From 3941be22d17836e1f8486e9e111267a9397c7615 Mon Sep 17 00:00:00 2001
From: Florian Pfaff <6773539+FlorianPfaff@users.noreply.github.com>
Date: Mon, 29 Jun 2026 11:09:04 +0200
Subject: [PATCH 1/2] Fix PyTorch FFT array-like input handling

---
 src/pyrecest/_backend/pytorch/fft.py | 50 +++++++++++++++++++++++-----
 1 file changed, 42 insertions(+), 8 deletions(-)

diff --git a/src/pyrecest/_backend/pytorch/fft.py b/src/pyrecest/_backend/pytorch/fft.py
index 262d6800c..f9c507da7 100644
--- a/src/pyrecest/_backend/pytorch/fft.py
+++ b/src/pyrecest/_backend/pytorch/fft.py
@@ -1,10 +1,44 @@
 # For ffts. Added for pyrecest.
 import torch as _torch
-from torch.fft import (
-    fftn,
-    fftshift,
-    ifftn,
-    ifftshift,
-    irfft,
-    rfft,
-)
+
+
+def _as_tensor(x):
+    return x if _torch.is_tensor(x) else _torch.as_tensor(x)
+
+
+def _resolve_dim_alias(dim, alias, alias_name, func_name, *, default=None):
+    if alias is None:
+        return default if dim is None else dim
+    if dim is not None and dim != alias:
+        raise TypeError(f"{func_name}() got both 'dim' and '{alias_name}'")
+    return alias
+
+
+def fftn(input, s=None, dim=None, norm=None, *, axes=None, out=None):
+    dim = _resolve_dim_alias(dim, axes, "axes", "fftn")
+    return _torch.fft.fftn(_as_tensor(input), s=s, dim=dim, norm=norm, out=out)
+
+
+def ifftn(input, s=None, dim=None, norm=None, *, axes=None, out=None):
+    dim = _resolve_dim_alias(dim, axes, "axes", "ifftn")
+    return _torch.fft.ifftn(_as_tensor(input), s=s, dim=dim, norm=norm, out=out)
+
+
+def rfft(input, n=None, dim=None, norm=None, *, axis=None, out=None):
+    dim = _resolve_dim_alias(dim, axis, "axis", "rfft", default=-1)
+    return _torch.fft.rfft(_as_tensor(input), n=n, dim=dim, norm=norm, out=out)
+
+
+def irfft(input, n=None, dim=None, norm=None, *, axis=None, out=None):
+    dim = _resolve_dim_alias(dim, axis, "axis", "irfft", default=-1)
+    return _torch.fft.irfft(_as_tensor(input), n=n, dim=dim, norm=norm, out=out)
+
+
+def fftshift(input, dim=None, *, axes=None):
+    dim = _resolve_dim_alias(dim, axes, "axes", "fftshift")
+    return _torch.fft.fftshift(_as_tensor(input), dim=dim)
+
+
+def ifftshift(input, dim=None, *, axes=None):
+    dim = _resolve_dim_alias(dim, axes, "axes", "ifftshift")
+    return _torch.fft.ifftshift(_as_tensor(input), dim=dim)

From b31bf9ab80cf1e53fd16f4d96b8d69322d8ad299 Mon Sep 17 00:00:00 2001
From: Florian Pfaff <6773539+FlorianPfaff@users.noreply.github.com>
Date: Mon, 29 Jun 2026 11:09:21 +0200
Subject: [PATCH 2/2] Add PyTorch FFT backend regression test

---
 .../test_pytorch_fft_numpy_signature.py       | 71 +++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 tests/backend_support/test_pytorch_fft_numpy_signature.py

diff --git a/tests/backend_support/test_pytorch_fft_numpy_signature.py b/tests/backend_support/test_pytorch_fft_numpy_signature.py
new file mode 100644
index 000000000..7424fafcd
--- /dev/null
+++ b/tests/backend_support/test_pytorch_fft_numpy_signature.py
@@ -0,0 +1,71 @@
+from __future__ import annotations
+
+import importlib.util
+
+import pytest
+
+from tests.support.backend_runner import run_backend_code
+
+
+def test_pytorch_fft_accepts_array_like_inputs_and_numpy_axis_aliases():
+    if importlib.util.find_spec("torch") is None:
+        pytest.skip("PyTorch is not installed")
+
+    result = run_backend_code(
+        "pytorch",
+        """
+import numpy as np
+import pyrecest.backend as backend
+
+values = [[1.0, 2.0], [3.0, 4.0]]
+fftn_result = backend.fft.fftn(values, axes=(0, 1))
+np.testing.assert_allclose(
+    backend.to_numpy(fftn_result),
+    np.fft.fftn(np.asarray(values), axes=(0, 1)),
+)
+
+rfft_result = backend.fft.rfft([0.0, 1.0, 0.0, -1.0], axis=0)
+round_trip = backend.fft.irfft(rfft_result, n=4, axis=0)
+np.testing.assert_allclose(
+    backend.to_numpy(round_trip),
+    np.asarray([0.0, 1.0, 0.0, -1.0]),
+)
+
+np.testing.assert_array_equal(
+    backend.to_numpy(backend.fft.fftshift([0, 1, 2, 3], axes=0)),
+    np.asarray([2, 3, 0, 1]),
+)
+np.testing.assert_array_equal(
+    backend.to_numpy(backend.fft.ifftshift([2, 3, 0, 1], axes=0)),
+    np.asarray([0, 1, 2, 3]),
+)
+print("ok")
+""",
+    )
+
+    assert result.returncode == 0, result.stderr
+    assert "ok" in result.stdout
+
+
+def test_pytorch_fft_rejects_conflicting_axis_aliases():
+    if importlib.util.find_spec("torch") is None:
+        pytest.skip("PyTorch is not installed")
+
+    result = run_backend_code(
+        "pytorch",
+        """
+import pyrecest.backend as backend
+
+try:
+    backend.fft.rfft([0.0, 1.0], dim=0, axis=1)
+except TypeError as exc:
+    assert "dim" in str(exc)
+    assert "axis" in str(exc)
+else:
+    raise AssertionError("conflicting dim/axis aliases were accepted")
+print("ok")
+""",
+    )
+
+    assert result.returncode == 0, result.stderr
+    assert "ok" in result.stdout