Re: Can you solve this interval problem using IEEE-stds 1788
Dear Mehran, Walter is not the only one confused by your email exchange, my impression is that most folks from the list are as confused as he is. As a result, we are unable to think about what you pointed out -- as you suggested -- since we do not understand what exactly you were trying to point out.
I think that what Walter proposed is a good idea: that you try to explain to him what exactly are your problems and your objections.
Once this is clarified, Walter and I would forward it to the whole list, and we will be glad to resume the discussions.
We will be waiting for the results of this discussion, different interval versions of equations are indeed an important topic, we have united solution, tolerance solution, etc., any progress in this will be very helpful
Vladik
P.S. If I am too pessimistic and there is someone on this list who understands what exactly you had in mind, I am asking this person to send the explanations to the list, ideally after checking with you first, Walter tried that and it looks like what he understood was not what you have in mind.
________________________________________
From: stds-1788@xxxxxxxx <stds-1788@xxxxxxxx> on behalf of Mehran Mazandarani <me.mazandarani@xxxxxxxxx>
Sent: Monday, November 28, 2016 1:05 AM
To: Walter Mascarenhas
Cc: stds-1788; Andrzej Piegat
Subject: Re: Can you solve this interval problem using IEEE-stds 1788
Dear Walter,
Yes. Your understand of the problem sent by me is wrong. The problem
sent to the group is quite different from what you imagined.
No problem, I continue our discussion in private. But I hope the
members of IEEE-stds 1788 think about what I pointed out and this key
point that whether the four basic operations defined in the standard
enable us to deal with real or at least designated problems, or not.
Indeed, my impression is not positive.
Warmest regards,
On 11/27/16, Walter Mascarenhas <walter.mascarenhas@xxxxxxxxx> wrote:
> Mehran,
>
> I probably misunderstood your point.
> I was under the impression that you considered a problem
> like this one:
>
> Suppose you are given intervals u and v and compute an
> interval x using intermediate interval expressions like
>
> a = u - 2 * v
> b = a - 3 * u + 2 * v
> c = a - v + u - b
> x = a + 3 v - b + c
>
> How can you compute a sharp bound on x?
>
> This is a linear programming problem, and I thought
> that you were considering this kind of question.
>
> If this is not the kind of question in which you
> are interested, then let me know in private, so
> that we can continue our conversation without
> bothering the other members of the list.
>
> walter.
>
>
>
>
>
>
>
>
>
> On Sun, Nov 27, 2016 at 10:50 AM, Mehran Mazandarani <
> me.mazandarani@xxxxxxxxx> wrote:
>
>> Dear Walter,
>> Thank you for your comment.
>> You may know, the equations are not equivalent based on the standard.
>> Please see the attachment. I was wondering how you could define an LP
>> problem to obtain the unique solution.
>> I would be pleased if you could let me know how you deal with that.
>> Then, I may answer to your question, i.e. "Would your method allow one
>> to obtain bound which could
>> not be obtained via linear programming?"
>>
>> Thank you so much.
>>
>> On 11/26/16, Walter Mascarenhas <walter.mascarenhas@xxxxxxxxx> wrote:
>> > Dear Mehran,
>> >
>> > For simple cases like the one you mention, the bounds
>> > you obtain can be obtained easily using linear programming.
>> > Would your method allow one to obtain bound which could
>> > not be obtained via linear programming?
>> >
>> > The point that bothers me is not that you could get bounds
>> > in simple contrived examples. As I said above, this can
>> > be easily done with linear programming.
>> >
>> > The point is how you would turn your approach in a tool
>> > for GENERAL USE. I still believe that you would need to do
>> > a lot of work in order to have a software that would take
>> > a GENERAL PROBLEM and obtain bounds to it.
>> >
>> > Dealing with particular cases is not, and should not be,
>> > the purpose of a standard.
>> >
>> > Until you present a software able to handle general
>> > examples, proposed by other people, I believe you
>> > will not convince many people about your approach.
>> > In particular, a particular example like the one you
>> > present if far, far, from showing the power of your approach.
>> >
>> >
>> > regards,
>> >
>> > walter.
>> >
>> >
>> >
>> > On Sat, Nov 26, 2016 at 12:14 PM, Mehran Mazandarani <
>> > me.mazandarani@xxxxxxxxx> wrote:
>> >
>> >> Dear Walter and member of IEEE-stds1788
>> >> Thank you for your comments.
>> >> I am waiting for the result of one of my paper in which we
>> >> investigated the interval computing in the context of fuzzy sets and
>> >> showed how the approach is able to address the practical cases.
>> >> Once I receive the final result, the paper will be sent to you.
>> >> Indeed, I have another approach to deal with interval arithmetic that
>> >> I'm working on.
>> >>
>> >> For the problem sent to the group, you can see the attachment
>> >> including the proposed approach for obtaining the solution.
>> >>
>> >>
>> >> Best wishes,
>> >>
>> >> On 11/25/16, Walter Mascarenhas <walter.mascarenhas@xxxxxxxxx> wrote:
>> >> > Mehran,
>> >> >
>> >> > I insist: unless you present a viable practical alternative
>> >> > to what is proposed in the standard your arguments will
>> >> > have little chance of being heard.
>> >> >
>> >> > You should also keep in mind that a standard is supposed
>> >> > to be a MINIMAL set of requirements, not a compilation of
>> >> > all conceivable alternatives. The more requirements you
>> >> > make in a standard, the less likely it will ever be implemented.
>> >> >
>> >> > In my opinion, the standard is already too complex and
>> >> > large as it is. Including new versions for the arithmetic operations
>> >> > would lead to something completely out of reach for the few souls
>> >> > who have the endurance to turn these abstractions into real
>> >> > software.
>> >> >
>> >> > At least for me, practice matters a lot, and I don't see how you
>> >> > could get a simple practical solution for the problems you pose.
>> >> >
>> >> > Of course, I may be wrong (and it would not be the first time),
>> >> > but
>> >> > I would like to see practical evidence that this is indeed the case.
>> >> >
>> >> > My wife is an experimental molecular biologist and she always
>> >> > tells me: "ideas are cheap, do the experiment". Along the years
>> >> > I found little reason to disagree with her.
>> >> >
>> >> > So, I ask you again: do the experiment and let us know the result.
>> >> >
>> >> > walter.
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >
>> >> > On Fri, Nov 25, 2016 at 7:58 AM, Mehran Mazandarani <
>> >> > me.mazandarani@xxxxxxxxx> wrote:
>> >> >
>> >> >> Dear Walter,
>> >> >> Thank you for message.
>> >> >> I wasn't going to turn the presented software by Oliver down, and
>> also
>> >> >> I have not written a software yet, maybe in the future. But, the
>> >> >> key
>> >> >> point I am going to mention is that whether a researcher could
>> >> >> handle
>> >> >> such a problem based on the IEEE-stds 1788 or not?
>> >> >> You know, software works based on algorithm and method created by
>> >> >> human. It also may think and learn based on human's algorithm and
>> >> >> methods defined for. Then, if we don't have correct method to deal
>> >> >> with problem, therefore software does not either.
>> >> >>
>> >> >> I think we should not only define the standard operators in
>> >> >> IEEE-stds
>> >> >> 1788, but also define the "standard form".
>> >> >>
>> >> >> Warmest regards,
>> >> >>
>> >> >> On 11/25/16, Walter Mascarenhas <walter.mascarenhas@xxxxxxxxx>
>> wrote:
>> >> >> > Mehran,
>> >> >> >
>> >> >> > How would your software solve this problem?
>> >> >> >
>> >> >> > Would your software have the same general
>> >> >> > purpose as Oliver's, that is, would it work in
>> >> >> > all possible cases? Would it handle
>> >> >> > nonlinear functions for instance?
>> >> >> >
>> >> >> > Would your software work if you only had
>> >> >> > information (1) ? If it would then I would
>> >> >> > like to see it working.
>> >> >> >
>> >> >> > What I particularly WOULD NOT like to
>> >> >> > see is the comparison of a software
>> >> >> > that exists only on your imagination
>> >> >> > with real code, which has been
>> >> >> > tested and used by other people.
>> >> >> >
>> >> >> > If you can provide such a software then
>> >> >> > I believe your contribution would be
>> >> >> > much appreciated by the community.
>> >> >> >
>> >> >> > Until you provide this software your
>> >> >> > claims will be questionable.
>> >> >> >
>> >> >> > So, write it up and let us see the result.
>> >> >> >
>> >> >> > Oliver has done such effort and I praise
>> >> >> > him for that. If you do something similar
>> >> >> > I will praise you too.
>> >> >> >
>> >> >> > walter
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> > On Fri, Nov 25, 2016 at 5:35 AM, Mehran Mazandarani <
>> >> >> > me.mazandarani@xxxxxxxxx> wrote:
>> >> >> >
>> >> >> >> Dear Member of IEEE-stds 1788,
>> >> >> >> I was wondering whether one can obtain a unique solution to the
>> >> >> >> following naturally equivalent problems based on IEEE-stds 1788
>> >> >> >> or
>> >> >> >> not:
>> >> >> >>
>> >> >> >> X is unknown; u and v are interval numbers and known
>> >> >> >>
>> >> >> >> 1) X-u=v
>> >> >> >> 2)X-u-v=0;
>> >> >> >> 3)X-v=u
>> >> >> >> 4)X=u+v
>> >> >> >>
>> >> >> >> solution X=?
>> >> >> >> Please do mention based on which definition of IEEE-stds 1788
>> >> >> >> the
>> >> >> >> presented problem has a unique solution.
>> >> >> >> Thank you so much for your kind comments in advance.
>> >> >> >> Warmest regards,
>> >> >> >> Mehran
>> >> >> >> --
>> >> >> >> Mehran Mazandarani
>> >> >> >> Department of Electrical Engineering
>> >> >> >> Ferdowsi University of Mashhad, Mashhad, Iran.
>> >> >> >> homepage:http://mehran.mazandarani.fumblog.um.ac.ir/
>> >> >> >> IEEE Member, me.mazandarani@xxxxxxxx
>> >> >> >>
>> >> >> >
>> >> >>
>> >> >>
>> >> >> --
>> >> >> Mehran Mazandarani
>> >> >> Department of Electrical Engineering
>> >> >> Ferdowsi University of Mashhad, Mashhad, Iran.
>> >> >> homepage:http://mehran.mazandarani.fumblog.um.ac.ir/
>> >> >> IEEE Member, me.mazandarani@xxxxxxxx
>> >> >>
>> >> >
>> >>
>> >>
>> >> --
>> >> Mehran Mazandarani
>> >> Department of Electrical Engineering
>> >> Ferdowsi University of Mashhad, Mashhad, Iran.
>> >> homepage:http://mehran.mazandarani.fumblog.um.ac.ir/
>> >> IEEE Member, me.mazandarani@xxxxxxxx
>> >>
>> >
>>
>>
>> --
>> Mehran Mazandarani
>> Department of Electrical Engineering
>> Ferdowsi University of Mashhad, Mashhad, Iran.
>> homepage:http://mehran.mazandarani.fumblog.um.ac.ir/
>> IEEE Member, me.mazandarani@xxxxxxxx
>>
>
--
Mehran Mazandarani
Department of Electrical Engineering
Ferdowsi University of Mashhad, Mashhad, Iran.
homepage:http://mehran.mazandarani.fumblog.um.ac.ir/
IEEE Member, me.mazandarani@xxxxxxxx