Since I began this blog I have mentioned many projects I planned to start, but have not yet followed up on them. I thought I would take this time to just give a brief overview of the main ones: SERO, Coulomb, and DeepNeurone. I am also going to talk about a new project of mine.
Recently my internet connection dropped out so I had a chance to work on DeepNeurone. Unfortunately due to the scope of the project it is unlikely I will get a working worm brain, let alone a human brain. I have, however, managed to write a computer program which is able to have neurones, synapses, spikes, rebalancing and reduction. If you look at the largeish output file below you can see the output of one of the runs.
In the run I aimed to create 50 neurones with synapses between them. The program prints out this layout but it is largely rubbish - it is just a random network the machine came up with. It then generates the thing you see in the file. This takes about 5 minutes to run. In initial runs it generally worked itself to a stable area (after about 30 runs) in which the most neurones it could fire were firing, but every neurone fired another which fired another which fired the original - basically, it wasn't moving. Then I thought that maybe it needs some random firings added in - so I made it randomly flip a neurone every once in a while. This produces the pattern you see.
This has largely gone on the back burner - I do not have the time, energy, or money to work on this project. It does not seem to have much use at the moment, as any components for it I am able to fabricate will generally not be powerful enough. Maybe in the future.
Yet again, mainly on the back burner. I might revisit this, but the majority of the code has just become unorganised. The simulation basically devolved very inaccurately due to the floating point issues I ran into.
And finally, some good news - TBD CHemkit.
The final entry on this list is a new project. I have no idea what to call it yet, but it is quite an interesting idea. I do chemistry, and occasionally I need to balance an equation, calculate a reactions products or do atom economy/percentage yield calculations. This can get a bit overwhelming (and I have a paper shortage :D), but the majority of tools I have found are generally either expensive or just drawing tools. My idea is to write a custom built chemistry program whereby you can enter some products, some reactants, and it can calculate the molar ratios. It can then operate on this and work out atom economies, etc. Eventually I hope to make it be able to work out reaction products (but this could be hard - I can see how I could do it for simple compounds, but as you get larger it would be difficult). I have no idea if this will be a command line program, a website (might be nice?) or a GUI app. Should be fun regardless.