Embedding Computations & Code

Session 3

Saturday, the 7th of February, 2026

Mickaël CANOUIL, Ph.D.

Independent Contractor

Session Overview

Session 3: Embedding Computations & Code

Computing Environments
- Setup R, Python, and Julia for Quarto.
- Engine configuration and code cell structure.
- Understanding multi-language workflows.

Execution Options
- Code cell options using comment + pipe syntax.
- Control code visibility, evaluation, and output.
- Global execution settings in YAML.

Cache and Freeze
- Cache for development speed vs freeze for collaboration.
- Understanding Jupyter and Knitr caching systems.
- CLI commands for cache and freeze management.

Parameters
- Creating dynamic document variations.
- Defining parameters in Python and R.
- Command-line parameter passing and YAML files.

Session Objectives

This session explores Quarto’s computational capabilities, focusing on seamless multi-language workflows, code cell options, and integration patterns for R and Python data science stacks.

Learning Objectives

By the end of this session, participants will be able to:

Execute R , Python , Julia code.
Configure code cells using “comment symbol” + “pipe” (|) syntax.
Control code visibility and execution.
Cache computations for improved performance.

Computing Environments

Setup R, Python, and/or Julia for Quarto

# Install renv if not already installed
install.packages("renv")

# Initialise project
renv::init()

# Install packages
renv::install("rmarkdown")

# Save current state
renv::snapshot()

bash

# Install uv (if not installed)
pip install uv

# Initialise project
uv init --no-package --vcs none

# Create virtual environment
uv venv

# Activate environment
source .venv/bin/activate

# Add packages
uv add jupyter papermill

julia

# Start Julia in your project directory
using Pkg

# Activate current directory as project
Pkg.activate(".")

# Add packages
Pkg.add("IJulia")

# Instantiate environment
Pkg.instantiate()

Engine Configuration

Configure engines in your document header:

knitr
jupyter
julia

yaml

---
engine: knitr
---

yaml

---
engine: jupyter
jupyter: python3
---

yaml

---
engine: julia
---

Tip

Set explicitly the engine in your document header instead of relying on auto-detection which does not work for inline code (i.e., `{language} ...`).

What is a Code Cell?

A code cell (also called a code chunk) is a section of executable code (defined by ```{language}) embedded within your Quarto document.

Code cells allow you to:

Execute programming code directly in your document.
Display both the code and its output.
Create reproducible analyses and reports.

Code Cell Structure

qmd

1```{python}
2#| label: my-plot
#| echo: true
#| fig-cap: "Sample visualization"

3import matplotlib.pyplot as plt
import numpy as np

x = np.linspace(0, 10, 100)
y = np.sin(x)

plt.plot(x, y)
plt.title("Sine Wave")
plt.show()
```

1: Computing language (e.g., python, r, julia).
2: Code cell options (e.g., label, echo, eval).
3: Code content (your actual programming code).

Why Use Code Cells?

Reproducibility

Code and results stay together.
Easy to re-run analyses.
Version control friendly.

Documentation

Combine narrative and code.
Show methodology clearly.
Create living documents.

Execution Options

Setting Execution Options Globally

yaml

execute:
  ...

Table 1: execute options

Option	Description
`eval`	Evaluate the code chunk (if `false`, just echos the code into the output).
`echo`	Include the source code in output
`output`	Include the results of executing the code in the output (`true`, `false`, or `asis` to indicate that the output is raw markdown and should not have any of Quarto’s standard enclosing markdown).
`warning`	Include warnings in the output.
`error`	Include errors in the output (note that this implies that errors executing code will not halt processing of the document).
`include`	Catch all for preventing any output (code or results) from being included (e.g., `include: false` suppresses all output from the code block).
`renderings`	Specify rendering names for the plot or table outputs of the cell, e.g., `[light, dark]`. See Renderings.

Code Cells Options

The “comment symbol” + “pipe” (|) syntax allows to pass options to code cells.

Code Visibility Control:

qmd

```{r}
1#| echo: false
2#| eval: false
3#| include: false
4#| code-fold: true
```

1: Hide code, show output.
2: Show code, don’t execute.
3: Execute code, hide everything.
4: Collapsible code blocks.

Output Control:

qmd

```{r}
1#| warning: false
2#| message: false
3#| error: true
4#| output: false
```

1: Suppress warnings
2: Suppress messages
3: Show errors for debugging
4: Hide all output

Cache and Freeze

Two Systems

Cache: Stores computation results

Speeds up development.
Works during iteration.
Engine-specific behaviour.

Freeze: Prevents re-execution

Enables collaboration.
Project-wide control.
Version control friendly.

Cache: Development Speed

Store expensive computation results to avoid re-running during development.

yaml

execute:
  cache: true

Two Types:

Jupyter Cache: Document-level (all-or-nothing).
Knitr Cache: Cell-level (granular control).

Freeze: Collaboration Control

Prevent documents from re-executing during project renders.

Results saved in _freeze directory (commit to version control).

yaml

execute:
 freeze: auto    # Re-render only when source changes
 freeze: true    # Never re-render during project render

Key Behavioural Differences

Action	Cache	Freeze
Individual document render	✅ Used	❌ Ignored
Project render	✅ Used	✅ Controls execution
Prose changes	No effect	No effect
Code changes	Invalidates cache	Triggers re-render (auto mode)

When to Use Each

Use Cache When

Developing multi format documents.
Iterating on analysis.
Working with expensive computations.
Editing prose frequently.

Use Freeze When

Working in teams.
Managing large projects.
Environment setup is complex.
Results need to be reproducible.

Essential CLI Commands

Cache Control

bash

quarto render --cache           # Force cache use
quarto render --no-cache        # Disable cache
quarto render --cache-refresh   # Force refresh

Freeze Management

bash

quarto render document.qmd      # Always executes (ignores freeze)
quarto render                   # Respects freeze settings

Golden Rules

Cache for Development.
Use when iterating on individual documents.

Freeze for Collaboration.
Use when working with teams or managing large projects.

Different Scopes.
- Cache: Session-level performance.
- Freeze: Project-level workflow management.

Version Control.
Always commit _freeze directory contents.

Parameters

What Are Parameters?

Parameters allow you to create dynamic variations of the same document:

Geographic reports → Different locations
Time-based analysis → Different periods
Scenario modelling → Different assumptions
Multi-client reports → Different datasets

Tip

Think of parameters as variables that make your documents code reusable and flexible.

Defining Parameters

python

#| tags: [parameters]
alpha = 0.1
ratio = 0.5
country = "UK"
year = 2024

Must use #| tags: [parameters] comment.
Provide sensible default values.
Parameters available in top-level environment.
Works with .qmd and .ipynb files.

yaml

---
title: "Sales Report"
params:
  region: "North"
  start_date: "2024-01-01"
  threshold: 1000
---

Access in R code:

print(params[["region"]])

Passing Parameters: Command Line

Use the -P (--execute-param) flag to override parameter values:

bash

# Single parameter
quarto render report.qmd --execute-param country:France

# Multiple parameters  
quarto render report.qmd -P year:2023 -P region:Europe

# With output specification
quarto render report.qmd -P country:Germany --to typst

Passing Parameters: YAML Files

Create reusable parameter sets with YAML files:

params.yml

country: "Spain"
year: 2023
budget: 50000

Terminal

quarto render report.qmd \
  --execute-params params.yml

Benefits

Version control, team sharing, complex configurations.

Practical Example

Setup (Jupyter/Python)

markdown

```{python}
#| tags: [parameters]
region = "All"
quarter = "Q1" 
min_sales = 1000
```

Using parameters

markdown

```{python}
if region != "All":
   df = df[df['region'] == region]

df_filtered = df[df['sales'] >= min_sales]
title = f"{region} Sales - {quarter}"
```

Command line usage

bash

quarto render sales.qmd \
  --execute-param region:North \
  --execute-param quarter:Q2

Results

Customised report for North region, Q2 data.

Best Practices

✅ Do

Provide meaningful defaults.
Use descriptive names.
Document parameter types.
Validate values in code.
Use YAML for complex configs.

❌ Avoid

Generic names (x, val).
Parameters without defaults.
Hardcoded values.
Overly complex structures.

Workflow Integration

Automated reporting

bash

# Daily reports
quarto render report.qmd -P date:"$(date +%Y-%m-%d)"

# Multiple outputs
for region in North South East West; do
  quarto render regional.qmd -P region:$region
done

Perfect for

CI/CD pipelines, batch processing, scheduled reports.

Hands-On Exercise: Adding Computational Power to Your Portfolio

Exercise Overview

Objective: Transform your Session 2 portfolio document into a computational showcase by adding live code execution, demonstrating the power of embedding computations in Quarto documents using modern Python tools.

Example Code: Exercises

bash

tar -xzf "03-exercises.tar.gz" -C "03-embedding-computations-code"

Part 1: Environment Setup

Choose your language and update your YAML header:

yaml

---
title: "My Computational Portfolio"
author: "Your Name"
engine: jupyter
jupyter: python3
execute:
  echo: true
  warning: false
---

Install packages:

bash

uv add polars plotnine # pip install polars plotnine

yaml

---
title: "My Computational Portfolio"
author: "Your Name"
engine: knitr
execute:
  echo: true
  warning: false
---

Install packages:

install.packages(c("dplyr", "ggplot2"))

yaml

---
title: "My Computational Portfolio"
author: "Your Name"
engine: julia
execute:
  echo: true
  warning: false
---

Install packages:

julia

using Pkg
Pkg.add(["DataFrames", "Plots", "StatsPlots"])

Part 2: Replace Static with Dynamic

Replace your skills table from Session 2:

Create a Polars DataFrame with your skills data.
Include columns: skill name, confidence level, hours practiced.
Use print() to display the data nicely.
Calculate total practice hours with pl.sum().
Show the result using inline code `{python} your_calculation`.

Create a data.frame with your skills data.
Include columns: skill name, confidence level, hours practiced.
Use knitr::kable() to display the table nicely.
Calculate total practice hours with sum().
Show the result using inline code `{r} your_calculation`.

Create a DataFrame with your skills data.
Include columns: skill name, confidence level, hours practiced.
Display the DataFrame directly.
Calculate total practice hours with sum().
Show the result using inline code `{julia} your_calculation`.

Part 3: Create Your First Visualization

Make a chart of your skills:

Use plotnine to create a bar chart.
Try geom_col() or geom_bar().
Add proper labels with labs().
Use theme_minimal() for clean styling.
Add a figure caption with #| fig-cap:.

Use ggplot2 to create a bar chart.
Try geom_col() or geom_bar().
Add proper labels with labs().
Use theme_minimal() for clean styling.
Add a figure caption with #| fig-cap:.

Use Plots.jl to create a bar chart.
Try bar() function.
Add proper labels with xlabel!() and ylabel!().
Add a title with title!().
Add a figure caption with #| fig-cap:.

Part 4: Experiment with Code Visibility

Try different code cell options:

Create a cell with #| eval: false (shows code, doesn’t run).
Make a collapsible section with #| code-fold: true.
Hide code with #| echo: false but show output.
Use #| include: false for data setup.

Create a cell with #| eval: false (shows code, doesn’t run).
Make a collapsible section with #| code-fold: true.
Hide code with #| echo: false but show output.
Use #| include: false for library loading.

Create a cell with #| eval: false (shows code, doesn’t run).
Make a collapsible section with #| code-fold: true.
Hide code with #| echo: false but show output.
Use #| include: false for package imports.

Success Criteria

✅ You’ve successfully completed the exercise if you can:

Transform static content into dynamic, computed content using Python , R , or Julia .
Create and display data visualisations.
Use inline code to insert computed values into prose.
Experiment with different code cell visibility options.

Reuse

CC BY-NC-SA 4.0

Session Overview

Session 3: Embedding Computations & Code

Session Objectives

Learning Objectives

Computing Environments

Setup R, Python, and/or Julia for Quarto

Engine Configuration

What is a Code Cell?

Code Cell Structure

Why Use Code Cells?

Reproducibility

Documentation

Execution Options

Setting Execution Options Globally

Code Cells Options

Cache and Freeze

Two Systems

Cache: Stores computation results

Freeze: Prevents re-execution

Cache: Development Speed

Freeze: Collaboration Control

Key Behavioural Differences

When to Use Each

Use Cache When

Use Freeze When

Essential CLI Commands

Cache Control

Freeze Management

Golden Rules

Parameters

What Are Parameters?

Defining Parameters

Passing Parameters: Command Line

Passing Parameters: YAML Files

Practical Example

Setup (Jupyter/Python)

Using parameters

Command line usage

Best Practices

✅ Do

❌ Avoid

Workflow Integration

Automated reporting

Hands-On Exercise: Adding Computational Power to Your Portfolio

Exercise Overview

Part 1: Environment Setup

Part 2: Replace Static with Dynamic

Part 3: Create Your First Visualization

Part 4: Experiment with Code Visibility

Success Criteria

Reuse

Copyright