Conceptual structure of an ndata file

The file consists of a header and a set of trials.

header

The header contains the following information:

• class - a string describing the nature of the data file.
• constant parameters - names, values, and types (i-integer, f-float, or c-char) for parameters that remain constant for the data file.
• variable parameters - names and types (i, f, or c) of parameters whose values changes across trials in the data file.
• event code tables - each table contains a list of code numbers and their meaning.
• number of trials - the number of data trials in the file.

trial

A trial consists of a trial header and a list of records.

trial header

• trial code - usage varies.
• time reference - A global time reference, if one is available, for the start of the trial.
• variable parameters - names and values for this trial. Types must match those in the file header.
• nrec - number of records in this trial.

record

• record type - one of the following values
  • 0 - point integer (e.g., a list of spike times)
  • 1 - float (e.g., membrane voltages)
  • 2 - point-value (e.g., integer time and float value)
  • 3 - event code (e.g., a list of event codes at integer times)
• record name - a name describing the record (e.g., unit0), which associates this record with similarly named records in other trials. A collection of records having the same name is called a channel.
• sampling - samples per second (e.g., 1000.0 for milliseconds)
• t0 - time origin of record data relative to the trial time reference.
• tn - time duration of the data (in sampling unints)
• n - number of data points.
• record data - the stored data depends on the record type.