Menu Data-Initial Values

Menu Data HyperManual Lsd Windows Main Browser Lsd FAQ's

Menu "Data" - Init. Values

This entry allows to determine the initial values to be used during the initial time step of the simulation. These values are inserted for a single Object type, and for all its instances. To set the initial number of instance of Objects see Object Number in menu Data(or click on the label of the Object in the above window in the upper left corner).
The data required to be initalized are the Parameters and the Variables used with lagged values (to be used in the early time steps of the simulation), therefore Variables used with no lags do not appear in the initial data window. Users can set only the initial values for Parameters and lagged Variables concerning the Object type shown in the Browser when this menu entry is invoked (see the Browser help page to move through the model). Variables that are used with one, two, or more lagged values will appear once for each lag, because it is necessary to have initial values for each of the lags.

The window is formatted as a spreadsheet: each line refers to one type of initial value (Parameter or one Variable lag) and each column refers to one instance of the Object containing that element. For example, in the window above it is highlighted the initial value for Parameter ms contained in an Object of type Seller. The exact Object instance to which the Parameters and Variables refer to is indicated with an automatic code (2-2 in the example) referred to on the columns' headers. In the following it is indicated how to identify the Object instance to which each cell refers to and the different ways to insert data.

Lsd Automatic Coding

In Lsd, Objects are assigned a unique label and can be replicated in as many instances as desired. Therefore, there will be many replica of Objects with the same label (and containing instances of Parameters and Variables with identical labels). This creates the problem of referring to one particular instance when this is necessary, typically for inserting initial values and for analysing the results. Since the any instance is virtually equal to every other instance of the same type, the only way to differentiate them is their relative position in the model.
A model in Lsd is structured as a hierarchy: on top is placed Object Root, which usually is left empty. Root can contain sets of descending Objects, which in turn may contain Objects, and so on. The automatic code assigns an integer number to an Object for each "ancestor". For example, consider a model with Objects Market, containing Objects Supply and Demand, the first of which contains Producer. The code identifying an instance of Object Producer will contain three groups of numbers: the first refers to Market, the second to Supply, and the last to Producer.

Inserting data manually

Users can insert values manually by moving through the cells (by default the system moves horizontally when the key Return is pressed). The window shown above is limited to show 100 instances. If the model contains more instances for some type of Objects, the it is necessary to use the Set All button, located in the beginning of the lines. Users can also initialize data using the interface showing one single Object instance (see Data Browse in menu Data for doing this in the general case).

SET ALL : Inserting data automatically

set all window

A simulation model needs initial values to start a simulation run. Initial values must be assigned to Parameters and lagged Variables, that is Variables used with a lag. In this latter case the user must assign an initial value to be uses in the first steps of the simulation. A given element to initialize (Parameter or lagged Variable) must be given one numerical value for each copy contained in the model, that is, for each copy of the Object containig the element.

Users can insert initial value one-by-one with a suitable interface (see Data Browse in menu Data). However, in most cases there are many copies of the element to initialize, and the initialization follows a specific function. For example, the user may want all the initial values in the 1000 copies of a Parameter set to the same value. In this cases, Lsd offers the "Set All" functionality, which allows to set all elements specifying an initialization function.

The user needs to specify the following options:

Initialization function to use for the initialization. For example, identical values for all the elements (option Equal to)
The data values used in the initialization funtion. Depending on the function chosen these values can vary. In the example of Equal to the value is the number to assign to all the elements.
The frequency of the initialization. Considering the series of all elements to initialize (from the first Object to the last), the user can choose to apply the initialization function to every element or to "jump" regularly some elements.
The extention of the initialization. The user can choose to apply the initialization funtion to all elements in the model, or only to a sub-group, specifying the first and last element part of the group.

Below you find the detailed description of the available options.

Initialization function

- Equal To: the inserted Value is assigned to every instance the element in the model;

- Range: the data values used concern the Minimum and Maximum. The system counts the number of elements to initialize, then assigns increasing values so that the very first element is set to Minimum, the last one to Maximum, and the intermediate values to equally increasing values. For example, if the model contains 100 elements, Miminum=1 and Maximum=100, then the sequence of elements will be set to: 1 2 3 4 ...

- Increasing: the values of the elements are set to increasing values beginning from Start and adding Step to each element. For example, if Start=12 and Step=0.5, then the elements will be initialized with the following values: 12 12.5 13 13.5 etc.

- Increasing (Groups): the same as Increasing, but the sequence of initial values is re-set to the Start value when a group of continguous objects is finfished and another group begins. For example, consider a model made of an Object Market containing a set of objects Firm. Suppose the model contains many Market's, and therefore many groups of objects Firm's. Initializing a parameter in Firm with Increasing (Group), setting Start=1 and Step=1, the initial values for the Firm's in the first Market will be set to 1 2 3 etc. When the initialization function reaches the first object Firm contained in the second Market, the sequence begins with 1 2 3 again.

- Random (Uniform): a random value is assigned to element to initialize. The random numbers are drawn from a uniform distribution whose minimum value is the first value (Min) and the maximum is the second value (Max).

- Random Integer (Uniform): a random integer value is assigned to element t initialize. The random integer values are drawn from a uniform distribution whose minimum value is the first value (Min) and the maximum is the second value (Max).

- Random (Normal): a random integer value is assigned to element t initialize. The random values are drawn from a normal distribution whose mean is the first value (Mean) and the standard deviation is the second (Std Dev)

- File: insert data from a file. The file must be a text file formed by a single column containing an initial label. The label is ignored and the file must contain as many lines as instances of the Objects to be initialize.

Random generator settings

When using one of the random initialization functions the user can specify a specific seed, in order to be able to replicate exactly the same random initialization if needed. Such setting does not affect the random values during a simulation run, unless the initialization is performed during a run.

Numerical data for initialization

These two entries allow users to insert the parameters for the initialization function chosen, The labels of the entries vary depending on the function chosen.

Frequency

The initialization functions are applied to the sequence of elements as they appear in the model. By default the function is used for every element encountered. The user can also choose to apply the function skipping a specified number of objects. For example, if one sets to apply the function every 2 objects, then the function will set the first element, skip the second, set the third an so on. In case of the box for "Fill in" is checked, the "Increasing" initialization function will fill also the intermediate elements with the latest values. For example, suppose to initialize an element with the "Increasing" initialization function, setting Start = 1 and Step=1. If you have the frequency every 2 objects and Fill in not checked, then you will get the series 1 0 2 0 3 0 4 0 etc. (supposing the element was set to 0 for all objects). If, with the same function and setting, you put Fill in checked, then you will get the series 1 1 2 2 3 3 4 4 etc.

Extension of initialization

The "Apply to every object" option applies the initialization function to the whole sets of elements in the model, that is over all the objects containing the element. Alternatively, the user can choose to apply the initialization only to a restricted set of object. The group of objects on which to apply the initialization must be indicated with the ordinal number of the first and last object on which to apply the initialization. For example, suppose the model contains 10 objects Market each containing 20 objects Firm. In order to initialize a Parameter in Firm, limiting to the copies of Firm in the second Market, the user needs to insert From=21 and To=40. Clicking on the entries with the right button of the mouse the user can insert the ordinal number of the objects by indicating the ordinal numbers of parent's. See here for details on this last option.

Update description

This option checked on causes the system to add a verbal description of the initialization to the description of the element. This description will appear in the description of the element in the browser and in the model report.