Merge branch 'main' into make-formatting-a-separate-action

.github/workflows/CI.yml

@@ -27,7 +27,7 @@ jobs:
           journal: joss
           # This should be the path to the paper within your repo.
           paper-path: paper/paper.md
-      - uses: actions/upload-artifact@v1
+      - uses: actions/upload-artifact@v4
         with:
           name: paper
           # This is the output path where Pandoc will write the compiled
@@ -42,7 +42,7 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.10'
+          - '1.11'
         os:
           - ubuntu-latest
         arch:
@@ -62,7 +62,7 @@ jobs:
         env:
           CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
       - name: Archive test artifacts
-        uses: actions/upload-artifact@v2
+        uses: actions/upload-artifact@v4
         with:
           name: test-output
           path: test/_output

Project.toml

@@ -1,7 +1,7 @@
 name = "RxInfer"
 uuid = "86711068-29c9-4ff7-b620-ae75d7495b3d"
 authors = ["Bagaev Dmitry <d.v.bagaev@tue.nl> and contributors"]
-version = "3.4.0"
+version = "3.7.1"
 
 [deps]
 BayesBase = "b4ee3484-f114-42fe-b91c-797d54a0c67e"
@@ -19,24 +19,33 @@ Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 ReactiveMP = "a194aa59-28ba-4574-a09c-4a745416d6e3"
 Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 Rocket = "df971d30-c9d6-4b37-b8ff-e965b2cb3a40"
+Static = "aedffcd0-7271-4cad-89d0-dc628f76c6d3"
 TupleTools = "9d95972d-f1c8-5527-a6e0-b4b365fa01f6"
 
+[weakdeps]
+ExponentialFamilyProjection = "17f509fa-9a96-44ba-99b2-1c5f01f0931b"
+
+[extensions]
+ProjectionExt = "ExponentialFamilyProjection"
+
 [compat]
 BayesBase = "1.1"
 DataStructures = "0.18"
 Distributions = "0.25"
 DomainSets = "0.5.2, 0.6, 0.7"
-ExponentialFamily = "1.2"
+ExponentialFamily = "1.5"
+ExponentialFamilyProjection = "1.1"
 FastCholesky = "1.3.0"
 GraphPPL = "~4.3.0"
 LinearAlgebra = "1.9"
 MacroTools = "0.5.6"
 Optim = "1.0.0"
 ProgressMeter = "1.0.0"
 Random = "1.9"
-ReactiveMP = "~4.2.0"
+ReactiveMP = "~4.4.1"
 Reexport = "1.2.0"
 Rocket = "1.8.0"
+Static = "0.8.10, 1"
 TupleTools = "1.2.0"
 julia = "1.10"
 
@@ -49,6 +58,7 @@ CpuId = "adafc99b-e345-5852-983c-f28acb93d879"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
+ExponentialFamilyProjection = "17f509fa-9a96-44ba-99b2-1c5f01f0931b"
 InteractiveUtils = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
 Logging = "56ddb016-857b-54e1-b83d-db4d58db5568"
 Optimisers = "3bd65402-5787-11e9-1adc-39752487f4e2"
@@ -62,4 +72,4 @@ Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 TestSetExtensions = "98d24dd4-01ad-11ea-1b02-c9a08f80db04"
 
 [targets]
-test = ["Test", "Pkg", "Logging", "InteractiveUtils", "TestSetExtensions", "Coverage", "CpuId", "Dates", "Distributed", "Documenter", "Plots", "BenchmarkCI", "BenchmarkTools", "PkgBenchmark", "Aqua", "StableRNGs", "StatsFuns", "Optimisers", "ReTestItems"]
+test = ["Test", "Pkg", "Logging", "InteractiveUtils", "TestSetExtensions", "Coverage", "CpuId", "Dates", "Distributed", "Documenter", "ExponentialFamilyProjection", "Plots", "BenchmarkCI", "BenchmarkTools", "PkgBenchmark", "Aqua", "StableRNGs", "StatsFuns", "Optimisers", "ReTestItems"]

README.md

@@ -160,40 +160,63 @@ result = infer(
 
 ![Coin Flip](docs/src/assets/img/coin-flip.svg?raw=true&sanitize=true "Coin-Flip readme results")
 
-# Where to go next?
+# Roadmap
 
-There are a set of [examples](https://reactivebayes.github.io/RxInfer.jl/stable/examples/overview/) available in `RxInfer` repository that demonstrate the more advanced features of the package. Alternatively, you can head to the [documentation](https://reactivebayes.github.io/RxInfer.jl/stable/) that provides more detailed information of how to use `RxInfer` to specify more complex probabilistic models.
+Our high-level project roadmap outlines the key milestones and focus areas for the upcoming years:
 
-Additionally, checkout our [video from JuliaCon 2023](https://www.youtube.com/watch?v=qXrvDVm_fnE) for a high-level overview of the package
+| Q1/Q2 2024          | Q3/Q4 2024                | 2025                | 
+|---------------------|---------------------------|--------------------|
+| 🧩 **Nested models with [GraphPPL.jl](https://github.com/reactivebayes/GraphPPL.jl)** ✅    | 🌐 **Graph structure visualization** | 🔀 **Stochastic Processes** |
+| 🔄 **Development of [ExponentialFamilyProjection.jl]()** ✅                | 🧠 **Automated inference with [ExponentialFamilyProjection.jl](https://github.com/reactivebayes/ExponentialFamilyProjection.jl)** | 🚀 **Robustness & Memory-efficiency** |
 
-<p align="center">
-    <a href="https://www.youtube.com/watch?v=qXrvDVm_fnE"><img width="460" src="https://img.youtube.com/vi/qXrvDVm_fnE/0.jpg"></a>
-</p>
+For a more granular view of our progress and ongoing tasks, check out our [project board](https://github.com/orgs/reactivebayes/projects/2/views/4) or join our 4-weekly [public meetings](https://dynalist.io/d/F4aA-Z2c8X-M1iWTn9hY_ndN).
 
-# Roadmap
+# External Contributors
 
-Our high-level project roadmap outlines the key milestones and focus areas for the upcoming years:
+RxInfer has benefited from the contributions and development efforts of external collaborators and organizations. We're grateful for their involvement in advancing the project.
+
+## Active Inference Institute
+
+Members of the [Active Inference Institute](https://www.activeinference.org/) have been working on improving the visualization capabilities of RxInfer/GraphPPL. Their efforts focus on developing better model visualization capabilities, creating various summary/subgraph visualization modalities, implementing different graph layout algorithms, and improving the ability to inspect and understand models.
 
-| Q1/Q2 2024          | Q3/Q4 2024                | 2025                | 2026                                      |
-|---------------------|---------------------------|--------------------|-------------------------------------------|
-| 🧩 **Nested models with [GraphPPL.jl](https://github.com/reactivebayes/GraphPPL.jl)** ✅    | 🛡️ **Robustness (NaN, Inf free)** | 🌐 **Stochastic Processes** | 🔄 **Automated inference with non-exponential family** 
-| 🔄 **Automated inference with [ExponentialFamily.jl](https://github.com/reactivebayes/ExponentialFamily.jl)**                 | 🧠 **Memory-efficiency** | 🚀 **Resource-adaptive inference** | 📊 **Inference over graph structure**
+For more details on their ongoing work, see the [RxInfer development project board](https://coda.io/d/RxInfer-2024-Active-Inference-Institute_ddtS-XZ4BJb/Developing-RxInfer-jl_sufeCeIa#_lutfq_7F).
 
-For a more granular view of our progress and ongoing tasks, check out our [project board](https://github.com/orgs/reactivebayes/projects/2/views/4) or join our 
-4-weekly [public meetings](https://dynalist.io/d/F4aA-Z2c8X-M1iWTn9hY_ndN).
+## Educational Content
+
+Educational content and tutorials related to RxInfer are being developed and can be found on [Learnable Loop](https://learnableloop.com/#category=RxInfer). These resources cover a range of topics including visualizing Forney Factor Graphs, sales forecasting with time-varying autoregressive models, hidden Markov models with control, and various applications of Active Inference across different domains.
 
 # Contributing
 
 We welcome contributions from the community. If you are interested in contributing to the development of `RxInfer.jl`, please check out our [contributing guide](https://reactivebayes.github.io/RxInfer.jl/stable/contributing/guide), the [contributing guidelines](https://reactivebayes.github.io/RxInfer.jl/stable/contributing/guidelines), or look at the [issues linked with the `good first issue` label](https://github.com/ReactiveBayes/RxInfer.jl/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22) to get started.
 
-# Ecosystem
+# Where to go next?
 
-The `RxInfer` framework consists of three *core* packages developed by reactivebayes:
+There are a set of [examples](https://reactivebayes.github.io/RxInfer.jl/stable/examples/overview/) available in `RxInfer` repository that demonstrate the more advanced features of the package. Alternatively, you can head to the [documentation](https://reactivebayes.github.io/RxInfer.jl/stable/) that provides more detailed information of how to use `RxInfer` to specify more complex probabilistic models.
+
+## Ecosystem
+
+The `RxInfer` framework consists of three *core* packages developed by [ReactiveBayes](https://github.com/reactivebayes/):
 
 - [`ReactiveMP.jl`](https://github.com/reactivebayes/ReactiveMP.jl) - the underlying message passing-based inference engine
 - [`GraphPPL.jl`](https://github.com/reactivebayes/GraphPPL.jl) - model and constraints specification package
 - [`Rocket.jl`](https://github.com/reactivebayes/Rocket.jl) - reactive extensions package for Julia 
 
+## JuliaCon 2023 presentation
+
+Additionally, checkout our [video from JuliaCon 2023](https://www.youtube.com/watch?v=qXrvDVm_fnE) for a high-level overview of the package
+
+<p align="center">
+    <a href="https://www.youtube.com/watch?v=qXrvDVm_fnE"><img style="width: 100%" src="https://img.youtube.com/vi/qXrvDVm_fnE/0.jpg"></a>
+</p>
+
+## Our presentation at the Julia User Group Munich meetup
+
+Also check out the recorded presentation at the Julia User Group Munich meetup for a more detailed overview of the package
+
+<p align="center">
+    <a href="https://www.youtube.com/watch?v=KuluqEzFtm8"><img style="width: 100%" src="https://img.youtube.com/vi/KuluqEzFtm8/0.jpg"></a>
+</p>
+
 # License
 
 [MIT License](LICENSE) Copyright (c) 2021-2024 BIASlab, 2024-present ReactiveBayes

codemeta.json

@@ -9,12 +9,12 @@
     "downloadUrl": "https://github.com/reactivebayes/RxInfer.jl/releases",
     "issueTracker": "https://github.com/reactivebayes/RxInfer.jl/issues",
     "name": "RxInfer.jl",
-    "version": "3.4.0",
+    "version": "3.7.1",
     "description": "Julia package for automated, scalable and efficient Bayesian inference on factor graphs with reactive message passing. ",
     "applicationCategory": "Statistics",
     "developmentStatus": "active",
     "readme": "https://reactivebayes.github.io/RxInfer.jl/stable/",
-    "softwareVersion": "3.4.0",
+    "softwareVersion": "3.7.1",
     "keywords": [
         "Bayesian inference",
         "message passing",

docs/Project.toml

@@ -1,9 +1,11 @@
 [deps]
 BayesBase = "b4ee3484-f114-42fe-b91c-797d54a0c67e"
+BenchmarkTools = "6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf"
 Cairo = "159f3aea-2a34-519c-b102-8c37f9878175"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 ExponentialFamily = "62312e5e-252a-4322-ace9-a5f4bf9b357b"
+ExponentialFamilyProjection = "17f509fa-9a96-44ba-99b2-1c5f01f0931b"
 GraphPPL = "b3f8163a-e979-4e85-b43e-1f63d8c8b42c"
 GraphPlot = "a2cc645c-3eea-5389-862e-a155d0052231"
 Optimisers = "3bd65402-5787-11e9-1adc-39752487f4e2"
@@ -12,6 +14,7 @@ ReactiveMP = "a194aa59-28ba-4574-a09c-4a745416d6e3"
 Rocket = "df971d30-c9d6-4b37-b8ff-e965b2cb3a40"
 RxInfer = "86711068-29c9-4ff7-b620-ae75d7495b3d"
 StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
+StatsPlots = "f3b207a7-027a-5e70-b257-86293d7955fd"
 
 [compat]
 Documenter = "1.0.0"

docs/make.jl

@@ -69,7 +69,10 @@ foreach(vcat(ExamplesOverviewPath, ExamplesCategoriesOverviewPaths)) do path
         @warn "`$(path)` does not exist. Generating an empty overview. Use the `make examples` command to generate the overview and all examples."
         mkpath(dirname(path))
         open(path, "w") do f
-            write(f, "The overview is missing. Use the `make examples` command to generate the overview and all examples.")
+            write(f, """
+            $(isequal(path, ExamplesOverviewPath) ? "# [Examples overview](@id examples-overview)" : "")
+            The overview is missing. Use the `make examples` command to generate the overview and all examples.
+            """)
         end
     end
 end
@@ -108,7 +111,9 @@ makedocs(;
                 "Streamline inference" => "manuals/inference/streamlined.md",
                 "Initialization"   => "manuals/inference/initialization.md",
                 "Auto-updates"   => "manuals/inference/autoupdates.md",
-                "Deterministic nodes" => "manuals/inference/delta-node.md"
+                "Deterministic nodes" => "manuals/inference/delta-node.md",
+                "Non-conjugate inference" => "manuals/inference/nonconjugate.md",
+                "Undefined message update rules" => "manuals/inference/undefinedrules.md"
             ],
             "Inference customization"   => [
                 "Defining a custom node and rules" => "manuals/customization/custom-node.md",

docs/src/manuals/comparison.md

@@ -49,19 +49,19 @@ Nowadays there's plenty of probabilistic programming languages and packages avai
 using RxInfer #hide
 
 @model function inner_inner(τ, y, x)
-    y ~ Normal(τ[1], τ[2] + x)
+    y ~ Normal(mean = τ[1], var = τ[2] + x)
 end
 
 @model function inner(θ, α)
-    β ~ Normal(0, 1)
-    α ~ Gamma(β, 1)
+    β ~ Normal(mean = 0.0, var = 1.0)
+    α ~ Gamma(shape = β, rate = 1.0)
     α ~ inner_inner(τ = θ, x = 3)
 end
 
 @model function outer()
     local w
     for i = 1:5
-        w[i] ~ inner(θ = Gamma(1, 1))
+        w[i] ~ inner(θ = Gamma(shape = 1.0, rate = 1.0))
     end
     y ~ inner(θ = w[2:3])
 end

docs/src/manuals/constraints-specification.md

@@ -177,19 +177,19 @@ Read more about the `@constraints` macro in the [official documentation](https:/
 
 ```@example manual_constraints
 @model function inner_inner(τ, y)
-    y ~ Normal(τ[1], τ[2])
+    y ~ Normal(mean = τ[1], var = τ[2])
 end
 
 @model function inner(θ, α)
-    β ~ Normal(0, 1)
-    α ~ Gamma(β, 1)
+    β ~ Normal(mean = 0.0, var = 1.0)
+    α ~ Gamma(shape = β, rate = 1.0)
     α ~ inner_inner(τ = θ)
 end
 
 @model function outer()
     local w
     for i = 1:5
-        w[i] ~ inner(θ = Gamma(1, 1))
+        w[i] ~ inner(θ = Gamma(shape = 1.0, rate = 1.0))
     end
     y ~ inner(θ = w[2:3])
 end
@@ -250,6 +250,28 @@ end
 ```
 More information can be found in the [GraphPPL documentation](https://reactivebayes.github.io/GraphPPL.jl/stable/plugins/constraint_specification/#Default-constraints).
 
+## Constraints on the data
+
+By default, `RxInfer` assumes that, since the data comes into the model as observed, the posterior marginal distribution of the data is independent from other marginals and is a Dirac-delta distribution. However, this assumption breaks when we pass missing data into our model. When the data is missing, we might have a joint dependency between the data and latent variables, as the missing data essentially behaves as a latent variable. In such cases, we can wrap the data in a `UnfactorizedData`. This will notify the inference engine that the data should not be factorized out and we can specify a custom factorization constraint on these variables using the `@constraints` macro. 
+
+```@docs
+UnfactorizedData
+```
+
+```@example constraints-specification
+unfactorized_example_constraints = @constraints begin
+    q(y[1:1000], μ, τ) = q(y[1:1000])q(μ)q(τ)
+    q(y[1001:1100], μ, τ) = q(y[1001:1100], μ)q(τ)
+end
+result = infer(
+    model       = iid_normal(),
+    data        = (y = UnfactorizedData(vcat(rand(NormalMeanPrecision(3.1415, 2.7182), 1000), [missing for _ in 1:100])),),
+    constraints = unfactorized_example_constraints, 
+    initialization = init,
+    iterations = 25
+)
+```
+
 ## Prespecified constraints
 `GraphPPL` exports some [prespecified constraints](https://reactivebayes.github.io/GraphPPL.jl/stable/plugins/constraint_specification/#Prespecified-constraints) that can be used in the `@constraints` macro, but these constraints can also be passed as top-level constraints in the `infer` function. For example, to specify a mean-field assumption on all variables in the model, we can use the `MeanField` constraint:
 

docs/src/manuals/customization/custom-node.md

@@ -2,6 +2,8 @@
 
 Welcome to the `RxInfer` documentation on creating custom factor graph nodes. In `RxInfer`, factor nodes represent functional relationships between variables, also known as factors. Together, these factors define your probabilistic model. Quite often these factors represent distributions, denoting how a certain parameter affects another. However, other factors are also possible, such as ones specifying linear or non-linear relationships. `RxInfer` already supports a lot of factor nodes, however, depending on the problem that you are trying to solve, you may need to create a custom node that better fits the specific requirements of your model. This tutorial will guide you through the process of defining a custom node in `RxInfer`, step by step. By the end of this tutorial, you will be able to create your own custom node and integrate it into your model.
 
+In addition, read another section on a different way of running inference with custom stochastic nodes without explicit rule specification [here](@ref inference-undefinedrules).
+
 ---
 
 To create a custom node in `RxInfer`, 4 steps are required:

docs/src/manuals/getting-started.md

@@ -305,4 +305,4 @@ result.posteriors[:θ]
 
 ## Where to go next?
 
-There are a set of [examples](@ref examples-overview) available in `RxInfer` repository that demonstrate the more advanced features of the package for various problems. Alternatively, you can head to the [Model specification](@ref user-guide-model-specification) which provides more detailed information of how to use `RxInfer` to specify probabilistic models. [Inference execution](@ref user-guide-inference-execution) section provides a documentation about `RxInfer` API for running reactive Bayesian inference. Also read the [Comparison](@ref comparison) to compare `RxInfer` with other probabilistic programming libraries.
+There are a set of [examples](@ref examples-overview) available in `RxInfer` repository that demonstrate the more advanced features of the package for various problems. Alternatively, you can head to the [Model specification](@ref user-guide-model-specification) which provides more detailed information of how to use `RxInfer` to specify probabilistic models. [Inference execution](@ref user-guide-inference-execution) section provides a documentation about `RxInfer` API for running reactive Bayesian inference. Also read the [Comparison](@ref comparison) to compare `RxInfer` with other probabilistic programming libraries. For advances use cases refer to the [Non-conjugate inference](@ref inference-nonconjugate) tutorial and inference [without defining the message update rules explicitly](@ref inference-undefinedrules).