torchode 0.2.0__tar.gz → 1.0.1__tar.gz

{torchode-0.2.0 → torchode-1.0.1}/CHANGELOG.md

@@ -5,6 +5,10 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## 1.0.0 - 2024-09-22
+
+After almost a year without any bugs found, we mark this as the first stable release!
+
 ## 0.2.0 - 2023-11-10
 
 ### Changed

torchode-1.0.1/CITATION.cff

@@ -0,0 +1,28 @@
+cff-version: 1.2.0
+title: torchode
+message: >-
+  If you use this software, please cite it using the
+  metadata from this file.
+type: software
+authors:
+  - given-names: Marten
+    family-names: Lienen
+    email: m.lienen@tum.de
+  - given-names: Stephan
+    family-names: Günnemann
+    email: s.guennemann@tum.de
+repository-code: "https://github.com/martenlienen/torchode"
+license: MIT
+preferred-citation:
+  type: conference-paper
+  title: "torchode: A Parallel ODE Solver for PyTorch"
+  authors:
+    - given-names: Marten
+      family-names: Lienen
+      email: m.lienen@tum.de
+    - given-names: Stephan
+      family-names: Günnemann
+      email: s.guennemann@tum.de
+  collection-title: "The Symbiosis of Deep Learning and Differential Equations II, NeurIPS"
+  year: 2022
+  url: "https://openreview.net/forum?id=uiKVKTiUYB0"

{torchode-0.2.0 → torchode-1.0.1}/PKG-INFO

@@ -1,16 +1,17 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: torchode
-Version: 0.2.0
+Version: 1.0.1
 Summary: A parallel ODE solver for PyTorch
 Author-email: Marten Lienen <m.lienen@tum.de>
 Requires-Python: >= 3.8
 Description-Content-Type: text/markdown
-Classifier: Development Status :: 4 - Beta
+Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Operating System :: OS Independent
 Classifier: Programming Language :: Python
+License-File: LICENSE
 Requires-Dist: sympy ~= 1.10
 Requires-Dist: torch >= 2.0
 Requires-Dist: torchtyping ~= 0.1.4

torchode-1.0.1/docs/backprop.md

@@ -0,0 +1,24 @@
+# Gradients & Backpropagation
+
+There are two ways to compute gradients of the dynamics of an ODE, so the neural network
+in the case of neural ODEs, with respect to the solution of the ODE. The first is to
+backpropagate straight through the solver. After all, an ODE solver is just a series of
+simple operations that define a dynamic computation graph that can be backpropagated
+through with pytorch's autograd. This is implemented in `to.AutoDiffAdjoint`, so called
+because it uses the autodiff/autograd mechanism. In general, this is the preferred method
+as long as you have enough memory, because it is fast and gives accurate gradients.
+
+If you run out of memory, you can compute gradients by solving the so called adjoint
+equations, which basically solve the ODE backwards and track gradients along the way. This
+is implemented in `to.BacksolveAdjoint`. Solving the adjoint equations requires the
+computation of gradients of the model at different steps in time, which
+`to.BacksolveAdjoint` implements with `torch.func`. If your model is not compatible and
+you get errors because of this, you can fall back to `to.JointBacksolveAdjoint`. This
+computes the model gradients with pytorch's usual autograd and should always work but
+comes with two caveats. However, to make this work, `to.JointBacksolveAdjoint` needs to
+solve the `n` independent adjoint equations jointly as one joint system that is jointly
+discretized. This breaks with torchode's approach of solving each ODE completely
+independently, because the joint discretization introduces a subtle coupling between the
+solutions of your batch of ODEs. Therefore, `to.JointBacksolveAdjoint` should be your
+backpropagation choice of last resort. Furthermore, it is only applicable if all ODEs in
+your batch have the same evaluation points.

torchode-1.0.1/docs/requirements.in

@@ -0,0 +1,2 @@
+mkdocs==1.6.1
+mkdocs-jupyter~=0.25

torchode-1.0.1/docs/requirements.txt

@@ -0,0 +1,226 @@
+#
+# This file is autogenerated by pip-compile with Python 3.14
+# by the following command:
+#
+#    pip-compile --output-file=docs/requirements.txt docs/requirements.in
+#
+asttokens==3.0.1
+    # via stack-data
+attrs==25.4.0
+    # via
+    #   jsonschema
+    #   referencing
+babel==2.18.0
+    # via mkdocs-material
+backrefs==6.2
+    # via mkdocs-material
+beautifulsoup4==4.14.3
+    # via nbconvert
+bleach[css]==6.3.0
+    # via nbconvert
+certifi==2026.2.25
+    # via requests
+charset-normalizer==3.4.6
+    # via requests
+click==8.3.1
+    # via mkdocs
+colorama==0.4.6
+    # via mkdocs-material
+comm==0.2.3
+    # via ipykernel
+debugpy==1.8.20
+    # via ipykernel
+decorator==5.2.1
+    # via ipython
+defusedxml==0.7.1
+    # via nbconvert
+executing==2.2.1
+    # via stack-data
+fastjsonschema==2.21.2
+    # via nbformat
+ghp-import==2.1.0
+    # via mkdocs
+idna==3.11
+    # via requests
+ipykernel==6.31.0
+    # via mkdocs-jupyter
+ipython==9.11.0
+    # via ipykernel
+ipython-pygments-lexers==1.1.1
+    # via ipython
+jedi==0.19.2
+    # via ipython
+jinja2==3.1.6
+    # via
+    #   mkdocs
+    #   mkdocs-material
+    #   nbconvert
+jsonschema==4.26.0
+    # via nbformat
+jsonschema-specifications==2025.9.1
+    # via jsonschema
+jupyter-client==8.8.0
+    # via
+    #   ipykernel
+    #   nbclient
+jupyter-core==5.9.1
+    # via
+    #   ipykernel
+    #   jupyter-client
+    #   nbclient
+    #   nbconvert
+    #   nbformat
+jupyterlab-pygments==0.3.0
+    # via nbconvert
+jupytext==1.19.1
+    # via mkdocs-jupyter
+markdown==3.10.2
+    # via
+    #   mkdocs
+    #   mkdocs-material
+    #   pymdown-extensions
+markdown-it-py==4.0.0
+    # via
+    #   jupytext
+    #   mdit-py-plugins
+markupsafe==3.0.3
+    # via
+    #   jinja2
+    #   mkdocs
+    #   nbconvert
+matplotlib-inline==0.2.1
+    # via
+    #   ipykernel
+    #   ipython
+mdit-py-plugins==0.5.0
+    # via jupytext
+mdurl==0.1.2
+    # via markdown-it-py
+mergedeep==1.3.4
+    # via
+    #   mkdocs
+    #   mkdocs-get-deps
+mistune==3.2.0
+    # via nbconvert
+mkdocs==1.6.1
+    # via
+    #   -r docs/requirements.in
+    #   mkdocs-jupyter
+    #   mkdocs-material
+mkdocs-get-deps==0.2.2
+    # via mkdocs
+mkdocs-jupyter==0.25.1
+    # via -r docs/requirements.in
+mkdocs-material==9.7.5
+    # via mkdocs-jupyter
+mkdocs-material-extensions==1.3.1
+    # via mkdocs-material
+nbclient==0.10.4
+    # via nbconvert
+nbconvert==7.17.0
+    # via mkdocs-jupyter
+nbformat==5.10.4
+    # via
+    #   jupytext
+    #   nbclient
+    #   nbconvert
+nest-asyncio==1.6.0
+    # via ipykernel
+packaging==26.0
+    # via
+    #   ipykernel
+    #   jupytext
+    #   mkdocs
+    #   nbconvert
+paginate==0.5.7
+    # via mkdocs-material
+pandocfilters==1.5.1
+    # via nbconvert
+parso==0.8.6
+    # via jedi
+pathspec==1.0.4
+    # via mkdocs
+pexpect==4.9.0
+    # via ipython
+platformdirs==4.9.4
+    # via
+    #   jupyter-core
+    #   mkdocs-get-deps
+prompt-toolkit==3.0.52
+    # via ipython
+psutil==7.2.2
+    # via ipykernel
+ptyprocess==0.7.0
+    # via pexpect
+pure-eval==0.2.3
+    # via stack-data
+pygments==2.19.2
+    # via
+    #   ipython
+    #   ipython-pygments-lexers
+    #   mkdocs-jupyter
+    #   mkdocs-material
+    #   nbconvert
+pymdown-extensions==10.21
+    # via mkdocs-material
+python-dateutil==2.9.0.post0
+    # via
+    #   ghp-import
+    #   jupyter-client
+pyyaml==6.0.3
+    # via
+    #   jupytext
+    #   mkdocs
+    #   mkdocs-get-deps
+    #   pymdown-extensions
+    #   pyyaml-env-tag
+pyyaml-env-tag==1.1
+    # via mkdocs
+pyzmq==27.1.0
+    # via
+    #   ipykernel
+    #   jupyter-client
+referencing==0.37.0
+    # via
+    #   jsonschema
+    #   jsonschema-specifications
+requests==2.32.5
+    # via mkdocs-material
+rpds-py==0.30.0
+    # via
+    #   jsonschema
+    #   referencing
+six==1.17.0
+    # via python-dateutil
+soupsieve==2.8.3
+    # via beautifulsoup4
+stack-data==0.6.3
+    # via ipython
+tinycss2==1.4.0
+    # via bleach
+tornado==6.5.5
+    # via
+    #   ipykernel
+    #   jupyter-client
+traitlets==5.14.3
+    # via
+    #   ipykernel
+    #   ipython
+    #   jupyter-client
+    #   jupyter-core
+    #   matplotlib-inline
+    #   nbclient
+    #   nbconvert
+    #   nbformat
+typing-extensions==4.15.0
+    # via beautifulsoup4
+urllib3==2.6.3
+    # via requests
+watchdog==6.0.0
+    # via mkdocs
+wcwidth==0.6.0
+    # via prompt-toolkit
+webencodings==0.5.1
+    # via
+    #   bleach
+    #   tinycss2

{torchode-0.2.0 → torchode-1.0.1}/mkdocs.yml

@@ -12,6 +12,7 @@ plugins:
 nav:
   - "Introduction": README.md
   - step-size-controllers.md
+  - backprop.md
   - jit.ipynb
   - extra-args.md
   - extra-stats.ipynb

{torchode-0.2.0 → torchode-1.0.1}/pyproject.toml

@@ -10,7 +10,7 @@ requires-python = ">= 3.8"
 license = {file = "LICENSE"}
 authors = [{name = "Marten Lienen", email = "m.lienen@tum.de"}]
 classifiers = [
-  "Development Status :: 4 - Beta",
+  "Development Status :: 5 - Production/Stable",
   "Intended Audience :: Developers",
   "Intended Audience :: Science/Research",
   "License :: OSI Approved :: MIT License",

{torchode-0.2.0 → torchode-1.0.1}/torchode/__init__.py

@@ -1,6 +1,6 @@
 """A parallel ODE solver for PyTorch"""
 
-__version__ = "0.2.0"
+__version__ = "1.0.1"
 
 from .adjoints import AutoDiffAdjoint, BacksolveAdjoint, JointBacksolveAdjoint
 from .interface import register_method, solve_ivp

{torchode-0.2.0 → torchode-1.0.1}/torchode/adjoints.py

@@ -207,10 +207,11 @@ class AutoDiffAdjoint(nn.Module):
             # Evaluate the solution at all evaluation points that have been passed in
             # this step.
             #
-            # We always build the interpolation and evaluate it, even if no evaluation
-            # points have actually been passed, because this avoids a CPU-GPU
-            # synchronization and for time series models we expect that most steps will
-            # pass at least one evaluation point across the whole batch (usually more).
+            # This causes a blocking CPU-GPU sync point at to_be_evaluated.any 
+            # when t_eval is not None, but deferring this sync doesn't seem to 
+            # yield a speedup. A sync is necessary at each evaluation point anyway
+            # since nonzero produces a variable-shape result.
+            # See https://github.com/martenlienen/torchode/issues/46
             if t_eval is not None:
                 to_be_evaluated = (
                     torch.addcmul(

{torchode-0.2.0 → torchode-1.0.1}/torchode/step_size_controllers.py

@@ -478,7 +478,7 @@ class IntegralController(nn.Module):
             args,
         )
 
-        d2 = norm((f1 - f0) * inv_scale) / dt0
+        d2 = torch.where(dt0 == 0, torch.inf, norm((f1 - f0) * inv_scale) / dt0)
 
         maxd1d2 = torch.maximum(d1, d2)
         dt1 = torch.where(

torchode-0.2.0/docs/requirements.in

@@ -1,2 +0,0 @@
-mkdocs==1.4.1
-mkdocs-jupyter~=0.21

torchode-0.2.0/docs/requirements.txt

@@ -1,154 +0,0 @@
-#
-# This file is autogenerated by pip-compile with python 3.10
-# To update, run:
-#
-#    pip-compile --output-file=docs/requirements.txt docs/requirements.in
-#
-attrs==22.1.0
-    # via jsonschema
-beautifulsoup4==4.11.1
-    # via nbconvert
-bleach==5.0.1
-    # via nbconvert
-certifi==2022.12.7
-    # via requests
-charset-normalizer==2.1.1
-    # via requests
-click==8.1.3
-    # via mkdocs
-defusedxml==0.7.1
-    # via nbconvert
-entrypoints==0.4
-    # via
-    #   jupyter-client
-    #   nbconvert
-fastjsonschema==2.16.2
-    # via nbformat
-ghp-import==2.1.0
-    # via mkdocs
-idna==3.4
-    # via requests
-jinja2==3.1.2
-    # via
-    #   mkdocs
-    #   mkdocs-material
-    #   nbconvert
-jsonschema==4.17.0
-    # via nbformat
-jupyter-client==7.4.4
-    # via nbclient
-jupyter-core==4.11.2
-    # via
-    #   jupyter-client
-    #   nbconvert
-    #   nbformat
-jupyterlab-pygments==0.2.2
-    # via nbconvert
-jupytext==1.14.1
-    # via mkdocs-jupyter
-lxml==4.9.1
-    # via nbconvert
-markdown==3.3.7
-    # via
-    #   mkdocs
-    #   mkdocs-material
-    #   pymdown-extensions
-markdown-it-py==2.2.0
-    # via
-    #   jupytext
-    #   mdit-py-plugins
-markupsafe==2.1.1
-    # via
-    #   jinja2
-    #   nbconvert
-mdit-py-plugins==0.3.1
-    # via jupytext
-mdurl==0.1.2
-    # via markdown-it-py
-mergedeep==1.3.4
-    # via mkdocs
-mistune==0.8.4
-    # via nbconvert
-mkdocs==1.4.1
-    # via
-    #   -r requirements.in
-    #   mkdocs-jupyter
-    #   mkdocs-material
-mkdocs-jupyter==0.22.0
-    # via -r requirements.in
-mkdocs-material==8.5.7
-    # via mkdocs-jupyter
-mkdocs-material-extensions==1.1
-    # via mkdocs-material
-nbclient==0.7.0
-    # via nbconvert
-nbconvert==6.5.4
-    # via mkdocs-jupyter
-nbformat==5.7.0
-    # via
-    #   jupytext
-    #   nbclient
-    #   nbconvert
-nest-asyncio==1.5.6
-    # via
-    #   jupyter-client
-    #   nbclient
-packaging==21.3
-    # via
-    #   mkdocs
-    #   nbconvert
-pandocfilters==1.5.0
-    # via nbconvert
-pygments==2.13.0
-    # via
-    #   mkdocs-jupyter
-    #   mkdocs-material
-    #   nbconvert
-pymdown-extensions==9.7
-    # via mkdocs-material
-pyparsing==3.0.9
-    # via packaging
-pyrsistent==0.19.1
-    # via jsonschema
-python-dateutil==2.8.2
-    # via
-    #   ghp-import
-    #   jupyter-client
-pyyaml==6.0
-    # via
-    #   jupytext
-    #   mkdocs
-    #   pyyaml-env-tag
-pyyaml-env-tag==0.1
-    # via mkdocs
-pyzmq==24.0.1
-    # via jupyter-client
-requests==2.28.1
-    # via mkdocs-material
-six==1.16.0
-    # via
-    #   bleach
-    #   python-dateutil
-soupsieve==2.3.2.post1
-    # via beautifulsoup4
-tinycss2==1.2.1
-    # via nbconvert
-toml==0.10.2
-    # via jupytext
-tornado==6.2
-    # via jupyter-client
-traitlets==5.5.0
-    # via
-    #   jupyter-client
-    #   jupyter-core
-    #   nbclient
-    #   nbconvert
-    #   nbformat
-urllib3==1.26.12
-    # via requests
-watchdog==2.1.9
-    # via mkdocs
-webencodings==0.5.1
-    # via
-    #   bleach
-    #   tinycss2