hmm resumed representation theory study this weekend.

hmm i think its not possible to finish all chapters by monday. 

But maybe, on next monday.

I also in tandem started to search existing libraries. 

In python scape, I previously experimented with Sage library some modules. 

Then  I also checked cpp/c libraries.

Then also checked the distributed system mechanism of erlang slightly and its cpp integration mechanisms.

If I werent so bored with scala, it might made sense to code as DSL some topics in this library mentioned to be devised.

But I am that bored of scala that I would rather write in DSL of compiler compiler languages :D (That which boringly generates boring java language either, aha no, that was antlr that converts to java, cpu ops conversion is usually manually coded to yacc)  but with interfaces of placing code integrations. 

So I started investigating how to design this library mentioned. 

e.g. how to define a ring. without using abstractions of Sage or other libraries. 

or how to define those DSL's or BNFs and their multiplication or summation operator (e.g. convolution operator alike for some algebras ) 

should I define instead with an ontology language? of course thats necessary. some conceptual definition set. 

should I define the definition in a knowledge db? 

or should i define at Yacc like tools initially to be later defined also ontological definitions?

or should I define it in prolog? 

so design of library currently has these design questions. 

I think it makes sense to surely define the ontology inside a knowledge db. 

and as native functioning, it should have a translator to yacc modules later. which would be most nearest to cpu level definition set. 

ok I decided this design decision. Lets define the DSL in a knowledge db. (with a self devised ontology language /definition sets). 

and for yacc alike  CPUnear implementations, it could be translated from there to the yacc definitions automatically.  no need to code it with yacc. lets define its ontology language constructs (BNF alike) in knowledge db. 

I also checked some libraries e.g. like openBLAS or Lapack,   I mean I had not checked yet but learnt that e.g. openBLAS might be good. 

sicne the matrix based repreasentations calculations could be fastly generated and computed from these ontology layers defined in knowledge db with an ontology layer of this optimizations library.

so this optimizations library even calculations might be done with OpenBLAS or so, its inference mechanisms decisive optimizations deciders would be defined with knowledge db layer. 

meanwhile, I saw sections in this book, like circulant matrix and its eigenvalue computations. yayyy eigenvalue computation systems for possible Lenz-Ising systems definitions alike.  I am not in that chapters yet.  I just sometimes fast forward and read other chapters slightly (due to curiosity )

hmm ***  I have to relearn the knowledge db lang which I had studied before. 

nevertheless its required. 

maybe I  postpone that partially to one half hour today and some half hour other day to revise.

yayyy would define the DSL items as nodes in knowledge graph and relations also. (e.g. "is" relation etc etc) 

This layer has not need to be by itself alot intelligent.  It would be a representation and optimizations layer alike. I mean e.g. theorem proving skills? -> it might slightly need to have some such skills basically & not very much still.  and it should also include native type optimizations knowledge already defined. 

but it should enable the later ml layers to build theorem proving skills type generality it should provide. 

although it should also have some basical inductive theorem proving skills at this layer. but not very much. that would be to be discovered such skills by other layers most. 

yepp these be the design principles of this ml layer.  an optimizations library layer (and representation also). a library like layer.