Add regression test data (#7)

Author: timmens

.envs/testenv.yml

@@ -30,3 +30,4 @@ dependencies:
   - pydata-sphinx-theme>=0.3.0
   - pip:
       - -e ../
+      - git+https://github.com/opensourceeconomics/lcm.git@simulation

README.md

@@ -1,10 +1,37 @@
-# Developer Repository: Life Cycle Models
+# LCM Development Repository
 
 [![image](https://img.shields.io/badge/code%20style-black-000000.svg)](https://github.com/psf/black)
 [![pre-commit.ci status](https://results.pre-commit.ci/badge/github/OpenSourceEconomics/lcm-dev/main.svg)](https://results.pre-commit.ci/latest/github/OpenSourceEconomics/lcm-dev/main)
 [![image](https://codecov.io/gh/OpenSourceEconomics/lcm-dev/branch/main/graph/badge.svg)](https://codecov.io/gh/OpenSourceEconomics/lcm-dev)
 
 This repository aims to facilitate the development of the package
-[lcm](https://github.com/OpenSourceEconomics/lcm). We do so by trying to increase the
-reproducibility of insights that led to changes in `lcm` and by providing developer
-documentation, designed to explain the inner workings of the `lcm` code base.
+[lcm](https://github.com/OpenSourceEconomics/lcm). In particular, here we
+
+- Generate data that is used in the testing process of `lcm`
+- Publish the developer documentation, which is designed to explain the inner workings
+  of the `lcm` code base
+- Create graphical comparisons between `lcm` and analytical solutions
+
+## Getting Started
+
+Get started by installing the conda environment. The conda environment will contain a
+local installation of `lcm`. Therefore, we require the following directory structure:
+
+```console
+parent_folder
+ |-lcm-dev
+ |-lcm
+```
+
+Then you can install the environment using
+
+```console
+cd /path/to/lcm-dev
+mamba env create
+```
+
+and generate all data and figures by running
+
+```console
+pytask
+```

codecov.yml

@@ -15,4 +15,6 @@ coverage:
         target: 10%
 ignore:
   - setup.py
-  - tests/**/*
+  - '**/__init__.py'
+  - '**/task_*.py'
+  - tests

environment.yml

@@ -32,3 +32,4 @@ dependencies:
   - pytest-xdist
   - pip:
       - -e .
+      - -e ../lcm

src/lcm_dev/task_regression_test.py

@@ -0,0 +1,53 @@
+"""Save LCM solution and simulation output for regression testing.
+
+This task needs to run again when the example model `PHELPS_DEATON` changes.
+
+"""
+import json
+
+import jax.numpy as jnp
+import pytask
+from lcm.entry_point import get_lcm_function
+from lcm.example_models import PHELPS_DEATON
+from pybaum import tree_map
+
+from lcm_dev.config import BLD
+
+
+@pytask.mark.produces(BLD.joinpath("regression_tests", "solution_and_simulation.json"))
+def task_save_lcm_output(produces):
+    model = {**PHELPS_DEATON, "n_periods": 5}
+
+    solve_model, _ = get_lcm_function(model=model, targets="solve")
+    simulate_model, _ = get_lcm_function(model=model, targets="simulate")
+
+    params = {
+        "beta": 1.0,
+        "utility": {"delta": 1.0},
+        "next_wealth": {
+            "interest_rate": 0.05,
+            "wage": 1.0,
+        },
+    }
+
+    solution = solve_model(params)
+
+    simulation = simulate_model(
+        params,
+        vf_arr_list=solution,
+        initial_states={
+            "wealth": jnp.array([1.0, 20, 40, 70]),
+        },
+    )
+
+    # Convert output to JSON serializable format
+    solution = [sol.tolist() for sol in solution]
+    simulation = tree_map(lambda x: x.tolist(), simulation)
+
+    out = {
+        "solution": solution,
+        "simulation": simulation,
+    }
+
+    with produces.open("w") as file:
+        file.write(json.dumps(out, indent=4))