Page 1 of 1

vector recorded from CVODE simulation

Posted: Mon Nov 26, 2007 3:46 pm
by eacheon
Hi,

I am trying out using CVODE in simulation. Without changing my hoc code, the time is recorded in one vector using Vector.record (I am not using local variable time step here).

My problem and confusing is that in the time vector (referred to as "tvec" in the following) I got a few time steps that are zeros, which means, tvec values at these steps are equal. I checked the voltage traces recorded with tvec, it has different values when t values are the same. For example:

Code: Select all

tvec                   soma.v(0.5)
...
3.00000000e+01         -69.99025646 
3.00000000e+01         -69.98228519
...
which value is the true value I should use that corresponds to the voltage at soma at this time?

Posted: Tue Nov 27, 2007 10:43 am
by hines
Events normally cause discontinuities in some parameter or state variable.
I assume you do not have an event that does that for voltage. Is that voltage
difference on the order of your
absolute tolerance? If you send me your
model I will determine the cause of the
discontinuity.

Posted: Tue Nov 27, 2007 1:32 pm
by eacheon
hines wrote:Events normally cause discontinuities in some parameter or state variable.
I assume you do not have an event that does that for voltage. Is that voltage
difference on the order of your
absolute tolerance? If you send me your
model I will determine the cause of the
discontinuity.

No I do not have such events for soma.v(0.5), though I do have one that triggers synaptic conductance on a dendrite.

After looking at these values closely, I found at the same time points the values of states are the SAME, instead of different as indicated in my previous posts. So there's no problem.