Merge branch 'main' into compathelper/new_version/2023-05-15-01-21-20…

…-644-00122993255

.github/workflows/CI.yml

@@ -20,7 +20,8 @@ jobs:
       matrix:
         version:
           - '1.6'
-          - '1.8'
+          - '1'
+          - 'nightly'
         os:
           - ubuntu-latest
         arch:

.github/workflows/Downstream.yml

@@ -15,7 +15,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        julia-version: [1,1.6]
+        julia-version: [1]
         os: [ubuntu-latest]
         package:
           - {user: nathanaelbosch, repo: Fenrir.jl, group: All}

Project.toml

@@ -1,13 +1,14 @@
 name = "ProbNumDiffEq"
 uuid = "bf3e78b0-7d74-48a5-b855-9609533b56a5"
 authors = ["Nathanael Bosch"]
-version = "0.11.1"
+version = "0.12.1"
 
 [deps]
 DiffEqBase = "2b5f629d-d688-5b77-993f-72d75c75574e"
 DiffEqDevTools = "f3b72e0c-5b89-59e1-b016-84e28bfd966d"
 ExponentialUtilities = "d4d017d3-3776-5f7e-afef-a10c40355c18"
 FastBroadcast = "7034ab61-46d4-4ed7-9d0f-46aef9175898"
+FastGaussQuadrature = "442a2c76-b920-505d-bb47-c5924d526838"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 FunctionWrappersWrappers = "77dc65aa-8811-40c2-897b-53d922fa7daf"
 GaussianDistributions = "43dcc890-d446-5863-8d1a-14597580bb8d"
@@ -34,11 +35,12 @@ DiffEqBase = "6.122"
 DiffEqDevTools = "2"
 ExponentialUtilities = "1"
 FastBroadcast = "0.2"
+FastGaussQuadrature = "0.5"
 ForwardDiff = "0.10"
 FunctionWrappersWrappers = "0.1.3"
 GaussianDistributions = "0.5"
 Octavian = "0.3.17"
-OrdinaryDiffEq = "6.49.1"
+OrdinaryDiffEq = "6.52"
 PSDMatrices = "0.4.2"
 PrecompileTools = "1"
 RecipesBase = "1"

docs/Project.toml

@@ -1,7 +1,9 @@
 [deps]
+Bibliography = "f1be7e48-bf82-45af-a471-ae754a193061"
 DiffEqUncertainty = "ef61062a-5684-51dc-bb67-a0fcdec5c97d"
 DifferentialEquations = "0c46a032-eb83-5123-abaf-570d42b7fbaa"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
+DocumenterCitations = "daee34ce-89f3-4625-b898-19384cb65244"
 Fenrir = "e9b4b195-f5cd-427c-8076-5358c553c37f"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 LiveServer = "16fef848-5104-11e9-1b77-fb7a48bbb589"

docs/make.jl

@@ -1,23 +1,50 @@
 using Documenter
 using ProbNumDiffEq
 
+using DocumenterCitations, Bibliography
+
+# DocumenterCitations.bib_sorting(::Val{:numeric}) = :nyt
+# function DocumenterCitations.format_bibliography_label(
+#     ::Val{:numeric},
+#     entry,
+#     citations::DocumenterCitations.OrderedDict{String,Int64},
+# )
+#     key = entry.id
+#     sorted_bibtex_keys = citations |> keys |> collect |> sort
+#     i = findfirst(x -> x == key, sorted_bibtex_keys)
+#     @info "format_bibliography_label" entry.id citations sorted_bibtex_keys i entry.date
+#     return "[$i]"
+# end
+# Bibliography.sorting_rules[:nyt] = [:authors; :date; :title]
+
+bib = CitationBibliography(
+    joinpath(@__DIR__, "src", "refs.bib"),
+    style=:numeric,
+    # style=:authoryear,
+)
+sort_bibliography!(bib.entries, :nyt)  # name-year-title
+
 makedocs(
+    bib,
     sitename="ProbNumDiffEq.jl",
-    format=Documenter.HTML(),
+    format=Documenter.HTML(
+        assets=String["assets/citations.css"],
+    ),
     modules=[ProbNumDiffEq],
     pages=[
         "Home" => "index.md",
         "Tutorials" => [
-            "Getting Started" => "getting_started.md"
-            "Second Order ODEs and Energy Preservation" => "dynamical_odes.md"
-            "Differential Algebraic Equations" => "dae.md"
-            "Parameter Inference" => "fenrir.md"
+            "Getting Started" => "tutorials/getting_started.md"
+            "Second Order ODEs and Energy Preservation" => "tutorials/dynamical_odes.md"
+            "Differential Algebraic Equations" => "tutorials/dae.md"
+            "Probabilistic Exponential Integrators" => "tutorials/exponential_integrators.md"
+            "Parameter Inference" => "tutorials/fenrir.md"
         ],
         "Solvers and Options" => [
             "solvers.md",
-            "diffusions.md",
-            "initialization.md",
             "priors.md",
+            "initialization.md",
+            "diffusions.md",
         ],
         "Benchmarks" => [
             "Multi-Language Wrapper Benchmark" => "benchmarks/multi-language-wrappers.md",
@@ -29,6 +56,7 @@ makedocs(
             "Filtering and Smoothing" => "filtering.md"
             "Implementation via OrdinaryDiffEq.jl" => "implementation.md"
         ],
+        "References" => "references.md",
     ],
 )
 

docs/src/assets/citations.css

@@ -0,0 +1,17 @@
+.citation dl {
+    display: grid;
+    grid-template-columns: max-content auto; }
+.citation dt {
+    grid-column-start: 1; }
+.citation dd {
+    grid-column-start: 2;
+    margin-bottom: 0.75em; }
+.citation ul {
+    padding: 0 0 2.25em 0;
+    margin: 0;
+    list-style: none;}
+.citation ul li {
+    text-indent: -2.25em;
+    margin: 0.33em 0.5em 0.5em 2.25em;}
+.citation ol li {
+    padding-left:0.75em;}

docs/src/diffusions.md

@@ -44,7 +44,7 @@ Or more compactly:
 | Time-fixed   | [`FixedDiffusion`](@ref)   | [`FixedMVDiffusion`](@ref)            |
 
 
-For more details on diffusions and calibration, check out this paper [[1]](@ref diffusionrefs).
+For more details on diffusions and calibration, check out this paper [bosch20capos](@cite).
 
 
 ## API
@@ -59,4 +59,9 @@ FixedMVDiffusion
 
 ## [References](@id diffusionrefs)
 
-[1] N. Bosch, P. Hennig, F. Tronarp: **Calibrated Adaptive Probabilistic ODE Solvers** (2021) ([link](http://proceedings.mlr.press/v130/bosch21a.html))
+```@bibliography
+Pages = []
+Canonical = false
+
+bosch20capos
+```

docs/src/index.md

@@ -8,20 +8,13 @@ The implemented _ODE filters_ solve differential equations via Bayesian filterin
 
 For a short intro video, check out our [poster presentation at JuliaCon2021](https://www.youtube.com/watch?v=EMFl6ytP3iQ).
 
----
-
-__For more probabilistic numerics check out the [ProbNum](https://probnum.readthedocs.io/en/latest/) Python package.__
-It implements probabilistic ODE solvers, but also probabilistic linear solvers, Bayesian quadrature, and many filtering and smoothing implementations.
-
----
-
 ## Installation
 
 Run Julia, enter `]` to bring up Julia's package manager, and add the ProbNumDiffEq.jl package:
 
 ```
 julia> ]
-(v1.8) pkg> add ProbNumDiffEq
+(v1.9) pkg> add ProbNumDiffEq
 ```
 
 ## Getting Started
@@ -30,28 +23,22 @@ For a quick introduction check out the "[Solving ODEs with Probabilistic Numeric
 
 ## Features
 
-  - Two extended Kalman filtering-based probabilistic solvers: the explicit [`EK0`](@ref) and semi-implicit [`EK1`](@ref).
-  - Adaptive step-size selection (by default with PI control)
-  - On-line uncertainty calibration, for multiple different measurement models
-  - Dense output
-  - Sampling from the solution
-  - Callback support
-  - Convenient plotting through a Plots.jl recipe
-  - Automatic differentiation via ForwardDiff.jl
-  - Supports arbitrary precision numbers via BigFloats.jl
-  - Specialized solvers for second-order ODEs (demo will be added)
-  - Compatible with DAEs in mass-matrix ODE form (demo will be added)
-
-
-## [References](@id references)
-
-  - N. Bosch, F. Tronarp, P. Hennig: **Pick-and-Mix Information Operators for Probabilistic ODE Solvers** (2022)
-  - N. Krämer, N. Bosch, J. Schmidt, P. Hennig: **Probabilistic ODE Solutions in Millions of Dimensions** (2021)
-  - N. Bosch, P. Hennig, F. Tronarp: **Calibrated Adaptive Probabilistic ODE Solvers** (2021)
-  - F. Tronarp, S. Särkkä, and P. Hennig: **Bayesian ODE Solvers: The Maximum A Posteriori Estimate** (2021)
-  - N. Krämer, P. Hennig: **Stable Implementation of Probabilistic ODE Solvers** (2020)
-  - H. Kersting, T. J. Sullivan, and P. Hennig: **Convergence Rates of Gaussian Ode Filters** (2020)
-  - F. Tronarp, H. Kersting, S. Särkkä, and P. Hennig: **Probabilistic Solutions To Ordinary Differential Equations As Non-Linear Bayesian Filtering: A New Perspective** (2019)
-  - M. Schober, S. Särkkä, and P. Hennig: **A Probabilistic Model for the Numerical Solution of Initial Value Problems** (2018)
-
-A much more detailed list of references, not only on ODE filters but on probabilistic numerics in general, can be found on the [probabilistic-numerics.org homepage](https://www.probabilistic-numerics.org/research/general/).
+- Two extended Kalman filtering-based probabilistic solvers: the explicit [`EK0`](@ref) and semi-implicit [`EK1`](@ref).
+- Adaptive step-size selection with PI control;
+  fully compatible with [DifferentialEquations.jl's timestepping options](https://docs.sciml.ai/DiffEqDocs/stable/extras/timestepping/)
+- Online uncertainty calibration for multiple different diffusion models (see "[Diffusion models and calibration](@ref)")
+- [Dense output](@ref)
+- Sampling from the solution
+- Callback support
+- Convenient plotting through a [Plots.jl](https://docs.juliaplots.org/latest/) recipe
+- Automatic differentiation via [ForwardDiff.jl](https://github.com/JuliaDiff/ForwardDiff.jl)
+- Arbitrary precision via Julia's built-in [arbitrary precision arithmetic](https://docs.julialang.org/en/v1/manual/integers-and-floating-point-numbers/#Arbitrary-Precision-Arithmetic)
+- Specialized solvers for second-order ODEs (see [Second Order ODEs and Energy Preservation](@ref))
+- Compatible with DAEs in mass-matrix ODE form (see [Solving DAEs with Probabilistic Numerics](@ref))
+
+
+## Related packages
+
+- [probdiffeq](https://pnkraemer.github.io/probdiffeq/): Fast and feature-rich filtering-based probabilistic ODE solvers in JAX.
+- [ProbNum](https://probnum.readthedocs.io/en/latest/): Probabilistic numerics in Python. It has not only probabilistic ODE solvers, but also probabilistic linear solvers, Bayesian quadrature, and many filtering and smoothing implementations.
+- [Fenrir.jl](https://github.com/nathanaelbosch/Fenrir.jl): Parameter-inference in ODEs with probabilistic ODE solvers. This package builds on ProbNumDiffEq.jl to provide a negative marginal log-likelihood function, which can then be used with an optimizer or with MCMC for parameter inference.

docs/src/initialization.md

@@ -48,8 +48,8 @@ Y^{(0)}(0) &= y_0, \\
 Y^{(1)}(0) &= f(y_0, 0).
 \end{aligned}
 ```
-And it turns out that we can also compute higher-order derivatives of ``y`` with the chain rule,
-which we can use to initialize ``Y^{(i)}(0)``.
+It turns out that we can also compute higher-order derivatives of ``y`` with the chain rule,
+and then use these to better initialize ``Y^{(i)}(0)``.
 This, done efficiently with Taylor-mode autodiff by using
 [TaylorIntegration.jl](https://perezhz.github.io/TaylorIntegration.jl/latest/),
 is what [`TaylorModeInit`](@ref) does.
@@ -65,7 +65,13 @@ TaylorModeInit
 ClassicSolverInit
 ```
 
+
 ## [References](@id initrefs)
 
-[1] N. Krämer, P. Hennig: **Stable Implementation of Probabilistic ODE Solvers** (2020) ([link](https://arxiv.org/abs/2012.10106))
-[2] M. Schober, S. Särkkä, and P. Hennig: **A Probabilistic Model for the Numerical Solution of Initial Value Problems** (2018) ([link](https://link.springer.com/article/10.1007/s11222-017-9798-7))
+```@bibliography
+Pages = []
+Canonical = false
+
+kraemer20stableimplementation
+schober16probivp
+```

docs/src/references.md

@@ -0,0 +1,5 @@
+# [References](@id references)
+
+```@bibliography
+*
+```