Approximately 2^(n^2)-2^(2n) - 2n as n increases?

On Mon, Aug 27, 2012 at 8:16 PM, Jim Bromer <[email protected]> wrote:
> Damn.
> Approximately 2^(n^2)-2^(2n) - n as n increases?
>
>
> On Mon, Aug 27, 2012 at 8:14 PM, Jim Bromer <[email protected]> wrote:
>> Approximately 2^(n^2)-2^(2n)-log(2)n as n increases?
>>
>>
>> On Mon, Aug 27, 2012 at 8:11 PM, Jim Bromer <[email protected]> wrote:
>>> 2^(n^2)-2^(2n)-log(e)n as n increases approximately? (That is just a
>>> guess. Something about the proportion of primes since the primes won't
>>> be detected.)
>>> Maybe.
>>>
>>> On Mon, Aug 27, 2012 at 8:07 PM, Jim Bromer <[email protected]> wrote:
>>>> That last messaged contained another error.
>>>> There are 2^(n^2) possible ways to fill the n^2 bits of the partial
>>>> products but there are 2^(2n) possible ways to do it with a
>>>> well-formed set (given a nXn bit multiplier). So the worse case is
>>>> that I might have to go through would be 2^(n^2)-2^(2n) trials before
>>>> I found a system that represented a number that could be factored. So
>>>> the conclusion is the same.
>>>> I think that is right - or at least it is getting closer to the ballpark.
>>>> Jim Bromer
>>>>
>>>> On Mon, Aug 27, 2012 at 7:44 PM, Jim Bromer <[email protected]> wrote:
>>>>> Ok the wheels are slipping but train is slowly getting up to steam.
>>>>>
>>>>> How many possible ways are there to set up the n^2 bits of the partial
>>>>> products?  2^(n^2).  Only n^2 are well-formed.  (This is for a nXn
>>>>> multiplication algorithm). So even if I had the dreamed of polynomial
>>>>> time solution to Boolean Satisfiabiity the worse case is that I might
>>>>> have to go through 2^(n^2)-n^2 trials before I got a factorization.
>>>>>
>>>>> So even though you do not have to make the multiplication algorithm
>>>>> reversible in order to derive a Boolean Formula that will detect a
>>>>> non-prime number (given a set of possible partial products), if the
>>>>> goal is to use the method with a polynomial time solution to Boolean
>>>>> Satisfiabilty then you are going to have to make it reversible (or
>>>>> something like that) if you are want to use it to find a factorization
>>>>> solution to any given number in polynomial time.
>>>>>
>>>>> Jim Bromer
>>>>> (I might be wrong...It has happened. I admit it.)
>>>>>
>>>>>
>>>>>
>>>>> On Mon, Aug 27, 2012 at 7:21 PM, Jim Bromer <[email protected]> wrote:
>>>>>> Unfortunately I was wrong about this, but the method does have a flaw.
>>>>>>
>>>>>> The partial products are fully specified in the sense that they detail
>>>>>> the relationships between each bit as it is used in the cross-products
>>>>>> in a consistent manner based on the bits of the multiplicand.  So if
>>>>>> you had a Boolean Solver then the Boolean formula of the cross
>>>>>> products (derived from a standard multiplication algorithm) could be
>>>>>> used to determine if there was a solution for a given system of
>>>>>> cross-products.
>>>>>>
>>>>>> However, if the system detected that there was no solution to a
>>>>>> problem (given a system of partial products) it would not definitely
>>>>>> represent a prime number since the given system could be poorly
>>>>>> formed.  On the other hand it should be able to detect a
>>>>>> factorization.  If it found that a system could be factored based on
>>>>>> the fact that there was a solution to the Boolean Formula, then the
>>>>>> given system of partial products could be added to detect the number
>>>>>> that can be factored.
>>>>>>
>>>>>> Jim Bromer
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Mon, Aug 27, 2012 at 6:48 PM, Jim Bromer <[email protected]> wrote:
>>>>>>> I was really surprised when I discovered that converting the
>>>>>>> cross-products, used to determine the partial products, of a binary
>>>>>>> multiplication into the form of Boolean logic was logically simple.
>>>>>>> It turns out that the methodology of the multiplication problem, when
>>>>>>> you are given the multiplicands is simple just as the multiplication
>>>>>>> algorithm is simple.  However, if you were to try to work backwards to
>>>>>>> use the partial products to try to determine the multiplicands the
>>>>>>> problem is much more difficult.  It is simple as long as poorly formed
>>>>>>> partial products have been filtered out beforehand.  The formalization
>>>>>>> of the problem, (expressed in pure Boolean form), is not quite so
>>>>>>> simple if the algorithm is supposed to detect or avoid poorly formed
>>>>>>> partial products.
>>>>>>>
>>>>>>> It took me a long time to figure this out so I won't be calling any of
>>>>>>> the other boys in this group slow anytime soon.  So anyway, if you
>>>>>>> wanted to use the multiplication algorithm as a factorization
>>>>>>> algorithm you would have to work the algorithm backwards in order to
>>>>>>> determine the multiplicands given the product, or to determine the
>>>>>>> multiplicands given a system of partial products.  The use of the
>>>>>>> partial products as the given isn't simple because if a bit in a
>>>>>>> partial product = 0 it could be because the corresponding bits in one
>>>>>>> or both of the multiplicands that produced the cross product were 0.
>>>>>>> That shows that there are three ways to explain a bit in a cross
>>>>>>> product that equals 0.  On the other hand if the bit in the cross
>>>>>>> product was 1 then we would know that both bits were 1, which shows
>>>>>>> that there should be a reasonably "easy" way to define the Boolean
>>>>>>> Logic of the partial products so that it would only allow correctly
>>>>>>> formed partial products to get past the Boolean Formula.
>>>>>>>
>>>>>>> This reasoning shows how complicated turning an algorithm into a true
>>>>>>> Boolean Formula can be.  First we start off with the standard
>>>>>>> multiplication algorithm, ok. But if you want to use that algorithm in
>>>>>>> an unconventional way, you have to make sure that the implicit
>>>>>>> relations are well-formed in the sense that the unconventional usage
>>>>>>> will not present the given values in an unacceptable poorly-formed
>>>>>>> way.  So even if you want to consider the problem in purely abstract
>>>>>>> way -using only Boolean variables- you have to be prepared for
>>>>>>> unexpected effects when your claim jumps the conventional rails.
>>>>>>>
>>>>>>> If you want to take the conversion of a multiplication function into a
>>>>>>> pure (variable) Boolean form and then use it to detect a factorization
>>>>>>> given a solution to the derived Boolean Formula, the multiplication
>>>>>>> method would first have to be expressed as a reversible function.
>>>>>>> Jim Bromer


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com

Reply via email to