, None]
+    if x.ndim != 2 or x.shape[0] < 1:
+        raise ValueError("settings must be an (n, d) array: one row "
+                         "per measurement")
+    if not np.all(np.isfinite(x)):
+        raise ValueError("settings must be finite")
+    return x

labplan/plan.py

@@ -0,0 +1,148 @@
+"""Plan the measurement before taking it.
+
+`information` answers, from the design alone, the question every
+instrument session should start with: can these planned measurements
+determine the parameters at all, and how small would the error bars
+come out? It is not an approximation of the fit -- it is the same
+(J^T W J) matrix the fit inverts, evaluated at your expected
+parameters, so on data the model describes the promise is kept
+exactly (the tests assert it).
+
+`design` picks the most informative subset of candidate measurements
+by the standard determinant criterion (D-optimal design; F.
+Pukelsheim, Optimal Design of Experiments, SIAM (2006)): each greedy
+pick most shrinks the joint parameter uncertainty. The greedy rule is
+transparent and each step can only add information (the rank-one
+determinant identity det(A + g g^T) = det(A)(1 + g^T A^-1 g)), but it
+is a good-practice heuristic, not a proof of the globally best
+subset.
+
+`repeats_for` is a closed form, not a search: repeating a design r
+times scales its covariance by exactly 1/r.
+"""
+from __future__ import annotations
+
+import numpy as np
+
+from .model import Model, _settings
+from .stats import SINGULAR_MSG, check_sigmas, invert_information
+
+__all__ = ["information", "design", "repeats_for"]
+
+
+def information(model: Model, theta, x, sigmas=None):
+    """Predicted error bars for a planned design.
+
+    theta : the working point (your expected parameters -- a previous
+        calibration, a datasheet, a cited set).
+    x : (n, d) planned measurement settings.
+    sigmas : expected 1-sigma reading error (scalar or (n,)). Without
+        it the answer is per unit measurement error, and real error
+        bars scale linearly with your sigma.
+
+    Returns dict(fisher, identifiable, condition_number, sigma, cov):
+    `sigma` maps each parameter to the error bar the weighted fit
+    would report, or is None when the design cannot tell the
+    parameters apart.
+    """
+    x = _settings(x)
+    sig = check_sigmas(sigmas, x.shape[0])
+    w = np.ones(x.shape[0]) if sig is None else 1.0 / sig
+    theta = np.asarray(theta, dtype=float).ravel()
+    jac = model.jacobian(theta, x) * w[:, None]
+    fisher = jac.T @ jac
+    identifiable, cond, cov, sigma = invert_information(
+        fisher, model.param_names)
+    if x.shape[0] < model.n_params:
+        identifiable, cov, sigma = False, None, None
+    return {"fisher": fisher, "identifiable": identifiable,
+            "condition_number": cond, "sigma": sigma, "cov": cov}
+
+
+def design(model: Model, theta, candidates, n_pick, sigmas=None):
+    """Pick the most informative subset of candidate measurements.
+
+    Greedy D-optimal selection over the candidate settings rows.
+    Returns dict(indices, fisher, condition_number, sigma) with the
+    chosen candidate indices in pick order. Refuses when even the
+    full candidate list cannot identify the parameters.
+    """
+    x = _settings(candidates)
+    n, p = x.shape[0], model.n_params
+    n_pick = int(n_pick)
+    if not p <= n_pick <= n:
+        raise ValueError(f"n_pick must be between {p} (the number of "
+                         f"parameters) and {n} (the number of "
+                         "candidates)")
+    sig = check_sigmas(sigmas, n)
+    full = information(model, theta, x, sig)
+    if not full["identifiable"]:
+        raise ValueError("even the full candidate list is not "
+                         "identifiable: " + SINGULAR_MSG)
+    w = np.ones(n) if sig is None else 1.0 / sig
+    theta = np.asarray(theta, dtype=float).ravel()
+    rows = model.jacobian(theta, x) * w[:, None]
+    # column-scaled (unit-free) greedy: identical scaling multiplies
+    # every candidate determinant by the same constant, so the picks
+    # are unchanged, while the tiny start-up regularizer stays
+    # meaningful in every direction
+    scale = np.sqrt(np.mean(rows * rows, axis=0))
+    scale[scale == 0.0] = 1.0
+    rs = rows / scale
+    eps = 1e-12 * float(np.max(np.sum(rs * rs, axis=1)))
+    fs = eps * np.eye(p)
+    chosen = []
+    for _ in range(n_pick):
+        best_j, best_det = -1, -np.inf
+        for j in range(n):
+            if j in chosen:
+                continue
+            det = float(np.linalg.slogdet(
+                fs + np.outer(rs[j], rs[j]))[1])
+            if det > best_det:
+                best_j, best_det = j, det
+        fs = fs + np.outer(rs[best_j], rs[best_j])
+        chosen.append(best_j)
+    fisher = (fs - eps * np.eye(p)) * np.outer(scale, scale)
+    _, cond, _, sigma = invert_information(fisher, model.param_names)
+    return {"indices": list(chosen), "fisher": fisher,
+            "condition_number": cond, "sigma": sigma}
+
+
+def repeats_for(target_sigma, plan):
+    """How many repeats of a planned design meet a target error bar?
+
+    Exact closed form: r identical repeats of a design multiply its
+    information by r, so every error bar shrinks by exactly
+    1/sqrt(r). `target_sigma` is the largest acceptable error bar of
+    any parameter, in that parameter's own units -- pass a dict
+    {name: target} to set per-parameter targets instead.
+
+    Returns (r, predicted) with `predicted` the per-parameter error
+    bars at r repeats. Refuses a non-identifiable plan: no number of
+    repeats can identify what one copy cannot.
+    """
+    if not plan.get("identifiable") or plan.get("sigma") is None:
+        raise ValueError("the plan is not identifiable; no number of "
+                         "repeats can identify what one copy of the "
+                         "design cannot -- change the design")
+    sigma = plan["sigma"]
+    if isinstance(target_sigma, dict):
+        ratios = []
+        for name, t in target_sigma.items():
+            if name not in sigma:
+                raise ValueError(f"unknown parameter {name!r} in "
+                                 "target_sigma")
+            t = float(t)
+            if not (np.isfinite(t) and t > 0.0):
+                raise ValueError("targets must be positive")
+            ratios.append(sigma[name] / t)
+        worst = max(ratios)
+    else:
+        t = float(target_sigma)
+        if not (np.isfinite(t) and t > 0.0):
+            raise ValueError("target_sigma must be positive")
+        worst = max(s / t for s in sigma.values())
+    r = max(1, int(np.ceil(worst ** 2)))
+    predicted = {k: v / np.sqrt(r) for k, v in sigma.items()}
+    return r, predicted

labplan/records.py

@@ -0,0 +1,90 @@
+"""A plain, checked file contract for measurement records.
+
+One CSV per record: setting columns x1..xd, the measured reading y,
+and optionally its 1-sigma error. Values are written with `repr`, so
+the round trip is bit-exact; the header and every value are checked
+on load, and a malformed file is refused, not guessed at -- the same
+contract style every package in this organization uses.
+"""
+from __future__ import annotations
+
+import csv
+
+import numpy as np
+
+from .model import _settings
+from .stats import check_sigmas
+
+__all__ = ["save_measurements_csv", "load_measurements_csv"]
+
+
+def _header(d, has_sigma):
+    cols = tuple(f"x{i + 1}" for i in range(d)) + ("y",)
+    return cols + ("sigma",) if has_sigma else cols
+
+
+def save_measurements_csv(path, x, y, sigmas=None):
+    """Write a measurement record; the exact inverse of
+    `load_measurements_csv`."""
+    x = _settings(x)
+    y = np.asarray(y, dtype=float).ravel()
+    if y.size != x.shape[0]:
+        raise ValueError("need one reading per settings row")
+    if not np.all(np.isfinite(y)):
+        raise ValueError("readings must be finite")
+    sig = check_sigmas(sigmas, x.shape[0])
+    header = _header(x.shape[1], sig is not None)
+    with open(path, "w", newline="") as fh:
+        wr = csv.writer(fh)
+        wr.writerow(header)
+        for i in range(x.shape[0]):
+            row = [repr(float(v)) for v in x[i]] + [repr(float(y[i]))]
+            if sig is not None:
+                row.append(repr(float(sig[i])))
+            wr.writerow(row)
+
+
+def load_measurements_csv(path):
+    """Read a measurement record written by `save_measurements_csv`.
+
+    Returns (x, y, sigmas) with sigmas None when the file has no
+    sigma column.
+    """
+    with open(path, newline="") as fh:
+        rows = list(csv.reader(fh))
+    if not rows:
+        raise ValueError("empty measurement file")
+    header = tuple(rows[0])
+    if len(header) < 2 or header[-1] not in ("y", "sigma"):
+        raise ValueError(f"unrecognized header {header}")
+    has_sigma = header[-1] == "sigma"
+    d = len(header) - (2 if has_sigma else 1)
+    if d < 1 or header != _header(d, has_sigma):
+        raise ValueError(f"measurement header must be "
+                         f"{_header(max(d, 1), has_sigma)}; got "
+                         f"{header}")
+    body = rows[1:]
+    if not body:
+        raise ValueError("measurement file has no data rows")
+    xs, ys, ss = [], [], []
+    for row in body:
+        if len(row) != len(header):
+            raise ValueError(f"row {row!r} does not match the header")
+        try:
+            vals = [float(v) for v in row]
+        except ValueError as exc:
+            raise ValueError(f"non-numeric value in row {row!r}") \
+                from exc
+        xs.append(vals[:d])
+        ys.append(vals[d])
+        if has_sigma:
+            ss.append(vals[d + 1])
+    x = np.array(xs)
+    y = np.array(ys)
+    sig = np.array(ss) if has_sigma else None
+    _settings(x)
+    if not np.all(np.isfinite(y)):
+        raise ValueError("readings must be finite")
+    if sig is not None:
+        check_sigmas(sig, y.size)
+    return x, y, sig

labplan/report.py

@@ -0,0 +1,99 @@
+"""The audit trail: what was fitted, to which data, by what, when.
+
+A calibration that cannot be traced is an opinion. `audit_record`
+turns a `FitResult` into a plain dictionary that says exactly what
+happened: the model's name and source, every fitted value with its
+error bar, the goodness of fit, the identifiability diagnostics, a
+sha256 digest of the exact data arrays the result answers for, the
+package versions that produced it, and a UTC timestamp. It is
+JSON-serializable as-is, so it can live next to the data forever;
+`report_text` renders the same record for a human or a logbook.
+
+Nothing in the record is re-derived at reading time: it is a
+statement of record, checked round-trip in the tests
+(json.dumps -> json.loads reproduces it exactly).
+"""
+from __future__ import annotations
+
+import datetime
+import json
+
+import numpy as np
+
+from .fit import FitResult
+
+__all__ = ["audit_record", "report_text"]
+
+
+def audit_record(result: FitResult, operator="", note=""):
+    """A JSON-serializable statement of record for one fit.
+
+    operator : who ran the calibration (free text, e.g. a name or an
+        instrument id); recorded verbatim.
+    note : anything else worth remembering, recorded verbatim.
+    """
+    from . import __version__
+    if not isinstance(operator, str) or not isinstance(note, str):
+        raise ValueError("operator and note must be strings")
+    rec = {
+        "record": "labplan.calibration",
+        "model": result.model_name,
+        "model_reference": result.reference,
+        "parameters": {
+            name: {"value": result.values[name],
+                   "sigma": result.sigma[name]}
+            for name in result.values
+        },
+        "covariance": [[float(v) for v in row] for row in result.cov],
+        "chi2": result.chi2,
+        "chi2_dof": result.chi2_dof,
+        "n_points": result.n_points,
+        "condition_number": result.condition_number,
+        "data_sha256": result.data_digest,
+        "software": {"labplan": __version__,
+                     "numpy": np.__version__},
+        "timestamp_utc": datetime.datetime.now(
+            datetime.timezone.utc).isoformat(timespec="seconds"),
+        "operator": operator,
+        "note": note,
+    }
+    # the record must survive serialization exactly
+    json.dumps(rec)
+    return rec
+
+
+def report_text(record):
+    """Render an audit record for a human or a logbook."""
+    if not isinstance(record, dict) \
+            or record.get("record") != "labplan.calibration":
+        raise ValueError("expected an audit_record dictionary")
+    lines = [
+        "labplan calibration record",
+        f"  model:      {record['model']}",
+        f"  source:     {record['model_reference']}",
+        f"  when (UTC): {record['timestamp_utc']}",
+    ]
+    if record.get("operator"):
+        lines.append(f"  operator:   {record['operator']}")
+    lines.append("  parameters:")
+    for name, pv in record["parameters"].items():
+        lines.append(f"    {name} = {pv['value']:.9g} "
+                     f"+/- {pv['sigma']:.3g}")
+    if record["chi2"] is not None:
+        lines.append(f"  chi2 / dof: {record['chi2']:.4g} / "
+                     f"{record['chi2_dof']} (near 1 per dof means "
+                     "model and stated errors agree)")
+    else:
+        lines.append("  chi2:       not available (no measurement "
+                     "errors were supplied; error bars are scaled "
+                     "from the residual scatter)")
+    lines.append(f"  points:     {record['n_points']}")
+    lines.append(f"  condition:  {record['condition_number']:.3g} "
+                 "(unit-free; large means barely identifiable)")
+    lines.append(f"  data sha256: {record['data_sha256']}")
+    sw = record["software"]
+    lines.append(f"  software:   labplan {sw['labplan']}, "
+                 f"numpy {sw['numpy']}")
+    if record.get("note"):
+        lines.append(f"  note:       {record['note']}")
+    return "\n".join(lines)

labplan/stats.py

@@ -0,0 +1,61 @@
+"""The shared exact statistics: weighted least squares, Fisher
+information, scale-invariant identifiability.
+
+Everything here is the standard machinery of independent Gaussian
+measurement errors, stated plainly (see e.g. D. R. Cox and D. V.
+Hinkley, Theoretical Statistics (1974), or any statistics text under
+"Cramer-Rao bound"): the information matrix of a design is J^T W J
+with J the model sensitivities and W = diag(1/sigma^2), the
+covariance of the weighted-least-squares estimate is its inverse, and
+both are the SAME matrix -- which is why labplan can promise, before
+any data exist, exactly the error bars the fit will report.
+
+Identifiability is judged on the correlation-scaled matrix
+D^-1 F D^-1 with D = sqrt(diag F): parameters carry arbitrary units,
+so the raw condition number partly measures the units, while exact
+functional degeneracies of the model survive the scaling and unit
+mismatches do not.
+"""
+from __future__ import annotations
+
+import numpy as np
+
+__all__ = ["invert_information"]
+
+COND_MAX = 1e10
+
+SINGULAR_MSG = ("these measurements cannot tell the parameters apart "
+                "(singular or near-singular information matrix): some "
+                "combination of parameters changes nothing the design "
+                "can see. Measure settings that respond differently "
+                "to each parameter -- `information` shows which "
+                "designs work before you spend the instrument time")
+
+
+def invert_information(fisher, names):
+    """Scale-invariant inversion of an information matrix.
+
+    Returns (identifiable, condition_number, cov, sigma_dict); cov and
+    sigma are None when the design is not identifiable.
+    """
+    fisher = np.asarray(fisher, dtype=float)
+    d = np.sqrt(np.diag(fisher))
+    if np.any(d <= 0.0) or not np.all(np.isfinite(d)):
+        return False, np.inf, None, None
+    fs = fisher / np.outer(d, d)
+    sv = np.linalg.svd(fs, compute_uv=False)
+    cond = float(sv[0] / sv[-1]) if sv[-1] > 0 else np.inf
+    if not (np.isfinite(cond) and cond <= COND_MAX):
+        return False, cond, None, None
+    cov = np.linalg.inv(fs) / np.outer(d, d)
+    err = np.sqrt(np.diag(cov))
+    return True, cond, cov, {n: float(s) for n, s in zip(names, err)}
+
+
+def check_sigmas(sigmas, n):
+    if sigmas is None:
+        return None
+    sig = np.broadcast_to(np.asarray(sigmas, dtype=float), (n,)).copy()
+    if np.any(sig <= 0.0) or not np.all(np.isfinite(sig)):
+        raise ValueError("sigmas must be finite and positive")
+    return sig

labplan-0.1.0.dist-info/METADATA

@@ -0,0 +1,208 @@
+Metadata-Version: 2.4
+Name: labplan
+Version: 0.1.0
+Summary: Plan the measurement, fit the model, keep the record - exact measurement planning, calibration and audit trails for any lab model
+Author: Tanvir Mahmud Mahim
+License: Apache-2.0
+Project-URL: Homepage, https://github.com/TaN-MM-Org/labplan
+Project-URL: Issues, https://github.com/TaN-MM-Org/labplan/issues
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Physics
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: numpy>=1.22
+Provides-Extra: test
+Requires-Dist: pytest>=7; extra == "test"
+Dynamic: license-file
+
+# labplan
+
+[![tests](https://github.com/TaN-MM-Org/labplan/actions/workflows/ci.yml/badge.svg)](https://github.com/TaN-MM-Org/labplan/actions)
+[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)
+
+Every measurement campaign asks the same four questions. Can the
+measurements I am about to take determine the numbers I care about --
+and how well? Which settings are worth the instrument time? Once the
+data exist, what are the numbers, with error bars that mean it? And a
+year from now, can anyone trace exactly what was fitted, to which
+data, by what? `labplan` answers all four for ANY instrument or
+experiment you can describe with a Python function -- and refuses,
+with an explanation, whenever the honest answer is "this design
+cannot tell".
+
+Nine research packages in this organization -- covering colour-centre
+spins, squeezed light, spin-squeezed clocks, Raman maps,
+single-photon emitters, superconducting detectors, band structure,
+semiconductor heterostructures and photonic fabrication -- each grew
+the same planning-and-calibration loop for its own physics. `labplan`
+is that loop extracted, generalized, and hardened into a single
+dependency-light package (NumPy only, Python 3.9-3.14): the tenth
+package is the pattern itself.
+
+## Install
+
+```
+pip install labplan        # NumPy only
+```
+
+## The loop in one example
+
+```python
+import numpy as np
+from labplan import (Model, information, design, fit,
+                     repeats_for, audit_record, report_text)
+
+# 1. Your model: anything that predicts a reading from parameters
+#    and settings. Here, a sensor line y = gain * x + offset.
+m = Model("sensor line",
+          lambda th, x: th[0] * x[:, 0] + th[1],
+          param_names=("gain", "offset"),
+          reference="sensor manual rev. 3, eq. (2)")
+
+# 2. Before measuring: would 12 planned settings determine the
+#    parameters, and how well, at 0.05 units of reading noise?
+x_planned = np.linspace(0.5, 5.0, 12)[:, None]
+plan = information(m, theta=[2.0, 0.0], x=x_planned, sigmas=0.05)
+print(plan["identifiable"], plan["sigma"])
+
+# ... or let labplan pick the best 5 of the settings you can reach,
+#     and price a target error bar in repeats (a closed form):
+pick = design(m, [2.0, 0.0], x_planned, n_pick=5, sigmas=0.05)
+r, predicted = repeats_for({"gain": 0.01}, plan)
+
+# 3. After measuring: fit, with the SAME matrix the plan promised.
+res = fit(m, x_planned, y_measured, theta0=[1.0, 0.0], sigmas=0.05)
+print(res.values, res.sigma, res.chi2)
+
+# 4. Keep the record: what, to which data (sha256), by what, when.
+rec = audit_record(res, operator="T. Mahim", note="bench 2, warm-up ok")
+print(report_text(rec))
+```
+
+The central promise: the planner's error bars and the fit's error
+bars are the same matrix, so what is promised before the measurement
+is what is reported after -- exactly, whenever the model describes
+the data. The tests assert that equality to machine precision.
+
+## When the model might be wrong
+
+Model-based error bars assume the model is right. For the day it is
+not, `conformal_quantile` supplies distribution-free prediction
+intervals from held-out calibration data, with an exact finite-sample
+guarantee that holds whatever the model and whatever the noise (split
+conformal prediction; Vovk, Gammerman and Shafer, Algorithmic
+Learning in a Random World, Springer (2005); Lei et al., J. Am.
+Stat. Assoc. 113, 1094 (2018); Angelopoulos and Bates,
+arXiv:2107.07511):
+
+```python
+from labplan import conformal_quantile, conformal_interval
+
+q = conformal_quantile(abs_errors_heldout, alpha=0.1)   # 90% level
+lo, hi = conformal_interval(new_predictions, q)
+```
+
+Its honest limits are stated in the docstring rather than hidden:
+the guarantee is marginal, and it needs the calibration data to be
+exchangeable with the new measurement -- last month's instrument
+state does not certify next month's drift.
+
+## What is inside
+
+- **`Model`**: your forward function with named parameters and a
+  mandatory `reference` -- provenance travels with every prediction,
+  the same rule every package in this organization applies.
+- **`information` / `design` / `repeats_for`**: predicted error bars
+  from the design alone (the Fisher information of independent
+  Gaussian measurements; any statistics text, under "Cramer-Rao
+  bound"); greedy D-optimal selection of the most informative
+  settings (Pukelsheim, Optimal Design of Experiments, SIAM (2006));
+  and the exact 1/sqrt(repeats) law, inverted in closed form.
+- **`fit`**: Levenberg-Marquardt weighted least squares in pure
+  NumPy, with exact known-noise covariance and a chi-squared check
+  when measurement errors are supplied, and residual-scaled error
+  bars (stated as such) when they are not.
+- **`conformal_quantile` / `conformal_interval` / `coverage_exact`**:
+  distribution-free intervals with the exact finite-sample coverage
+  formula exposed for checking.
+- **`audit_record` / `report_text`**: a JSON-serializable statement
+  of record -- model, source, values, error bars, goodness of fit,
+  identifiability, the sha256 of the exact data arrays, software
+  versions, UTC timestamp, operator -- and its human-readable
+  rendering.
+- **`save_measurements_csv` / `load_measurements_csv`**: a plain,
+  checked file contract whose round trip is bit-exact.
+
+## Refusals, not guesses
+
+A design that cannot tell the parameters apart is refused with an
+explanation, in the planner, the fit and the design tool alike --
+judged on a unit-free (correlation-scaled) information matrix, so
+mixed units can never fake or hide a degeneracy. Too few points, a
+non-converging fit, an uncertifiable conformal level, a malformed
+data file: each refuses with the reason and, where one exists, the
+remedy.
+
+## How it is checked
+
+14 tests (Python 3.9-3.14, run in CI on every push), every
+statistical claim pinned to a closed form, an exact identity, or
+seeded simulation against an exact formula -- never a stored number.
+Highlights: on a linear model the fit covariance equals the textbook
+closed form sigma^2 (X^T X)^-1 exactly, and the planner promises the
+same matrix; 400 seeded Monte-Carlo experiments match the reported
+error bars; an exactly degenerate model is refused via an exact rank
+argument; the greedy design obeys the rank-one determinant identity,
+reproduces its own rule, never loses to a random subset, and -- for
+the two-point line design -- matches the classical optimum found by
+exhaustion; the repeat law is asserted by tiling the design; the
+conformal quantile is the exact rank formula and seeded simulation
+matches the exact closed-form coverage inside the published
+two-sided guarantee; the audit record survives JSON round trip
+exactly and its digest pins the exact data; file round trips are
+bit-exact.
+
+## Honest limits
+
+Deliberate scope, designed out with reasons: the Gaussian
+error-bar machinery is exact for independent Gaussian measurement
+errors and first-order-accurate otherwise (the conformal tools are
+the assumption-free complement, and their own limits are stated);
+the greedy design is a transparent heuristic, not a proof of global
+optimality; no physics ships in this package at all -- your model
+and its `reference` carry the physics, and the nine physics packages
+of this organization remain the place where specific instruments'
+models live, each already wired into this same loop.
+
+## Support and governance
+
+Written and maintained by Tanvir Mahmud Mahim (Department of
+Electrical and Electronic Engineering, BRAC University), who reviews
+every change and takes the final decision on scope and releases.
+Design questions are discussed in the open in issues and pull
+requests, and the standing rule of
+[CONTRIBUTING.md](CONTRIBUTING.md) binds the maintainer exactly as it
+binds contributors: a change that touches the statistics arrives with
+a test, and a claim arrives with its source.
+
+Support runs through the
+[issue tracker](https://github.com/TaN-MM-Org/labplan/issues). Usage
+questions are welcome alongside bug reports; a docstring that left a
+unit or a convention unclear is treated as a documentation bug, not
+user error. While the version is below 1.0 the API may still move
+between minor versions; such changes are called out in the release
+notes.
+
+## License
+
+Apache-2.0.

labplan-0.1.0.dist-info/RECORD

@@ -0,0 +1,13 @@
+labplan/__init__.py,sha256=vwzYsh_vjFZcdGgE5gb4HI9OXLcvDrJy-8ACuJD4YOs,2464
+labplan/conformal.py,sha256=V57nOiHXhlcuhXrB0iRjKle3u3fMyZJIZ0BnHo1giIg,3668
+labplan/fit.py,sha256=p6BK732Scqnz34RGozsX2MG2iL76HnLHkOPukfzzdas,5840
+labplan/model.py,sha256=SPM3p9hcQ4rxtvPEgYzMkVpNYeyBOyjIHnzkjcKAigY,4497
+labplan/plan.py,sha256=t44NyKXM0aDnH5hZZm5TQ2VujfdPcrq1PaazG6GArHI,6353
+labplan/records.py,sha256=mkP1jFMPp0DYXrE68LQy9V9f6GcyAO9Y5xkjX1Gr1Bo,3115
+labplan/report.py,sha256=RueDQxiiLwqytxICymCrK4s1VFWaytnzfeoMqlqcm_0,3949
+labplan/stats.py,sha256=Gu_6nHR-daFqDmRGnKb3_RVCV5-1UBL26xWkOE4OoKA,2517
+labplan-0.1.0.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
+labplan-0.1.0.dist-info/METADATA,sha256=ls9uYJUP15GXDidjS3tUmM9iM9U-tmh-pV_Dp6Moe5U,9372
+labplan-0.1.0.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
+labplan-0.1.0.dist-info/top_level.txt,sha256=JhJuOv_Gavmvt5oPPnnT9SVWQNam6jJQRf2GVqbd9XQ,8
+labplan-0.1.0.dist-info/RECORD,,

labplan-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (84.0.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

labplan-0.1.0.dist-info/licenses/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

labplan-0.1.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+labplan