File contents

Routine Name:   setuptau

Description:    Sets up A and B tables of voltage-dependent gates based on
                generic equations describing the form of the tau (time
		constant) and minf (steady state activation) state constants.

Usage:          setuptau channel-element gate AA AB AC AD AF BA BB BC BD BF \
                   -size n -range min max

                channel-element   This must be a tabgate or a tabchannel i.e.
                                  a voltage-dependent gate or channel with
                                  tabulated activation functions.

                gate              The name of the gate (must be X, Y, or Z).

                AA-AF             Coefficients A to F of the tau table 
                                  (see below).

                BA-BF             Coefficients A to F of the minf table 
                                  (see below).

                -size n           Number of divisions in the table 
                                  (default = 3000).

                -range min max    Range of the table (default: min = -0.100; 
                                  max = 0.050).

                
                This routine makes it easy to set up the A and B tables of 
                tabulated channels or gates (tabchannel or tabgate objects) 
                when the equations describing the state constants tau and 
                minf are of the form:

                y(x) = (A + B * x) / (C + exp((x + D) / F))

                Many standard channels have tau and minf parameters that can
		be cast into this form.  In these cases to set up the tables
		we can simply call setuptau with the parameter values as
		arguments to the function.  Since there are both tau and minf
		variables, we use the coefficients AA-AF to refer to the tau
		variable and BA-BF to refer to the minf variable.

		setuptau calls the TABCREATE action of the tabchannel or
		tabgate to allocate tables with n divisions (n + 1 entries)
		representing x values from min to max.  It then evalutes the
		functions at these points to fill the tables.

                tabgates have tables for the alpha and beta rate variables, so
                the tau and minf values are used to calculate alpha = minf/tau
                and beta = (1 - minf)/tau, and fill these tables.  The gates of
                a tabchannel each have an A table that holds alpha values, and
                a B table that contains alpha + beta.  These tables are filled
                with A = minf/tau and B = 1/tau.

                The tables are by default set up in "no interpolation" mode, 
                which means that the process of finding the correct table
                value is simply a lookup operation.  With 3000 divisions in 
                the table (i.e. the table size is 3001) the lookup usually
                provides sufficient accuracy.  If not, you can override this 
                behavior; see the tabchannel documentation.  The range of the
                activation variable is between -0.100 and 0.050 by default.
                This is adequate for most voltage-dependent channels, but 
                can be overridden using the -range option.

Example:	see Scripts/neurokit/prototypes/newbulbchan.g

Notes:		For tabgate elements, which represent only a single gate, the
		 "gate" argument is ignored, but it should still be given as
		X, Y, or Z.  The setupalpha command is similar to setuptau,
		but uses the rate variables alpha and beta instead.

          	If the tau and minf state constant of your channel cannot be
		described using the above equation, you must fill the tables
		with one of the other methods described in the tabchannel
		documentation, or The Book of GENESIS, Chapter 19 (2nd ed.).

                The state equation described above has a removable singularity
                when C = -1 at the point x = -D (which is, unfortunately, a
                common case).  In this case the routine may generate
                inaccurate results due to roundoff errors.  We STRONGLY
                RECOMMEND that if you must use a C value of -1 you check the
                resulting activation curves by using Neurokit (in the "Edit
                Channel" mode) to see if they look reasonable.  If they do
                not then you will have to define the channel using a script
                function as mentioned above.  Such a function will result in
                a slower setup time for the channel but will be much more
                accurate.

See also:       tabchannel, tabgate, setupalpha, tweakalpha, tweaktau, Tables