synchronisation

[Sir-Bip-Bop]

Measure the synchrony of the system, and compare the result depending if the gap junction existed or not

• [x] need to simulate for a long time - 100 spikes at least
• [x] start working with a constant intensity and noise
• [x] chi measurement
• [x] van Rossum measurement
• [x] spike count measurement
• [ ] studies depending on magnitude of intensity
• [x] studies depending on magnitude of noise
• [x] make raster plot of spike times
• [x] calculate synchrony for completely out of phase systems

[Sir-Bip-Bop]

problems for van_Rossum distance for out of phase systems

[Sir-Bip-Bop]

Update from march 8th meeting:

• [x] Change the intensity value to try to get all the firing rates to be roughly equal (Look at ML)
• [x] Debug Van_Rossum measurement (definition of time array was done incorrectly)

Once that's done, set up the synchrony measurements experiments:

• [x] Benchmark each of the three measurements with a constant intensity, and create 'fake' completely synchronised and completely out of phase data. Experiments to do with each of the models:
• [x] Set up simulations for a different number of seeds to compute the mean and variance of the synchrony measurement.
• [x] Compare results gap junction on vs. gap junction off
• [x] Simulate and study the changes in function of the gap junction strength
• [x] Simulate and study the changes in function of noise amplitude (variance on the noise generation) As a side note, study whether Chi measurements should drop to 0 or not

[Sir-Bip-Bop]

Update from Pi day:

• [x] Fix the spiking conditions of ML and HH
• [x] Change the time constant fro the measurement of the Van Rossum distance
• [x] Re-do the synchronisation measurements

[Sir-Bip-Bop]

Update from March 29th:

• [x] Change decay value for Van Rossum (adjusting to dt)
• [x] Scale Van Rossum by firing rate
• [x] Run simulations for longer and see if there is a mismatch (not much to fix on that)
• [x] Change noise to (1 + 0.01*noise)
• [x] Plot a relative Chi measure (in function of Chi at k = 0)
• [x] Bring the original firing state back
• [x] Study multi-neuron measurements
• [x] Make sure that the noise is not giving fake spikes

[Sir-Bip-Bop]

Update from April 11/12th:

• [x] Paralelise the code
• [x] Use relative values for both the Van Rossum distance and Chi
• [ ] Implement R statistics
• [ ] Implement ISI
• [x] Adjust the external drive - k relation
• [x] Check that the IF firing rate is not due to numerical errors

[Sir-Bip-Bop]

Update from April 20:

• [ ] Fix the implementation of the Reliability measurement
• [ ] Implement ISI measurements
• [x] Adjust with the width of the noise for the Izhikevich model
• [x] Reduce simulation time but increase the simulation number by a greater amount
• [x] Introduce the multiprocessing system to the whole simulation - Start utilising synchrony_parallel
• [x] Save the voltage traces and scaterplots of the simulations for a comparison afterwards (and to try to understand the origin of the differences)

[Sir-Bip-Bop]

Update from April 25/26:

• [x] Chi measurement: Track average voltage and variance independently
• [x] Van Rossum: plot f(x) - g(x)
• [x] Models: Add a threshold to the spiking of HH and IZH and check if that changes anything
• [ ] Cut off the convolution tail: Resulted in a lot of problems when computing the distances
• [ ] Make a function to compute the interspike time

[Sir-Bip-Bop]

Update from May 10:

• [x] Revise all the units to show everything in miliseconds
• [x] Plot the componenets of the Chi measurement
• [ ] Look at the LIF vs. HH model
• [ ] Adjust spikelet height of all the models (and compare against noise amplitude)
• [x] Make the IF noise narrower
• [ ] Prepare the experiment for larger networks