so lets write code with pyspark even being slower than normal spark, since i dont have time to investigate nonpython libraries i want fast data creation.

hmm then at the same time lets revise some basic neural networks topics e.g. generalization doer or also convolutiopnal nns etc etc or encoder decoder types etc etc.

since i think whilst i study revise topology i coould keep neural networks running to investigate whilst before moving to generative options which needs topology knowledge. so to follow pursue 2 solution options at the same time its better to start the nonlabeled data nn methods like encoders decoders or them being tied to convolutional nns etc. 

yayyy downloading pyspark to generate data which would be data points encoded somehow. not decided not designed yet.

so i wish i make this big data part of solution reality in some time and also in meanwhile find figure out nn components to test/try with then tie them then hmm in that post time revise topology study so that tomorrow i can start investigating also some generative approaches with some intiial predefined templates definitions. 

lets go full noninformation having approach of the data set first. for topology lets first depend on convolutional networks e.g. or so. e.g. instead of informative concepts like simplicial complexes lets go leave it to neural network first to discover but then put last layer to prize similarity to discover similarities. or to put encoder decoders to investigate underlying generative structure.  and to investigate later those underlying params neural network vlaues itself with another neural network. but this latter one being less optimistic or lets say not optimistic but i am right now more curious to convolutional geometry study. of encoding geometry such wise i mean. 

but the huge data size makes this a quite challenge. 

usually convolutional networks needs labeled data but this time it should be able to define without labeled data so that be an open design question how to do that. not that very challenging yet but just a task that needs to be solved.

 hmm starting with preparing the data now. 

inner structure of equivalence classes of integers some rfepresentation type