yep revised chapter 5 (excluding monoids one theorem since i feel tired(due to sinusitis), would check it later) then now switching to (or not, dunno, since i feel tired due to sinusitis ) boolean algebra related chapter.
hmm today had not passed efficient. I should already have had finished the boolean algebra related chapters.
yep ahead to be learnt topics are initially these boolean algebraic systems (and first order and higher order concepts e.g. Tarski, Kripke etc) and then later intuitionistic maths chapter.
hmm i think i might need to rebuy some group theory books hmm also modules/fields study were incomplete. but those not urgent for this week.
hmm imho it makes sense to first complete DSL study for logic and initial NLP translations first before group theoretic library study.
then for that I first need to learn also hmm intuitionistic maths and coq/agda/unimath libraries also and some compiler book study to help me write DSL more correctly.
hmm so i first need to finish these studies to then move to DSL for logic systems and for NLP translations etc. then i should code the category theory diagram graph structure in big data then imho thats time to revisit group theory/modules/fields theories/representation theory methods
hmm imho project's kind of alot either add or adhd creating phase ended(endlessly trying to study to phd msc topics to learn some introduction to phd msc maths topics). but very hardwork required phase started of e.g. NLP translation studies etc etc.
it were before like, i had to study to many books maths topics. started with measure theory (had not finished, only started) then topology i think, then group theory, and real analysis, and some such many topics in also linguistics I also had to check awhile just trying to setup design of the project.
e.g. its like a project even design of requires studying to msc/phd and bsc topics. some quite a lot design challenge.
before now it were slow paced for this iteration. whihc i slowly checked representation theory referenced from phd group theory book and were slowly iterating since it were so discrepant from the project's coding phase the required msc/phd topics study sometimes and that creates some disconnectness between project's iteration and the tasks. but nevertheless its required to study msc/phd topics to me since i were ignorant in maths msc/phd topics and its required to be somewhat knowledgeable in designing such ai/ml engineering project.
but there were some disconnect mode of studied msc/phd books were slightly alot much effort that is not directly binded to the project's tasks very directly. so that creates a lot add/adhd for thinking diverging from projects required study tasks. or just feeling lost inside tons of phd/msc topics/books to learn read. somekind of moment of disconnect from iteration feeling of project iterations alike phase last iteration until now were. one before iteration before that were unlike that, it were linguistics studies including phase,and so, also some coding studies also including phase there were in between.
so now a very hardwork requiring iteration started. since DSL study would be resumed and also coding of graph structure either.
and also remaining topics studies e.g. heyting stuff or tarski system or kripke etc or compiler book, all would need to be studied in short period e.g. unlike representation theory which i would start then pause then restart and it would be like endless studying since i were also working to data engineering in day time and i were studying to project in evenings or on weekends.
since i curently work fully to this project, i have better chance to iterate faster. but still couldnt iterate fast when studying to Tom Leinster's book. since category theory is really a complete different maths universe imho. getting used to arrow concepts epics/monics etc is really something conceptually different than previous math methods to us we computer engineering side. since i mean even the group theory book other than commutativity diagrams, wouldnt have alot references to diagrams initially (bsc bookone) so this field is a very new domain and it really were severely challenging to try to self didactively learn of category theory topics. its just a new paradigm imho to we engineering people whom knew before only as linear algebra/calculus as the maths. (like if i am expert in calculus? no:D but still category theory paradigms are quite unlike, it requires at some point some msc level maths knowledge/perspective as understood) so being quite ignorant in this abstract side of maths, (still quite ignorant yet) (only studied to introduction of, e.g. dunno of galois connections yet etc etc, just only studied yet only some of introduction knowledge to these topics) it took quite a lot effort to get just introduce myself to abstract math topics some. so thereby even if i worked full time to ml algos project when studying to category theory last mont, studying to tom Leinster's book felt very hard type of task to me, since category theory really is quite a new paradigm/math universe. so i think i spent a lot time to just only finish tom leinster book first. like maybe 20 days or so. it were very hard challenge to me. since there is by then still there were some disconnect feeling to project's actual DSL study. e.g. couldnt design the commutativity diagrams yet from what i studied there directly first. when topoi were studied, disconnect disappeared. I before stdying to category theory, initially were intending to design abstractions with group theoretic and representatino theory wise constructs. but that initiative also turned out not very efficient method. then there were some unhappy situation of still happening disconnect feeling, some blur of actual how to design DSLs part. some ideas in mind. but all unclear. I mean some representation tehory wise diagram one side, some method ideas in some other drawing to DSL studies. but not any clear approach that is convincing to be having enough representation coverage/capacity directly. so felt like still very disconnected from the design of DSL since it were a mixture of ideas but not clear. and not suited for th abstraction required at many cases. so I been like kind of trying to find the most efficient abstraction for the propject tasks with some unclarity in yet designed thought approaches. but as soon as topoi were learnt, such unclarity has resolved. snce this is the best approach to design abstraction.
e.g. i were like unhappy that:
I were postponing how to bind ontology information to DSLs. initially when in last iteration, i revisited some NLP studies and worked to select some NLP libraries for part of speech parsing etc, and constituency /dependency info alike, and I had thought of some ideas (which required very much neural net training times (Whihc time is not a sparsely available resource of project, we dont have enough time is the default time setting of project) ) which would require a lot effort etc.
then i thought lets postpone it. i would somehow find a DSL wise binding to ontology e.g. is-a relation alike.
but this were creating some kind of disconent in my mind either. for not having solved how to bind the ontology db information to project's design of NLP yet :)
but after studying to topois, all these unclarity is gone.
Its like a project that design of constantly had been decided along the project's schedule. and there is some project iterations like.
e.g. module a layer a.1 -> ok i design in this sprint.
module a layer a.2 -> I do not have any idea to how to design now and postpone to following iteration :)
alike very agile iterations it has. but still it requires some level of ready design decisions also. its not possible to design it very most agile. since module a layer a.2's design decisions might be requiring some architectural design pattern in previous iteration's design decisions.
so it were very stressful. to have modules layers which i postponed design decisions of with having initially not any much idea of how to merge to DSL. :) so that lack of knowledge of design decisions were a constant stress factor in my mind :)
it were a very funny project in the sense ->
i mean all iterations has this aspect. of many modules i initially have no design idea of first initially, then study a batch of courses/msc/phd books to then decide some design principals alike.
what a challenge its to, to study without having anyn knowledge or knowing whether if the msc topic i learn would be beneficial or not to the project's tasks :)
i mean in any project you have some iterations righ? in this project due to this aspect, in many moments you completely lose the iteration aspect. e.g. it happens since also i in day time were working to data engineering and in evenings or weekends trying to study to this project. there would many times time points happens like -> I start some msc book and finish some chapters. then some pause to project. :) then i restart then have to restudy since i would pause like a month or 2 weeks and would forget some of topics i learn. and then repeating such loop in many books. :) it were not any easy to stdudy such project in evenings only and weekends only. since i mean you lose the iteration of project even.
its quite a challenge :) such a very agile project :)
not only that, I many times felt lost in what to study., since i mean I had phd/msc books library. from commutative algebra to algebraic topology. I mean how to iterate my studies. and how to not over study e.g. study a phd topic that is not directly relevant to project some phase. that were also very challenge to me since I were very ignorant in maths phd/msc topics at all. there fore to setup some study plan were also some not an easy task to me. since i wouldnt know which book to study first. so its has had some such challenges resulting from self didactive project study aspect.
i mean its not the standard engineering project you have architectural templates ready invented before alike. a high level NLP libs integration/ontology engineering/and various such tasks project but even requires alot msc/phd topics studying. (due to my previous very ignorance in msc/phd topics in maths)
plus my ignorance in maths msc/phd topics (abstract maths) were also a very disadvantage initially. (which i tried to fix by studying to msc/phd topics to be able to design even.)
but here we are finally at some iteration which we could say as one iteration before 0.1 version of project.
Yorumlar
Yorum Gönder