Add invertible Asinh processor

sdm/processing/__init__.py

@@ -38,6 +38,7 @@
     RobustScale,
     MinMaxScale,
     FlipSign,
+    Asinh,
     DropConstantColumns,
     PCA,
     RandomProjection,
@@ -93,6 +94,7 @@
     "RobustScale",
     "MinMaxScale",
     "FlipSign",
+    "Asinh",
     "DropConstantColumns",
     "PCA",
     "RandomProjection",

sdm/processing/numerical/__init__.py

@@ -15,6 +15,7 @@
 from sdm.processing.numerical.robust_scale import RobustScale
 from sdm.processing.numerical.min_max_scale import MinMaxScale
 from sdm.processing.numerical.flip_sign import FlipSign
+from sdm.processing.numerical.asinh import Asinh
 from sdm.processing.numerical.constant import DropConstantColumns
 from sdm.processing.numerical.pca import PCA
 from sdm.processing.numerical.random_projection import RandomProjection
@@ -35,6 +36,7 @@
     "RobustScale",
     "MinMaxScale",
     "FlipSign",
+    "Asinh",
     "DropConstantColumns",
     "PCA",
     "RandomProjection",

sdm/processing/numerical/asinh.py

@@ -0,0 +1,18 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+from sdm import Stype, TableTensor
+from sdm.processing import InvertibleMixin, Processor
+
+
+class Asinh(Processor, InvertibleMixin):
+    r"""Apply the inverse hyperbolic sine to numerical values."""
+
+    handles_stypes = (Stype.numerical,)
+    requires_fit = False
+
+    def _transform(self, table: TableTensor) -> TableTensor:
+        return table.replace_blocks(numerical=table.numerical.asinh())
+
+    def _inverse_transform(self, table: TableTensor) -> TableTensor:
+        return table.replace_blocks(numerical=table.numerical.sinh())

test/processing/numerical/test_asinh.py

@@ -0,0 +1,17 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+import torch
+
+from sdm import TableTensor
+from sdm.processing import Asinh
+
+
+def test_asinh_transform_and_inverse() -> None:
+    table = TableTensor.from_tensor(torch.randn(3, 2))
+    processor = Asinh()
+
+    out = processor.transform(table)
+    torch.testing.assert_close(out.numerical, table.numerical.asinh())
+    out = processor.inverse_transform(out)
+    torch.testing.assert_close(out.numerical, table.numerical)

test/processing/test_contract.py

@@ -131,6 +131,7 @@ def _make_processor_pair(
     ProcessorCase(sp.RobustScale()),
     ProcessorCase(sp.MinMaxScale()),
     ProcessorCase(sp.FlipSign()),
+    ProcessorCase(sp.Asinh()),
     ProcessorCase(sp.DropConstantColumns()),
     ProcessorCase(sp.PCA(2)),
     ProcessorCase(sp.RandomProjection(2)),