Vector/matrix: subtracting linear response

Exercises

1) Run plotdata.hoc in the course/vec directory. This reads the contents of data4.dat and displays four curves: the voltage responses to hyperpolarizing current pulses of -9, -18, -27, and -36 nA. Scale these curves according to the magnitude of the current pulse to see if the response was linear.

Hints

A good way to test a Vector method is to pick a curve into the clipboard, apply the method, and either printf the result or plot it in a NEURONMainMenu/Display tool. For example, if we pick the 36nA curve and put it in a Vector/Display and then pick the 9nA curve, scale it with

	hoc_obj_.mul(4)
and then put it in the tool, we suddenly realize that the curves need to be shifted to 0 before being scaled. One way to shift would be merely to select a point on the resting curve and subtract that value from the entire curve, ie
	hoc_obj_.sub(hoc_obj_.x[10])
If the resting value is noisy, we could subtract the mean value over a short range as in
	hoc_obj_.sub(hoc_obj_.mean(5,15))
To modify a column in a matrix, get it into a Vector with m.getcol(i) and put a Vector into the column with m.setcol(i, vector). i.e one can scale a column in one statement with
m.setcol(i, m.getcol(i).sub(m.getcol(i).mean(5,15)).div(9*i))
but long nested chains get easier to write than to read and it is generally better to do only one or two Vector operations per statement.

2) This has nothing to do with data but it will give you practice in Vector manipulations. Set up a simulation with several action potentials, e.g. with a long but low amplitude current pulse in an HH patch. Run it with the variable time step method. Now, plot log10(dt) as a function of t.

Hints

Use the clipboard. The t values are in hoc_obj_[1]. Calculate the dt values in hoc_obj_ (a synonym for hoc_obj_[0]) and use a Vector/Display to plot it.

Some pertinent Vector methods are
vec.c return a clone (new identical copy) of the vector instance
vec.deriv(1, 1) v.x[i+1]-v.x[i] -> v.x[i] , size is 1 less than before
vec.resize(vec.size-1) reduce size of vector by 1
vec.log10 log10(v.x[i]) -> v.x[i] , log10(0) is an error
vec.remove(index_vector) remove all elements at values given by index_vector
index_vector = vec.where("==", 0) indices where v.x[i] == 0

If removing 0 elements seems too complicated, it is less elegant but just as effective to add 1e-9 to every element before taking the log.


NEURON hands-on course
Copyright © 1998, 1999 by N.T. Carnevale and M.L. Hines, all rights reserved.