Skip to content

Tutorial: Interactive Modeling with Dyno Lab

This tutorial walks through using Dyno Lab (jupyterlab_dyno), the interactive JupyterLab extension for Dyno, to explore, calibrate, and simulate a Real Business Cycle (RBC) model in real time.

Prerequisites & Setup

Ensure you have jupyterlab_dyno installed in your environment:

pixi add --channel https://repo.prefix.dev/econforge jupyterlab_dyno

Launch JupyterLab:

pixi run -e dev jupyter lab

JupyterLab will start and open your browser to the workspace.

Step 1: Open an RBC Model

In the JupyterLab file browser on the left, create or open a model file named rbc.dyno:

# 1. Calibration
alpha <- 0.33
beta  <- 0.99
delta <- 0.025
rho   <- 0.95
eta   <- 1.0

# 2. Steady State
k[~] <- ((1/beta - (1-delta)) / alpha)**(1 / (alpha - 1))
y[~] <- k[~]^alpha
i[~] <- delta * k[~]
c[~] <- y[~] - i[~]
z[~] <- 0.0

# 3. Dynamic Equations
z[t] = rho * z[t-1] + e_z[t]
y[t] = exp(z[t]) * k[t-1]^alpha
k[t] = (1-delta)*k[t-1] + i[t]
y[t] = c[t] + i[t]
beta * (c[t+1]/c[t])^(-1) * (alpha * y[t+1]/k[t] + 1 - delta) = 1

# 4. Stochastic Shocks
e_z[t] <- N(0.01)

Upon opening rbc.dyno, Dyno Lab configures a split-pane layout:

  • Left Pane — Code Editor: Code editor with syntax highlighting for .dyno and .mod files.
  • Right Pane — Dyno Report: Interactive report viewer connected to a background Python kernel, displaying steady-state values, eigenvalues, and IRFs.

Step 2: Live Model Exploration

Dyno Lab updates the report automatically as you modify model specifications.

Experiment 1: Modifying Shock Persistence

  1. In the editor, change the persistence parameter rho from 0.95 to 0.70: 
    rho <- 0.70
  2. After a brief pause, the background solver recomputes the decision rule. The impulse responses on the right update to reflect the faster decay back to steady state.

Experiment 2: Modifying Capital Depreciation

  1. Change delta from 0.025 to 0.050: 
    delta <- 0.050
  2. The report updates:
  3. Steady-state capital \(k^*\) falls to reflect higher depreciation.
  4. Investment \(i^*\) adjusts to maintain the stationary capital stock.
  5. The report view maintains its scroll position during re-computation.

Step 3: Using the Dyno Options Sidebar

Solution parameters can be adjusted via the sidebar interface:

  1. Click the Dyno Options tab in the JupyterLab sidebar (or the Options button in the report toolbar).
  2. Change Simulation Type between Deviation, Log-Deviation, or Level.
  3. Adjust the Horizon (e.g. from 40 to 80 periods) to examine longer-term dynamics.
  4. Enable Steady state only when calibrating parameters to inspect steady-state values without running perturbation solvers.

Step 4: Diagnostic Highlighting

Dyno Lab flags syntax and mathematical errors directly in the editor:

  1. In the code editor, introduce an undefined variable (e.g. c[t] = y[t] - i[t] + unknown_var[t]).
  2. The background solver identifies the undefined symbol and highlights the offending line in the editor gutter.
  3. Correcting the line clears the diagnostic marker and restores the report.

Next Steps

  • For an overview of Dyno Lab features and installation, see the Dyno Lab Guide.