gfit-related terms

a collection of experiments compatible with a model.

a variable representing observed values of an experiment or simulation by a model. Dependent variables may contain statistical weights.

a scalar, one number, part of a variable. Variables contain zero (rarely), one, or many elements.

an input variable representing an intrinsic property of a system, which can be estimated by regression analysis. Note that designation estimatable depends on the goal of any given experiment.

a set of variables representing a group of related observations and conditions. Input variables contain sufficient information for one simulation (one model call). Variables included in the experiment should be defined in model description. The designer of the model has some flexibility in dividing data between experiments.

an input variable representing a precisely known condition of an experiment. Note that designation independent depends on the goal of any given experiment.

an integer valued scalar input variable that may be used for controlling dimensions of other variables. The value of index variable cannot be estimated.

a variable passed to a model and used for simulation. Its elements originate from experiment conditions, parameters, or both.

See Also variable, independent variable, estimatable variable, index variable.

a computer program that uses input variables to simulate one or many aspects of system's behavior, expressed as one or more dependent variables. gfit does not stipulate the type of an algorithm the model can use for simulation. The properties of the model including its input and output variables, are listed in model description.

meta information attached to the model that allows gfit to use the model for simulations. Model description includes model name, version, general human-readable comments about the purpose of the model and its algorithm, and information about the model's input and output variables. For each variable the model description specifies a name, type, a range of acceptable values, and dimensions. Model description defines every dimension of a variable. In the simplest case, dimension is fixed to a certain value – unity is the common default value. Variables may change their size depending on experimental data and user actions. Dimension of a variable usually changes in agreement with dimensions of other variables. Model description defines linear relationships between different dimensions and between a dimension and an index variable.

an object that controlls content of one or many input variables of a model. Element can be accessed by a user, or by an optimization engine. Each parameter is connected to one or more elements in different input variables and different experiments. Changing one parameter may affect simulations of one or many experiments. Connections between parameters and variable elements are defined through GUI based on the knowledge and assumptions about the experimental system.

one or many numbers describing an aspect of an experimental system, as defined in the model description . Variables store experiment information or a simulation result. A variable may contain a single element (scalar variable – zero dimensions), a vector of elements (one dimension), a matrix (two dimensions), or a multidimensional array of numbers.

Depending on model description, each dimension of the variable can be fixed, vary as a function of a different dimension of same or different variable, vary as a function of an index variable, or vary freely within bounds.