index.qmd

---
title: "the odin-monty toolkit"
author: "Rich FitzJohn, Marc Baguelin, Ed Knock"
subtitle: "MRC Centre for Global Infectious Disease Analysis, Imperial College London"
date: "2025-02-13"
format:
    revealjs:
        logo: "images/logo.png"
        slide-number: c/t
        footer: "[the odin-monty toolkit](https://mrc-ide.github.io/odin-monty/)"
editor: visual
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)
r_output <- function(path, highlight = NULL) {
  if (is.null(highlight)) {
    prefix <- "```r"
  } else {
    prefix <- sprintf('```{.r code-line-numbers="%s"}', highlight)
  }
  writeLines(c(prefix, readLines(path), "```"))
}
set.seed(1) # always the same
knitr::knit_hooks$set(small_margins = function(before, options, envir) {
  if (before) {
    par(mar = c(4, 4, .1, .1))
  }
})
```

## Interactive session

```{r, echo=FALSE}
#| fig-width: 8
code <- qrcode::qr_code("https://github.com/mrc-ide/odin-monty-hpru-2025?tab=readme-ov-file#interactive-session")
plot(code)
```

[`github.com/mrc-ide/odin-monty-hpru-2025`](https://github.com/mrc-ide/odin-monty-hpru-2025)

## odin - the beginnings (2016-2019)

-   **`odin`** created to integrate ODEs (e.g. for compartmental models) in R with a domain-specific language (DSL)
-   Limited support for difference (discrete-time) equations
-   Automatic translation to C/C++/JavaScript; efficient solutions in little code
-   Used in models of malaria, HIV, ebola and other diseases
-   No support for inference

## COVID-19 response

-   **`mcstate`** and **`dust`** emerged during the COVID-19 pandemic
-   Collaborative work by the UK real-time modelling & research software engineers teams at Imperial College
-   Tailored to handle and fit complex state-space models at scale (e.g. Knock et al. 2021)

## Challenges addressed by `dust` and `mcstate`

-   **Efficient** (marginal) likelihood computation for large stochastic models via particle filters
-   Exploit parallelism (OpenMP & across chains)
-   Limited toolkit for inference for models created in `odin`

## Recent developments

- Design of a new architecture, rewiring data, model and parameters
  - rewrite from ground up, keeping best ideas and discarding some baggage
- Design of a new interface

  - new small BUGS-inspired DSL for priors
  - modular approach to inference
  - inverts the dependency stack
- Development of automatic differentiation algorithms

## Key components

::::: columns
::: {.column width="60%"}
![](images/dependencies_odin_monty.png)
:::

::: {.column width="40%"}

- `monty` models can be built from
    - `odin` and data
    - custom DSL
    - by composition
- MCMC samplers work on any `monty` model
- Concept of "packer"

:::
:::::

## Introducing `monty`

-   **`monty`**: A framework to handle statistical models (defined by their domain and density)
-   Aims to offers multiple sampling solutions
-   Modular approach so that `odin` and `monty` "collaborate" but remain independent

## Different samplers

::: {layout-ncol=2}

![HMC](images/HMC.png)

![PT](images/PT.png)

:::

## Introducing automatic differentiation {.smaller}

::::: columns
::: {.column width="60%"}
![](images/graph_one_step.png)

:::

::: {.column width="40%"}

```{r}
#| results: "asis"
r_output("models/sir_4_AD.R")
```

:::
:::::

## Introducing automatic differentiation {.smaller}

![](images/chained_graph.png)

- `odin` DSL is small
- Ladder shaped computation graph
- Regular parameter inputting (red)
- Regular state outputting (green)

## Reverse automatic differentiation {.smaller}

![](images/SSM_autodiff.png)

Goal: a large class of `odin` models and the DSL `monty` models can autodifferentiate

## The odin-`monty` vision

-   Provide a **modern modeling ecosystem**:
    1.  Powerful DSL (`odin2`+`monty`)
    2.  Streamlined parallel simulation & filtering engine (`dust2`)
    3.  User-friendly statistical integration platform (`monty`)
-   Empower researchers to build robust models quickly, accelerating scientific insights

# In action!

(switch to RStudio and Ed here)

```{r, echo=FALSE}
#| fig-width: 8
code <- qrcode::qr_code("https://github.com/mrc-ide/odin-monty-hpru-2025?tab=readme-ov-file#interactive-session")
plot(code)
```

[`github.com/mrc-ide/odin-monty-hpru-2025`](https://github.com/mrc-ide/odin-monty-hpru-2025)

# Toward the future

<!-- ## `odin2` and `dust2` (Rich) -->

<!-- -\> @Rich for most people `odin2` and `dust2` are the same than `odin` and `dust`: can you summarise how the change make the whole toolkit more efficient? the parallelisation of parameter+stochastic draw groups is one I see and better memory usage between "fixed data" and changing parameters but feel free to put what you think -->

## 2025 Roadmap {.smaller}

- Finish development of automatic differentiation
- NUTS and HMC algorithms
- GPU acceleration via CUDA graph API
- MPI-based backend for many-node
- Restore missing features from `mcstate`/`dust` and original `odin`

## Resources

-   We will provide a demonstration
-   You can find more in [our online `odin` & `monty` book](https://mrc-ide.github.io/odin-monty/)

## `odin`-`monty` Workshop {.smaller}

::::: columns
::: {.column width="60%"}
📅 **24–25 March 2025**
📍 **White City, Imperial College London**

✅ **Hybrid participation available**
✅ **Travel grants available**
✅ **Lightning talk slots!**

🔗 **Register now [here](https://docs.google.com/forms/d/e/1FAIpQLSePERaKXjMMYK0re1mh--2nNM5PCUZaRrd-SL_zSqAYY-pk5A/viewform)!**

⚠️ **Deadline: 19/02 – Don't miss out!**
:::

::: {.column width="40%"}
```{r, echo=FALSE}
#| fig-width: 8
code <- qrcode::qr_code("https://docs.google.com/forms/d/e/1FAIpQLSePERaKXjMMYK0re1mh--2nNM5PCUZaRrd-SL_zSqAYY-pk5A/viewform")
plot(code)
```
:::
:::::