Re: the cubic Concept.

mayton wrote:

3. Search in an incomplete dial-up of attributes. Instead of search of a point we will search for set of the facts on a hyper-straight line or hyperplane and  . O (1) turns to linear or polynomial.

If a dial-up of attributes always full I already above suggested to consider  as a multidimentional array of bits
At the big size  probably filter Bluma somehow approaches.

Re: the cubic Concept.

mayton, esteem about a card of Carnot.

https://ru.wikipedia.org/wiki/_ wrote:

Cuba of Carnot  - a graphic method of minimization of the switching (Boolean) functions, providing the relative simplicity of operation with the big expressions and elimination of potential races [1]. Represents operations of a pairwise incomplete patching and the elementary absorption. Cards of Carnot are considered as the truth diagram of function rebuilt in appropriate way. cards of Carnot can be considered as certain plane scanning of a n-dimensional Boolean cube.

Re: the cubic Concept.

mayton wrote:

Digits. For a cube with the storage expenditure in 2Gbyte. (~17 179 869 184 boolean an element)...

2 Gb are a lot of for  but if to use  a file in storage it is possible to involve 2 Tb easily. Disk storage inexpensive, for this task suffices HDD.
Big rarefied files from OS zeroes are able to create, i.e. filling will be rather fast. At usage the file page by page will be displayed in storage, i.e.  it is not required much.

Re: the cubic Concept.

mayton wrote:

I will add that a cube does not love.
3. Search in an incomplete dial-up of attributes. Instead of search of a point we will search for set of the facts on a hyper-straight line or hyperplane and  . O (1) turns to linear or polynomial.

One of basic bases of cubes, this obtaining of a cutoff.... Instead of derivings of a specific point (Obs)

Re: the cubic Concept.

Roman Mejtes wrote:

it is passed...
One of basic bases of cubes, this obtaining of a cutoff.... Instead of derivings of a specific point (Obs)

You speak about OLAP. I speak about other data structure.

Re: the cubic Concept.

Dima T wrote:

it is passed...
If a dial-up of attributes always full I already above suggested to consider  as a multidimentional array of bits
At the big size  probably filter Bluma somehow approaches.

On the first point - yes. 100 %.
On filters. Almost yes. Provided that we do partial . Access to a DB on the former is necessary.
I not against Bluma but at first want to settle possibilities a pile of the first setting.

Re: the cubic Concept.

Dima T wrote:

mayton, esteem about a card of Carnot.
it is passed...

Carnot is a graphic representation of a boolean function. The method  demands involvement of the person of the expert which this minimization does manually.
The alternative is Kvajn which is automated but not the fact that constructs compact function. Actually here it is possible to raise the question about the basic possibility that that to minimize.

Re: the cubic Concept.

The specific target, let at least for an example, strongly would simplify decision search.

Re: the cubic Concept.

hVostt wrote:

the Specific target, let at least for an example, strongly would simplify decision search.

.

Re: the cubic Concept.

mayton wrote:

it is passed...
On the first point - yes. 100 %.

If yes, I do not have other ideas unless to be engaged in invention of selection of storage under  since zeroes there will be much more than . One of variants already offered

mayton wrote:

On filters. Almost yes. Provided that we do partial . Access to a DB on the former is necessary.
I not against Bluma but at first want to settle possibilities a pile of the first setting.

on the big sizes Blum most that that is necessary, let a small part it is necessary  on a DB, but differently  .

Re: the cubic Concept.

Dima T wrote:

it is passed...
on the big sizes Blum most that that is necessary, let a small part it is necessary  on a DB, but differently  .

I will add: you seem to me filter Bluma invent, but such which gives guaranteed "is/is not present" instead of "Probably/is not present". I would go by the way of probability increase "Probably is".
If and the initial data at you changes a little, it is possible to play functions  Bluma that at "Probably is" there was a minimum "really is not present". I.e. or at once you receive the answer "is not present", or you do request to a DB and you minimize the answer "is not present" at the expense of selection of functions .

Re: the cubic Concept.

Dima T wrote:

it is passed...
2 Gb are a lot of for  but if to use  a file in storage it is possible to involve 2 Tb easily. Disk storage inexpensive, for this task suffices HDD.
Big rarefied files from OS zeroes are able to create, i.e. filling will be rather fast. At usage the file page by page will be displayed in storage, i.e.  it is not required much.

On file system (random access) it is necessary to think of clever buffering. At initial initialization.

Re: the cubic Concept.

ViPRos wrote:

it is passed...
.

I will try to make an educational example. But you probably not in course of my tradition to create Friday topics of brainstorming session. I share thoughts instead of business by setting.
do not wait from me . Dream. Add.

Re: the cubic Concept.

mayton wrote:

it is passed...
On file system (random access) it is necessary to think of clever buffering. At initial initialization.

The essence of this idea in that that will think of it OS, i.e. it is not necessary  a question which you set.

Re: the cubic Concept.

Dima T wrote:

it is passed...
The essence of this idea in that that will think of it OS, i.e. it is not necessary  a question which you set.

It do not agree.

Re: the cubic Concept.

mayton wrote:

it is passed...
It do not agree.

The idea   certainly can be developed, but it is difficult and expensive, therefore I and suggested to use that that already is, i.e. OS means.

Re: the cubic Concept.

Dima T wrote:

it is passed...
The idea   certainly can be developed, but it is difficult and expensive, therefore I and suggested to use that that already is, i.e. OS means.

Give idea of a file cube we postpone for some time. Simply at me many thoughts under the very first version are real. And I for myself did not think over them.
Let's return to a disk then. Later.

Re: the cubic Concept.

mayton wrote:

I will try to make an educational example. But you probably not in course of my tradition to create Friday topics of brainstorming session. I share thoughts instead of business by setting.

Without an example to solve tasks it from area bring that,  that.
Because the first answer from Dima T results the finished decision of your task. As much as necessary measured array in the modern information system will be an one-dimensional array with offsets, therefore is told about a bit card is that head above which you will not spring. Technical questions, type how to bypass restriction on the RAM, too were offered - reflection on a file.
But if it is all not then let's consider in a context of any specific target. Type here there is it and it, and it is necessary and so. Then the sense will think.

mayton wrote:

for Vobshchem do not wait from me . Dream. Add.

Then it is necessary in area , machine intelligence, quantum , etc. All the same it cannot be checked up and in what to be convinced. But to dream - as much as necessary

Re: the cubic Concept.

hVostt wrote:

Then it is necessary in area , machine intelligence, quantum , etc. All the same it cannot be checked up and in what to be convinced. But to dream - as much as necessary

Well... As a topic-starter I all the same suggest to be restricted to a subject of the first post.
But if you have what to tell on AI or  - I ask create a new branch.
I will probably be connected.

Re: the cubic Concept.

Concerning filter Bluma. I result a certain subjective comparative analysis of a cube and filter Bluma
In the form of a label. On each point I am ready to give the argumentation. Some points which
I marked stars (*) detailing and I demand it  a bit later in process of time.

Re: the cubic Concept.

(*) Memory consuming MBHC it is defined by the formula. . The formula of product of cardinalities of all measurements of tuples.
Plus waybills on usage of reference manuals of measurements.
Let's admit at us there is a label of the facts with 4 columns for 80000 lines. Despite great volume
Segment of the data - partite values of cardinalities are equal 8, 5, 3, 10. Fields have the various
Data types. A line. Date. Number. And one more number. But the type is not especially interesting to us. We create 4 reference manuals.
For display of lines of dates and unique numbers to a restricted range of indexes (it is desirable since zero).
0. 7, 0. 4 e.t.c.  a cube with volume on 8*5*3*10 = 1200 bits or 150 byte. It also is ours memory consuming
And certain  in storage which size I now cannot count a state of 4 reference manuals
But you can estimate proceeding from average metrics for grocery bases. What there? Names? The codes of the goods?
Invent.

Re: the cubic Concept.

(**) Support  Iterations in a multidimentional Boolean cube is possible.
At the very beginning of a topic I wrote that is impossible. Or it is difficult.
look. A cube basic purpose - storage and extraction of the facts in a mode rigid OLTP.
we assume pointwise requests of a type

``````SELECT booleanValue FROM CUBE1
WHERE (field1, field2, field3, field4) in (1,4,2,0)``````

What sense to do reboric algorithms in such conditions? I do not know. Can it is simple COUNT () all facts.
Invent. Time complexity will be linear if we  field1 and will grow if to discard
Other retrieval delimiters. A word search in the full coincidence of all keys of a tuple - is preferable.
Iteration - is possible but the cube loses meaning. Then it is better to search in a DB.

Re: the cubic Concept.

(***) Support of interval requests.
In the beginning of a topic I mentioned that that cube axes  to enumeration values. Enumeration.
But basically if our value is a sredne-revenue that at us there is a possibility
To conduct a rank or comparing of tuples cubed on one axis.
For example to select all facts where the sredne-revenue from a range from 50 000\$ to 950 000\$.
Certainly the original request should be transformed through
Classes of incomes in a certain integer value of an index. Here for effect achievement
We should organize physically a cube thus that the facts of this range lay nearby.

``````SELECT decodeDimension1 (), decodeDimension2 () FROM CUBE2
WHERE map2Dimension4Range (50 000\$, 950 000\$)``````

Re: the cubic Concept.

Further (****) the Extension of a segment of the data for filter Bluma.
I here will long not describe the theory. Can esteem Vicks. Filter Bluma in essence does not extend.
It it is possible  anew and using an original DB  all amount the facts in a new segment.
Adding of the new facts - probably but leads to selectivity loss.
I entered the simple enough metrics. An amount of zero and individual bits in a card should
To be approximately 50 %. As soon as  reach half in process of filling - it is a high time to give
Signal  on reorganization.
Here. Concerning MBHC and extensions in one measurement. I think probably. If in example CUBE2
We  hardly there is more than storage (with a store) also we cut a cube on hyperplane which correspond to average revenues. And thus  crude storage of an operating system is filled by a cube on 80 % that to us hinders nothing
To format its continuation under the following hyperplane and to consider that a cube it was expanded.
Why we not  all at once? I suggest to think and of the possible extension of other measurements
In conditions when we do not know as handbook data will grow. Purely technically it is simple driving
Bits in  (with  padding) and all other. Business tiresome and  and I now simply do not want
Of it to think but I state that technically - probably. If strongly it is necessary - that we expand.

Re: the cubic Concept.

Further (*****) compression Possibility.
It is a sore point. It is possible to tell a feeble place of algorithm. I will bring under it practical
Basis. The educational circuit of type of Borej and patience to estimate only is necessary to me as lay down
In a cubic cube I will not build a label while. I have enough For an estimation for example
that-de Borej in SQLite and to count its cardinalities.
Not to flood words I enter function

``def cubeCardinality (tableRef) =....``

And on it I will refer as to the main metrics of efficiency of a cube regarding storage.
Well and for simplicity still number of lines

``def rowCount (tableRef) =....``

Probably most part of labels even does not lay down. Especially it concerns real fields.
- it is not general-purpose.