Re: Next generation COBOL?

2005-11-27 01:22:10 MET

Oops - past midnight, and I ought to make another
posting to c.l.c ...

Steve Richfie1d wrote:

Herwig,

I attended a talk and had some discussions afterwards with a group who was working on a quantum computer for the NSA. The question came up: Immediate problems aside, what COULD quantum computers potentially do for us better than conventional computers?

The only apparent answer seemed to be factoring large numbers for crypto applications, which rather limits the number of potential sales to the number of countries. While there are a LOT of other conceivable applications, they all fail because either the setup would be impossibly difficult or because conventional computers are fast enough, or there is some other computational technology (e.g. analog) that would work well. 

Fact is, quantum computers were amongs us for eons. We
have a simple device which aids the photon to compute
a superposition of all it's possible ways, thus
emerging nothing else as the laws of ordinary imaging
optics. The device is called "eye", and most of us
have two of them.

You know the way game programmers compute those pictures
visible on the screen? There are Raytracing algorithms
and all that stuff - and you have still deficits,
compared with reality. The reason is, nature does not
do raytracing. She does Quantum Computing - with each
photon.

That hints the way to go. Quantum Computing is for
simulations. Also, the concept of a Bit and Quantum
Computing do not merge very well - after all, an
observable of a physical system oftenly has more
than two values. The QuBit just is the simplest quantum
system to study, that's all (the unitary evolution
of a single QuBit is a complex-valued 2*2 matrix,
the isolated QuBit itsef dwells in a Hilbert Space
of *only* 4 dimensions.).

I expect that further research will reveal quantum
algorithms which do not deal (only) with QuBits.

Even David Deutsch himself changes his mind about
the future now and then. Recently, he anounced
that "Fully fledged quantum computation is within
sight" - he wrote:

   I have recently changed my mind about the time scale
   for achieving fully fledged quantum computation. On
   my blog, I have stuck my neck out and suggested that
   it might be less than a decade away.

http://www.qubit.org/people/david/index.php?blog=20050830143405

   Cluster quantum computation is the reason.

The article he mentions is

http://xxx.lanl.gov/abs/quant-ph/0508218

Be careful: contains math in quantities maybe degrading
your and my health (I have not yet fully digested it).

However, perhaps you know some way past this logjam to be able to construct practical future quantum computers to solve real-world problems? 

I don't. Except, of course, the old age quantum fund, a
concept for which I hold the rights in Germany - that
is certainly a "real world problem", but, to my mind,
ethically unaceptable. Also, I would not call it "computing".

It sounds like you missed the SPEED in speed read above. Sure, once you are concerned with details of functionality then COBOL mires down, but it is sure fast as finding the place where you need to drill down. There is a LOT of brain that has been trained to read English fast, so why not use it?! 

Technical stuff I can read speedier in formulas
rather that in English or German. Commercial stuff - well,
I do not like commercial stuff. I do not understand my
own tax declaration.

My Ada-experience, however, is 10 years old, so there is some
loss in the virtuosity with that language.


ADA is far superior to C for writing reliable code. I wrote the ADA linker for ROLM Corp's ADA compiler. 

I was ten years in the European Ada Compiler Project
- from inception to it's grave. Because that compiler
was written in Ada, I was, then, more fluent in Ada
than in any other language - which is to be expected,
I guess.

Quite so. Worse: The same language means different thing to
different persons. All those of us who are/were married know that.

A problem not (yet?) shared by programming languages.


This of course is the essence of conversion and porting, where the same statements often mean different things to different computers. 

And this is one reason for the invention of
programming languages. And it is one reason
of lengthy discussions about the meaning of
a Standard for programming languages. I have
seen that in Ada and, recently, in OO-COBOL.

BTW: It was and is the same as in a marriage:
He who has the loudest voice or collects the
most allies in this newsgroup decides how the
standard is to be interpreted. The compiler,
once implemented, has his own mind also.

Programming Languages for quantum Computers, however,
look very different from programming languages for
classical computation.


Has someone proposed such a language? That WOULD be interesting. 

Yes. Quite some time ago I found an austrian
scientist who had published an open source
programming language for quantum computers.
I have still an installation right here on my
computer. It is called QCL (Quantum computing
Language) by Bernhard Oemer <oemer@xxxxxxxxxxxxxxxx>
I copy in here the URLs from the README-file:

  http://tph.tuwien.ac.at/~oemer/doc/qcldoc/index.html
  http://tph.tuwien.ac.at/~oemer/doc/qcldoc/qcldoc.ps

  http://tph.tuwien.ac.at/~oemer/doc/quprog/index.html
  http://tph.tuwien.ac.at/~oemer/doc/quprog/qcldoc.ps

I do not know whether these URLs are still correct -
the time stamp of the files is 2001-01-17 ...

What he did is creating a formalism with which to set up
a few QuBits, initialize their amplitude and chase them
throuch a network of various Quantum Gates. In the end,
the QuBits are measured, and an ordinary randum number
generator is used to emulate the behaviour of a quantum
measurement.

Because the Quantum Computation is emulated with
classical computation throughout, you end up with
a lot of amplitudes - even with only 30 QuBits,
you can run out of memory easily, and you need
much computing time - a real QC would not
suffer from that.

His (and my) answer is the Many Worlds Interpretation of
Quantum Mechanics


In real-world simulations, it often boils down to the one-dimensional numbers REALLY being approximate coordinates in a multi-dimensional space that is long and narrow, with the number being the position along the major axis. Only when you consider that those approximate numbers are really hyperspheres of uncertainty in he multidimensional space can you determine the potential extend of the uncertainty in your computations. 

Uncertainity in Quantum Physics does not result
from imprecise measurements, and, consequentially,
the inability to do exact simulations with inexact
boundary conditions.

Uncertainty in Quantum Physics results from the
fact that the building blocks of reality are not
particles, are not precise positions and imulses,
are not facts. The building block of reality is
nothing that we know in everyday life.

The building block of reality is the quantum state.
"Asking" a quantum state for a non-quantum quantity
such as position yields the known behaviour of the
uncertanity principle, the emergence of probability,
and the emergence of many (classical) reality
branches. All these are artefacts of the fact that
our minds (as well as our computers) do not work
with quantum concepts but with the condensed
concept of "fact" - which we call a bit.

BTW: even time may be an emergent phenomena. That
has far reaching consequences which I will not
discuss here - for one thing because it's very late.

In short, no you cannot predict the EXACT future, but you CAN often predict the future along with an indication of accuracy, that obviously deteriorates the further out you extrapolate. The problem with current computers is that they do NOT automatically determine the uncertainty when it is entirely possible/practical to do so. The reason that things are now as they are is because the wrong people were in the IEEE-754 standards committee. This was a "volunteer" organization, meaning that only those who were paid by special interests to promote EXISTING standards participated, thereby guaranteeing our present mess. 

As I say, it makes no sense predicting a future of which
are many. For any prediction which is physically
possible there is a future which you enter with a
certain probability.

But in the limits of predictable processes - celestial
mechanics, for instance, as a prime and shining example,
you are right indeed. Probably intervall arithmetic
might be useful, or adaptive precission, or what have
you - I do not follow that area very closely. I don't
see why the IEEE standard for floating point calculations
should look exactly the way it does.

... 

There is *no* way to predict a future - because there is
always a bunch of futures!

We can be pretty sure that tomorrow's temp will be below 100C - but how much below? The challenge is NOT in predicting the future, but rather in precisely stating how accurate your prediction is. 

Quite so. And there *is* a very small probability that
tomorrow's temperature will be *above* 100C. The core of
the sun might detonate, for instance.

Actually, my employer pays me for thinking about COBOL and related
subjects.

This DOES sound interesting. Perhaps there is some overlap in our motivations? 

I don't know - does looking for a better paid job
qualify sufficiently as motivation?

Forgive me any imprecision in my wording - it is
really late.

Herwig

--
**********************************************************
*  http://www.quantenrente.de    Josella Simone Playton  *
*  http://www.Josella-Simone-Playton.de  Herwig Huener   *
**********************************************************

.

Relevant Pages

  • Rasetti: Quantum Computers based on TQFT
    ... QC can solve problems which are not in P in polynomial time. ... So the idea is that there may be more types of quantum computers than ... in as much we want to regard analog computers as computers when it ...
    (sci.physics.research)
  • Re: Rasetti: Quantum Computers based on TQFT
    ... > new and more powerful types of quantum computers. ... > polynomials is not in P. ... Analog computers tend not to work on hard problems, or when they do, ...
    (sci.physics.research)
  • Re: Incompleteness vs. Mechanical Reasoning
    ... perfectly possible future computers have free will in this sense. ... to assert that a computer's future action (e.g. will it halt or not ... NAFL, in which I have formulated my definition of free will. ... are quantum computers as defined via the NAFL model of computation ...
    (sci.logic)
  • Re: *Quantum Computing* expert Bill Munro
    ... these researchers called on scientists around the world to ... ]develop ciphers resistant to attack by quantum computers. ... Advances in mathematics can always destroy cyphers-- public or private ...
    (sci.crypt)
  • Re: consciousness
    ... computers are faced with unexpected conditions all the time. ... great deal of effort in most modern programming languages is to deal ... I suspect you are correct, which is why I suspect real AI ... an emergent property of complexity. ...
    (talk.origins)