---
title: "Simulating relational events"
author: "Francisco Richter"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Simulating relational events}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r}
library(amorem)
```

## Simulating actor covariates

We can start by defining sender and receiver sets and generating
exogenous covariates. Static covariates return wide matrices, while
time-varying covariates are emitted in tidy format.

```{r}
senders <- paste0("s", 1:3)
receivers <- paste0("r", 1:2)

covs <- simulate_actor_covariates(
  senders = senders,
  receivers = receivers,
  covariate_names = c("activity", "popularity"),
  time_points = 0:5,
  rho = 0.7,
  sd = 0.2,
  seed = 2024
)
head(covs$sender_covariates)
```

## Simulating event sequences

Using the covariates, we simulate 20 relational events. Sender effects
control how activity shapes outgoing intensity, and receiver effects do
the same for popularity.

```{r}
static_sender <- reshape(
  covs$sender_covariates,
  direction = "wide",
  idvar = "actor",
  timevar = "covariate"
)

events <- simulate_relational_events(
  n_events = 20,
  senders = senders,
  receivers = receivers,
  baseline_rate = 2,
  sender_covariates = static_sender[, c("value.activity", "value.popularity")],
  sender_effects = c(0.8, -0.2),
  allow_loops = FALSE
)
head(events)
```