cinpla / neuronify Goto Github PK

It should be possible to zoom and scroll in the plots.

It should be possible to add notes and draw arrows on the canvas

Neurons that are bistable, either in terms of firing or polarization. A neuron may switch states upon receiving excitatory or inhibitory input.

Many units are currently arbitrary. Units such as time and conductances should be made more visible and realistic.

Spike timing dependent. Check if presynaptic spike is soon followed by a postsynaptic spike. If so, the weighting is increased. See chapter 7.5 in Sterratt. Other learning rules may also be applied.

Listening to the network firing could be interesting. The first implementation would be to add a sound when each cell is firing.

In tutorial mode, the image should be fixed (no or restricted panning), showing only part of the circuit with notes and explanations (see also #39 for notebook features).

This mode should be used to give a tutorial to Neuronify in general and to introduce the different circuits.

The cell border could be a bit thicker to make it more visible.

They should disappear completely, but they dont

Add easter eggs:

Konami code - start a molecular dynamics simulation using the neurons as particles.

A mobile user typically expects to keep working from where the app was left the last time.

Selecting a group of entities and saving them into a module. The modules should appear in the creation menu along with the usual entities (neurons, voltmeters, etc).

Such a button would be useful.

It should be possible to see the difference between inhibitory and excitatory neurons by looking at their endpoints. They could be fork-like or ball-like.

Add neurons with post-inhibitory rebound:
http://en.wikipedia.org/wiki/Central_pattern_generator#Mechanism_of_rhythmic_generators:_post-inhibitory_rebound

We should have a colorblind mode - where the excitatory/inhibitory features are not only distinguished by colors.

Once sounds have been implemented, different cell types should be assigned different sounds. Optionally, the user could be able to select a sound, pitch and volume.

Would be nice to be able to choose the time constants of the synapses.

On mobile devices, save and load should be implemented without a file manager. The best option is probably to use save slots.

Now its complete js, which lags (a lot). Should be some openGL solution.

Connection to voltmeter depends on radius, it makes it so the connection is far from the voltmeter when connected to the top or bottom side.

Add morse decoder in neuronify

Poisson generator should switch color when inhibitory.

The base neuron class is now very specialized. It should be a thin shell that is easily expanded by new features in separate classes. This is preferred to having a large, monolithic Neuron class.

Only sensors should be able to connect to neurons, not the other way around.

Depends on #47

Plotting is currently the biggest resource hog. We should reduce the number of points being plotted.

A module can be defined as a single unit and represented by a "black box" in the workspace. The user defines special input/output connections in the module that may be used to connect different modules in a simple way.

Voltmeter currently only supports four connected neurons. This may be improved by increasing the number of LineSeries or by making a more general implementation.

We should have an undo-button.

The current Plot.qml class is a joke. The way it works in terms of obtaining and storing data should be thoroughly redesigned. The new design should allow for upcoming features such as zooming, scrolling and saving.

We should make a type of neuron that spikes randomly (poisson process).

There should be an option to snap neurons to a grid (either visible or not) for easier creation of larger simulations.

It is easier to hold the phone in portrait mode with one hand.

Would be nice to be able to reset the membrane potentials to resting values by pushing a button for (i) individual and (ii) ALL neurons.

Saving of plots and data from the oscilloscopes. Perhaps also from all neurons?

Each poisson generator should be able the represent N independent poisson processes.

The axons are currently a bit hard to see on high-dpi displays. They could be a bit thicker.

The current Voltmeter and Plot implementation will draw two curves in the same plot on top of each other in z-direction. It should be possible to plot the two with a shared x-axis, but separate y-axes. That is, on top of each other in y-direction.

Listening to all cells can be a bit noisy. Perhaps adding a sensor that plays a sound depending on the distance to the nearby cells could be a way to solve this.

Selected items should be on top of the z stack so that they are painted on top of all other objects. This makes it easier to click their MouseArea too.

As caused by opening of calcium channels, neurons should be able to exhibit long-lasting depolarization after receiving high-frequency input. This may be deactivated again by inhibition.

Remove current advanced controls (or move to separate menu) such as slider that changes from excitatory to inhibitory and rather add separate neurons to the creation menu.

Windows:

When in menu, pressing select simulation crashes neuronify.