working fullfied (kind of), better units

.gitignore

@@ -4,6 +4,7 @@
 
 plots*
 /make_*.py
+/debug*.py
 Archive
 *.mp4
 *.png

src/scgenerator/evaluator.py

@@ -1,7 +1,6 @@
 import itertools
 from collections import defaultdict
 from dataclasses import dataclass
-from operator import itemgetter
 from typing import Any, Callable, Optional, Union
 
 import numpy as np
@@ -259,8 +258,8 @@ class Evaluator:
                     value = default
                     self.logger.info(prefix + f"using default value of {value} for {target}")
                     self.set_value(target, value, 0)
-
-            assert target == self.__curent_lookup.pop()
+            last_target = self.__curent_lookup.pop()
+            assert target == last_target
             self.__failed_rules[target] = []
 
             if value is None and error is not None:

src/scgenerator/math.py

@@ -354,3 +354,48 @@ def _polynom_extrapolation_in_place(y: np.ndarray, left_ind: int, right_ind: int
     y[:left_ind] = r_left * (y[left_ind] - y[left_ind + 1]) + y[left_ind]
     y[right_ind:] = r_right * (y[right_ind] - y[right_ind - 1]) + y[right_ind]
     return y
+
+
+def envelope_ind(
+    signal: np.ndarray, dmin: int = 1, dmax: int = 1, split: bool = False
+) -> tuple[np.ndarray, np.ndarray]:
+    """
+    returns the indices of the top/bottom envelope of a signal
+
+    Parameters
+    ----------
+    signal : np.ndarray, shape (n,)
+        signal array (must be sorted)
+    dmin, dmax : int, optional
+        size of chunks for lower/upper envelope
+    split: bool, optional
+        split the signal in half along its mean, might help to generate the envelope in some cases
+        this has the effect of forcing the envlopes to be on either side of the dc signal
+        by default False
+
+    Returns
+    -------
+    np.ndarray, shape (m,), m < n
+        lower envelope indices
+    np.ndarray, shape (l,), l < n
+        upper envelope indices
+    """
+
+    local_min = (np.diff(np.sign(np.diff(signal))) > 0).nonzero()[0] + 1
+    local_max = (np.diff(np.sign(np.diff(signal))) < 0).nonzero()[0] + 1
+
+    if split:
+        dc_value = np.mean(signal)
+        local_min = local_min[signal[local_min] < dc_value]
+        local_max = local_max[signal[local_max] > dc_value]
+
+    if dmin > 1:
+        local_min = local_min[
+            [i + np.argmin(signal[local_min[i : i + dmin]]) for i in range(0, len(local_min), dmin)]
+        ]
+    if dmax > 1:
+        local_max = local_max[
+            [i + np.argmax(signal[local_max[i : i + dmax]]) for i in range(0, len(local_max), dmax)]
+        ]
+
+    return local_min, local_max

src/scgenerator/operators.py

@@ -630,7 +630,7 @@ class FullFieldSPM(AbstractSPM):
         self.factor = self.fraction * chi3 * units.epsilon0
 
     def __call__(self, state: CurrentState) -> np.ndarray:
-        return self.fraction * state.field2 * state.field
+        return self.factor * state.field2 * state.field
 
 
 ##################################################

src/scgenerator/parameter.py

@@ -20,7 +20,6 @@ from .evaluator import Evaluator
 from .logger import get_logger
 from .operators import (
     AbstractConservedQuantity,
-    EnvelopeLinearOperator,
     LinearOperator,
     NonLinearOperator,
 )
@@ -423,12 +422,13 @@ class Parameters:
     def __repr_list__(self) -> Iterator[str]:
         yield from (f"{k}={v}" for k, v in self.dump_dict().items())
 
-    def dump_dict(self, compute=True) -> dict[str, Any]:
+    def dump_dict(self, compute=True, add_metadata=True) -> dict[str, Any]:
         if compute:
             self.compute_in_place()
         param = Parameters.strip_params_dict(self._param_dico)
-        param["datetime"] = datetime_module.datetime.now()
-        param["version"] = __version__
+        if add_metadata:
+            param["datetime"] = datetime_module.datetime.now()
+            param["version"] = __version__
         return param
 
     def compute_in_place(self, *to_compute: str):
@@ -440,7 +440,8 @@ class Parameters:
     def compute(self, key: str) -> Any:
         return self._evaluator.compute(key)
 
-    def pformat(self) -> str:
+    def pretty_str(self) -> str:
+        """return a pretty formatted string describing the parameters"""
         return "\n".join(
             f"{k} = {VariationDescriptor.format_value(k, v)}" for k, v in self.dump_dict().items()
         )
@@ -597,7 +598,7 @@ class Configuration:
             self.variationer.append(config.variable)
             self.fiber_paths.append(
                 utils.ensure_folder(
-                    self.final_path / fiber_folder(i, self.name, config.fixed["name"]),
+                    self.final_path / fiber_folder(i, self.name, Path(config.fixed["name"]).name),
                     mkdir=False,
                     prevent_overwrite=not self.overwrite,
                 )

src/scgenerator/physics/__init__.py

@@ -58,7 +58,7 @@ def material_dispersion(
         disp = np.zeros(len(w))
         ind = w > 0
         disp[ind] = material_dispersion(
-            units.To.m(w[ind]), material, pressure, temperature, ideal, False
+            units.m.inv(w[ind]), material, pressure, temperature, ideal, False
         )
         return disp
     else:
@@ -102,7 +102,7 @@ def find_optimal_depth(
     w = w_c + w0
     disp = np.zeros(len(w))
     ind = w > (w0 / 10)
-    disp[ind] = material_dispersion(units.To.m(w[ind]), material)
+    disp[ind] = material_dispersion(units.m.inv(w[ind]), material)
 
     propagate = lambda z: spectrum * np.exp(-0.5j * disp * w_c ** 2 * z)
     integrate = lambda z: math.abs2(np.fft.ifft(propagate(z)))

src/scgenerator/physics/units.py

@@ -2,7 +2,11 @@
 # For example, nm(X) means "I give the number X in nm, figure out the ang. freq."
 # to be used especially when giving plotting ranges : (400, 1400, nm), (-4, 8, ps), ...
 
+from __future__ import annotations
+from collections import defaultdict
+
 from typing import Callable, TypeVar, Union
+from functools import wraps
 from operator import itemgetter
 import numpy as np
 from numpy import pi
@@ -25,17 +29,38 @@ provided you decorate it with @unit and provide at least a type and a label
 types are "WL", "FREQ", "AFREQ", "TIME", "PRESSURE", "TEMPERATURE", "OTHER"
 """
 _T = TypeVar("_T")
+_UT = Callable[[_T], _T]
 
 
-class From:
-    pass
+def chain(c1: _UT, c2: _UT) -> _UT:
+    def chained_function(x: _T) -> _T:
+        return c1(c2(x))
+
+    return chained_function
+
+
+class UnitMap(dict):
+    def __setitem__(self, new_name, new_func):
+        super().__setitem__(new_name, new_func)
+        already_here = [(name, func) for name, func in self.items() if isinstance(name, str)]
+        for name, func in already_here:
+            super().__setitem__((name, new_name), chain(self[name].inv, new_func))
+            super().__setitem__((new_name, name), chain(self[new_name].inv, func))
+
+
+units_map: dict[str, dict[Union[str, tuple[str, str]], Unit]] = defaultdict(UnitMap)
 
 
 class To:
-    pass
+    def __init__(self, name: str, tpe: str):
+        self.name = name
+        self.type = tpe
 
-
-units_map = dict()
+    def __getattr__(self, key: str):
+        try:
+            return units_map[self.type][self.name, key]
+        except KeyError:
+            raise KeyError(f"no registered unit named {key!r} of type {self.type!r}") from None
 
 
 def W_to_Vm(n0: float, A_eff: float) -> float:
@@ -54,22 +79,36 @@ def W_to_Vm(n0: float, A_eff: float) -> float:
     return 1.0 / np.sqrt(A_eff * 0.5 * epsilon0 * c * n0)
 
 
+class Unit:
+    __func: _UT
+    to: To
+    inv: _UT
+    name: str
+    label: str
+
+    def __init__(self, func: _UT, inv: _UT, name: str, label: str, tpe: str):
+        self.__func = func
+        self.inv = inv
+        self.to = To(name, tpe)
+        self.name = name
+        self.label = label
+        self.type = tpe
+        self.__name__ = name
+
+    def __call__(self, x: _T) -> _T:
+        """call the original unit function"""
+        return self.__func(x)
+
+
 def unit(tpe: str, label: str, inv: Callable = None):
-    def unit_maker(func):
+    def unit_maker(func) -> Unit:
         nonlocal inv
         name = func.__name__
         if inv is None:
             inv = func
-        setattr(From, name, func.__call__)
-        setattr(To, name, inv.__call__)
-        func.type = tpe
-        func.label = label
-        func.name = name
-        func.inv = inv
-        if name in units_map:
-            raise NameError(f"Two unit functions with the same name {name!r} were defined")
-        units_map[name] = func
-        return func
+        unit_obj = wraps(func)(Unit(func, inv, name, label, tpe))
+        units_map[tpe][name] = unit_obj
+        return unit_obj
 
     return unit_maker
 
@@ -89,6 +128,11 @@ def um(l: _T) -> _T:
     return 2 * pi * c / (l * 1e-6)
 
 
+@unit("FREQ", r"Frequency $f$ (Hz)", lambda w: w / (2 * pi))
+def Hz(f: _T) -> _T:
+    return 2 * pi * f
+
+
 @unit("FREQ", r"Frequency $f$ (THz)", lambda w: w / (1e12 * 2 * pi))
 def THz(f: _T) -> _T:
     return 1e12 * 2 * pi * f
@@ -306,7 +350,7 @@ class PlotRange(tuple):
     def __iter__(self):
         yield self.left
         yield self.right
-        yield self.unit.__name__
+        yield self.unit.name
 
     def __repr__(self):
         return "PlotRange" + super().__repr__()

src/scgenerator/plotting.py

@@ -419,6 +419,7 @@ def transform_2D_propagation(
     params: Parameters,
     log: Union[int, float, bool, str] = "1D",
     skip: int = 1,
+    y_axis=None,
 ) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
     """transforms raws values into plottable values
 
@@ -455,8 +456,8 @@ def transform_2D_propagation(
     is_complex, x_axis, plt_range = prep_plot_axis(values, plt_range, params)
     if is_complex:
         values = abs2(values)
-
-    y_axis = params.z_targets
+    if y_axis is None:
+        y_axis = params.z_targets
 
     x_axis, values = uniform_axis(x_axis, values, plt_range)
     y_axis, values.T[:] = uniform_axis(y_axis, values.T, None)
@@ -1102,7 +1103,7 @@ def partial_plot(root: os.PathLike):
     )
 
     t, z, values = transform_2D_propagation(
-        np.fft.ifft(raw_values),
+        params.ifft(raw_values),
         PlotRange(-10, 10, "ps"),
         params,
         log=False,

src/scgenerator/scripts/__init__.py

@@ -127,7 +127,7 @@ def plot_init(
         lbl = plot_1_dispersion(lim_disp, tl, tr, style, lbl, params, loss_ax)
         lbl = plot_1_init_spec_field(lim_field, lim_spec, bl, br, style, lbl, params)
         all_labels.append(lbl)
-        print(params.pformat())
+        print(params.pretty_str())
     finish_plot(fig, tr, all_labels, params)
 
 
@@ -142,8 +142,8 @@ def plot_1_init_spec_field(
 ):
     field = math.abs2(params.field_0)
     spec = math.abs2(params.spec_0)
-    t = units.To.fs(params.t)
-    wl = units.To.nm(params.w)
+    t = units.fs.inv(params.t)
+    wl = units.nm.inv(params.w)
 
     lbl += f" max at {wl[spec.argmax()]:.1f} nm"
 
@@ -219,7 +219,7 @@ def plot_1_dispersion(
     right.set_ylabel(units.D_ps_nm_km.label)
 
     # plot beta2
-    left.plot(units.To.nm(params.w[m]), units.beta2_fs_cm.inv(beta_arr[m]), label=" ", **style)
+    left.plot(units.nm.inv(params.w[m]), units.beta2_fs_cm.inv(beta_arr[m]), label=" ", **style)
     left.set_ylabel(units.beta2_fs_cm.label)
 
     left.set_xlabel(units.nm.label)

src/scgenerator/utils.py

@@ -311,8 +311,18 @@ def ensure_folder(path: Path, prevent_overwrite: bool = True, mkdir=True) -> Pat
         path = path.parent / (folder_name + f"_{i}")
 
 
-def branch_id(branch: tuple[Path, ...]) -> str:
-    return branch[-1].name.split()[1]
+def branch_id(branch: Path) -> tuple[int, int]:
+    sim_match = branch.parent.name.split()[0]
+    if sim_match.isdigit():
+        s_int = int(sim_match)
+    else:
+        s_int = 0
+    branch_match = re.search(r"(?<=b_)[0-9]+", branch.name)
+    if branch_match is None:
+        b_int = 0
+    else:
+        b_int = int(branch_match[0])
+    return s_int, b_int
 
 
 def find_last_spectrum_num(data_dir: Path):
@@ -519,5 +529,5 @@ def simulations_list(path: os.PathLike) -> list[Path]:
     for pwd, _, files in os.walk(path):
         if PARAM_FN in files:
             paths.append(Path(pwd))
-    paths.sort(key=lambda el: el.parent.name)
+    paths.sort(key=branch_id)
     return [p for p in paths if p.parent.name == paths[-1].parent.name]

src/scgenerator/variationer.py

@@ -149,6 +149,8 @@ class VariationDescriptor(BaseModel):
             p = Path(value)
             if p.exists():
                 return p.stem
+        elif callable(value):
+            return getattr(value, "__name__", repr(value))
         return str(value)
 
     class Config:

testing/configs/Chang2011Fig2.toml

@@ -8,12 +8,12 @@ width = 30e-15
 core_radius = 10e-6
 model = "marcatili"
 gas_name = "argon"
-pressure = 1e5
+pressure = 3.2e5
 
 length = 0.1
-interpolation_range = [100e-9, 3000e-9]
+interpolation_range = [120e-9, 3000e-9]
 full_field = true
-dt = 0.1e-15
+dt = 0.05e-15
 t_num = 32768
 z_num = 128
-step_size = 2e-6
+step_size = 10e-6

testing/test_full_field.py

@@ -1,36 +1,61 @@
 import warnings
-
-import matplotlib.pyplot as plt
 import numpy as np
+import rediscache
 import scgenerator as sc
 from customfunc.app import PlotApp
-from scgenerator.physics.simulate import RK4IP
-from customfunc import pprint
+from scipy.interpolate import interp1d
+from tqdm import tqdm
 
-warnings.filterwarnings("error")
+# warnings.filterwarnings("error")
+
+
+@rediscache.rcache
+def get_specs(params: dict):
+    p = sc.Parameters(**params)
+    sim = sc.RK4IP(p)
+    return [s[-1] for s in tqdm(sim.irun(), total=p.z_num)], p.dump_dict()
 
 
 def main():
+
     params = sc.Parameters.load("testing/configs/Chang2011Fig2.toml")
-    x = params.l * 1e9
-    o = np.argsort(x)
-    x = x[o]
+    specs, params = get_specs(params.dump_dict(add_metadata=False))
+    params = sc.Parameters(**params)
+    rs = sc.PlotRange(100, 1500, "nm")
+    rt = sc.PlotRange(-500, 500, "fs")
+    x, o, ext = rs.sort_axis(params.w)
+    vmin = -50
+    with PlotApp(i=(int, 0, params.z_num - 1)) as app:
+        spec_ax = app[0]
+        spec_ax.set_xlabel(rs.unit.label)
+        field_ax = app[1]
+        field_ax.set_xlabel(rt.unit.label)
+        x: float = 4.5
+        y = sc.units.m.to.nm(x)
 
-    plt.plot(x, sc.abs2(params.spec_0[o]))
-    state = sc.operators.CurrentState(
-        params.length, 0, params.step_size, 1.0, params.ifft, params.spec_0
-    )
-    # expD = np.exp(state.h / 2 * params.linear_operator(state))
-    # plt.plot(x, expD.imag[o], x, expD.real[o])
-    plt.plot(x, sc.abs2(params.nonlinear_operator(state))[o])
-    plt.yscale("log")
-    plt.xlim(100, 2000)
-    plt.show()
+        @app.update
+        def draw(i):
+            spec, *fields = compute(i)
+            spec_ax.set_line_data("spec", *spec, label=f"z = {params.z_targets[i]*1e2:.0f}cm")
+            for label, x, y in fields:
+                field_ax.set_line_data(label, x, y)
 
-    # for *_, spec in RK4IP(params).irun():
-    #     plt.plot(w[2:-2], sc.abs2(spec[ind]))
-    #     plt.show()
-    #     plt.close()
+        print(params)
+
+        @app.cache
+        def compute(i):
+            xt, field = sc.transform_1D_values(params.ifft(specs[i]), rt, params)
+            x, spec = sc.transform_1D_values(sc.abs2(specs[i]), rs, params, log=True)
+            # spec = np.where(spec > vmin, spec, vmin)
+            field2 = sc.abs2(field)
+            bot, top = sc.math.envelope_ind(field2)
+            return (x, spec), ("field^2", xt, field2), ("envelope", xt[top], field2[top])
+
+            # bot, top = sc.math.envelope_ind(field)
+            # bot = interp1d(xt[bot], field[bot], "cubic", bounds_error=False, fill_value=0)(xt)
+            # top = interp1d(xt[top], field[top], "cubic", bounds_error=False, fill_value=0)(xt)
+
+            # return ((x, spec), ("upper", xt, top), ("field", xt, field), ("lower", xt, bot))
 
 
 if __name__ == "__main__":