Code review/overview

This week me, Daniel and one of his PhD students, Chaitanya met at Nencki Institute of Experimental Biology and made a code review. We found few small flaws in the code, for example the new source distribution I introduced didn’t place the sources completely evenly.

We also discussed the 3D laplace equation for spherical-symmetric sources, using step-case. This alternative approach is about to be added to the 3D KCSD variant.

New functionality

Analytic relation between LFP and CSD

Daniel suggested that the package should offer a possibility to add a case in 3D kCSD, where it is possible to find an analytic solution of the relation between sources and potential.

He sent me a nice pdf with the idea explained. And the conculsion is that:

$$V(r) = \left \lbrace \begin{array}{ll} \displaystyle \frac{C}{4 \pi \sigma r} & \textrm{for} \,\, r > R \\[1em] \displaystyle \frac{3 C}{8 \pi \sigma R^3}(R^2-r^2/3) & \textrm{for} \,\, r < R \end{array} \right .$$

where C=1 is the total current, R is source radius, r is actual position and $\sigma$ is conductance.

Tasks before the evaluation

We discussed a list of tasks that should be done before the final evaluation. It consists of:

• improving validation method in 2d/3d against model data

• creating validation framework

• recalculating normalization constants in basis functions

• comparing images matlab/python visually (with provided data)

• time series management

• cross validation for R and h parameters at once

• tests on bigger datasets: 100, 1000, 10k electrodes and time frames
• describing workflows
• package deployment