Glossary

Model structure

Array: A variable expanded across named dimensions and elements. Also called a subscripted variable.

Auxiliary: A calculated variable that transforms inputs without accumulating them. Also called a converter.

Connector: A dependency arrow showing that one variable is used by another. Also called a link or information arrow.

Flow: A rate that changes a stock. Also called a rate.

Module: A reusable or grouped model structure with defined connections. Also called a submodel or component.

Stock: An accumulation changed over time by flows. Also called a level, state variable, or reservoir.

Equations and behavior

Equation: The expression used to calculate a variable.

Feedback loop: A closed causal chain. A reinforcing loop amplifies change; a balancing loop opposes change. Aliases include positive loop and negative loop.

Graphical function: A point-defined input-output relationship with interpolation. Also called a lookup, table function, or graph function.

Initial value: A stock's value at simulation start.

Time step (DT): The interval between simulation calculations.

Run specs: Start time, end time, time step, and related simulation settings. Also called simulation control parameters.

Analysis

Behavior reproduction: Calibration against reference behavior or imported data.

Calibration: Searching for parameter values that reproduce target data or behavior. Also called fitting or parameter estimation.

Dimensional consistency: Compatibility of units across an equation.

Evaluation budget: Maximum number of model evaluations used by an optimization.

Halton sequence: A low-discrepancy sequence used to cover a parameter space.

Latin Hypercube: A stratified sampling method that spreads samples across each parameter range. Abbreviation: LHS.

MAPE: Mean absolute percentage error; lower values generally indicate closer proportional fit.

Monte Carlo: Repeated simulation using sampled parameter values. Abbreviation: MC.

Markov Chain Monte Carlo: Sampling where proposed values depend on the current sample. Abbreviation: MCMC.

Optimization: Searching parameter ranges to maximize or minimize a payoff.

Parameter: A model input varied by sensitivity analysis or optimization.

Payoff: The score an optimization maximizes or minimizes. Also called objective, objective function, cost, fitness, or loss.

Response: The model output measured by a sensitivity analysis.

R-squared (R²): A measure of how much variation in reference data is represented by fitted results; interpret it with error measures and visual checks.

Sensitivity analysis: Repeated simulation that tests how uncertain inputs affect outputs. Also called uncertainty analysis or parameter sweep.

Validation: Structured checks for model quality and available evidence; it is not proof of correctness.

Files and security

Canvas export: A visual image or PDF of the diagram, not an editable model.

Encrypted project: A .yano project whose model, simulation runs, and imported media are protected by a password and optional Touch ID unlock.

MDL: A Vensim text model format.

XMILE: An interchange format for system dynamics models; common extensions are .xmile, .xml, and .stmx.

Yano project: The native .yano package containing model structure, saved simulation runs, and imported media. JSON export contains model structure only.