ENH: StochasticFreeFormFins for Monte Carlo simulations

CHANGELOG.md

@@ -32,6 +32,7 @@ Attention: The newest changes should be on top -->
 
 ### Added
 
+- ENH: `StochasticFreeFormFins`, so free-form fin sets can be used in Monte Carlo simulations. The outline is randomized as a block, since a shape is only meaningful as a complete set of points: every coordinate is perturbed by its own draw, the fin root is held on the body line, and a list of candidate outlines can have a different number of points in each. [#953](https://github.com/RocketPy-Team/RocketPy/issues/953)
 - ENH: Support for Open Meteo API in the `Environment` class, adding the `open_meteo` and `open_meteo_ensemble` atmospheric models. Pressure-level forecasts, past forecasts (from 2021 onwards) and ensembles are read straight from a keyless JSON API, with no external files and no netCDF/OPeNDAP dependency. [#520](https://github.com/RocketPy-Team/RocketPy/issues/520) [#1119](https://github.com/RocketPy-Team/RocketPy/pull/1119)
 - ENH: Add simplified opening shock force estimation [#1092](https://github.com/RocketPy-Team/RocketPy/pull/1092)
 - ENH: Add Qodo PR-Agent workflow using Google Gemini [#1089](https://github.com/RocketPy-Team/RocketPy/pull/1089)
@@ -47,9 +48,11 @@ Attention: The newest changes should be on top -->
 
 ### Fixed
 
+- BUG: Pick between the candidate values of a list input with the stochastic model's own seeded generator. `random.choice` was used, which draws from the interpreter-wide stream that `_set_stochastic` does not reseed, so a fixed seed did not reproduce the values chosen from a list, and Monte Carlo workers forked from one process walked a single shared stream instead of sampling independently. Fixed-seed baselines that vary a list input change. [#953](https://github.com/RocketPy-Team/RocketPy/issues/953)
 - BUG: Report the atmospheric model time period and ensemble member count for lower-case model types. `set_atmospheric_model` documents `type` as case-insensitive, but `Environment.info()` and `all_info()` compared against capitalised literals, so `type="ensemble"` printed no time period and no member count, and skipped the ensemble comparison plot. [#520](https://github.com/RocketPy-Team/RocketPy/issues/520)
 - BUG: Give each `CustomSampler` input its own deterministic stream, and seed samplers sharing one generator once as a group. Existing fixed-seed `CustomSampler` baselines change, and samplers built on the legacy `RandomState` must move to `default_rng` because seeds now carry the full 128 bits. [#1102](https://github.com/RocketPy-Team/RocketPy/pull/1102)
 - BUG: Accept the callable parachute triggers `StochasticParachute` documents, and reject the ones it cannot mean. An invalid string, an empty list or a boolean now fails during validation instead of reaching `Parachute` or becoming a one-metre height trigger. [#1103](https://github.com/RocketPy-Team/RocketPy/pull/1103)
+- BUG: Accept a deterministic aerodynamic surface in `StochasticRocket.add_nose`, `add_trapezoidal_fins`, `add_elliptical_fins` and `add_tail`. Each wrapped the surface with a `component=` keyword none of the stochastic classes accept, so passing anything other than an already-stochastic surface raised a `TypeError`. [#953](https://github.com/RocketPy-Team/RocketPy/issues/953)
 - BUG: rocket with a late-starting thrust curve never leaves the rail [#1085](https://github.com/RocketPy-Team/RocketPy/pull/1085)
 
 ## [v1.13.0] - 2026-07-21

docs/reference/classes/monte_carlo/stochastic_models/index.rst

@@ -19,6 +19,7 @@ input parameters, enabling robust Monte Carlo simulations.
    stochastic_nose_cone
    stochastic_trapezoidal_fins
    stochastic_elliptical_fins
+   stochastic_free_form_fins
    stochastic_tail
    stochastic_rail_buttons
    stochastic_rocket

docs/reference/classes/monte_carlo/stochastic_models/stochastic_free_form_fins.rst

@@ -0,0 +1,5 @@
+Stochastic Free Form Fins
+-------------------------
+
+.. autoclass:: rocketpy.stochastic.StochasticFreeFormFins
+    :members:

docs/user/stochastic.rst

@@ -93,6 +93,43 @@ passed in a few different ways:
     gives you the full control of how the samples are generated. See 
     :ref:`custom_sampler` for more details.
 
+.. note::
+    The formats above assume each argument holds a single number. The
+    ``shape_points`` of :class:`rocketpy.stochastic.StochasticFreeFormFins` is
+    the exception: a fin outline is only meaningful as a complete set of points,
+    so the deviation given applies to the outline as a block, with every
+    coordinate of every point perturbed by its own draw. The fin root is held on
+    the body line, so a point nominally at ``y = 0`` keeps that value and no
+    point ends up inside the airframe.
+
+    Because the deviation has to centre on the nominal coordinate, only the
+    distributions that read their first argument as that centre can be used here:
+    *"normal"*, *"gumbel"*, *"laplace"* and *"logistic"*. The others take bounds
+    (*"uniform"*) or shape parameters (*"wald"*, *"gamma"*, ...), which a set of
+    coordinates cannot be, and are rejected when the object is created.
+
+    A list means either one fixed outline, used as given, or a list of candidate
+    outlines to choose between, which do not have to have the same number of
+    points::
+
+        # One millimetre of deviation on every coordinate
+        StochasticFreeFormFins(free_form_fins=fins, shape_points=0.001)
+
+        # One fixed outline, not randomized
+        StochasticFreeFormFins(
+            free_form_fins=fins,
+            shape_points=[(0, 0), (0.08, 0.1), (0.12, 0)],
+        )
+
+        # Choose between two outlines
+        StochasticFreeFormFins(
+            free_form_fins=fins,
+            shape_points=[[(0, 0), (0.08, 0.1), (0.12, 0)], [(0, 0), (0.06, 0.12), (0.12, 0)]],
+        )
+
+    A ``CustomSampler`` given for this argument has to yield a whole outline per
+    sample, since what it returns replaces the outline instead of perturbing it.
+
 .. note::
     In statistics, the terms "Normal" and "Gaussian" refer to the same type of \
     distribution. This distribution is commonly used and is the default for the \

rocketpy/__init__.py

@@ -62,6 +62,7 @@
     StochasticEllipticalFins,
     StochasticEnvironment,
     StochasticFlight,
+    StochasticFreeFormFins,
     StochasticNoseCone,
     StochasticParachute,
     StochasticRocket,

rocketpy/stochastic/__init__.py

@@ -9,6 +9,7 @@
 from .stochastic_aero_surfaces import (
     StochasticAirBrakes,
     StochasticEllipticalFins,
+    StochasticFreeFormFins,
     StochasticNoseCone,
     StochasticRailButtons,
     StochasticTail,