Drivers

LiCoRICE uses the concept of drivers to interface with external systems. LiCoRICE includes drivers for a number of default inputs and outputs and gives users the ability to write custom drivers for additional I/O support. Code for default drivers can be found under the licorice/templates/source_drivers and licorice/templates/sink_drivers directories.

Sync vs Async

By default, LiCoRICE input and output driver code is run synchronously inside the respective source or sink. Drivers may also be run asynchronously by specifying the async: True flag on that driver’s external signal. This creates two processes for the asynchronous sources or sink: one async reader or async writer which reads in or writes data and handles buffering; and another realtime stamper which defines tick boundaries in the data by updating buffer housekeeping variables according to the realtime clock.

In-built Drivers

Source Drivers

line

Coming soon.

parport

Coming soon.

pipe

Coming soon.

pygame_joystick

Coming soon.

udp

Coming soon.

zmq

Coming soon.

Sink Drivers

disk

The disk sink driver defines a specialized logger output that creates SQLite database files for use in recording data from systems neuroscience experiments. Disk sinks should be defined as asynchronous since I/O operations will generally not fall within the realtime tick boundaries.

The disk driver’s behavior is defined in the YAML model file by an args dict on the sink output signal as well as a log dict on the signal definitions of the input signal. It creates a tick table by default which logs one row per LiCoRICE tick and includes 4 timing signals for which definitions are found in Modules Properties.

An example of the disk sink output signal args is as follows:

out:
    name: sqlite_log_out_signal
    async: True
    args:
        type: disk
        tick_table: tick
        save_file: ./data

And this args dict has the full specification:

Name	Type	Description
type	string	Must be set to disk
tick_table	string	Table name for tick-level table. This name must be valid in SQLite since it is spliced into SQL command without any error checking
save_file	string	Name of file for where data for the experiment should be stored. 48 character limit. DO NOT add .db to the end

Disk driver sinks may also take input signals that run at different frequencies to the LiCoRICE model and therefore would be cumbersome to store in the tick-level table. By specifying a log dict on the signal definition and setting the table keyword, signals can be logged in their own table with one row per timestep. If multiple non-tick signals are stored in the same table, they must be time-locked.

An example of the signal-level log dict is as follows:

log:
    enable: True
    table: two_khz
    save_file: ./data

This log dict may be simply a boolean to use disk defaults or can specify the arguments in a dict with the following keys:

Name	Type	Description
enable	boolean	Whether or not this signal should be logged. Defaults to True
type	string	scalar, vector, msgpack, or text. Defaults to auto which will pick the best among the first three options given signal shape and dtype.
table	string	The name of the table for this signal to be logged in. If this value is the same across multiple signals, then those signals must be time-locked meaning they produce the same amount of data on the same ticks

NOTE: Although all the keys are optional, at least one key must be specified in the dict so that the YAML syntax is valid

Using multiple disk sinks at once has not been tested.

line

Coming soon.

parport

Coming soon.

pipe

Coming soon.

udp

Coming soon.

vis_pygame

Coming soon.

zmq

Coming soon.