Researchers show classical computers can keep up with, and surpass, their quantum counterparts
https://www.sciencedaily.com/releases/2024/02/240209134402.htm
Researchers adopt innovative method to boost speed and accuracy of traditional computing

On 12/02/2024 05:26, Jan Panteltje wrote:

> Researchers show classical computers can keep up with, and surpass, their quantum counterparts
> https://www.sciencedaily.com/releases/2024/02/240209134402.htm
> Researchers adopt innovative method to boost speed and accuracy of traditional computing

It is a bold claim but not backed up by any convincing evidence.

I can believe that classical computing and in particular NN type AI can
be speeded up by making some gross heuristic approximations that are
usually true. Ignoring almost irrelevant noisy information may work.

Nothing can surpass an N bit quantum computer for factoring products of
impossibly long primes. The whole of modern public key cryptography is
predicated on that task being well beyond present day computing power. A
decent length quantum register computer could change that overnight.

Dedicated hardware can always do better than general purpose computers
at specific tasks but that is a different issue altogether.

Turing's Bombe or Collosus would have beaten anything less than a 386 PC
at code breaking despite them having plug boards, paper tape, relays and
valve logic. They were incredibly cunning designs able to short circuit
the codebreaking by ruling out big chunks of the search space.

--
Martin Brown

On 2/12/24 11:24, Martin Brown wrote:
> On 12/02/2024 05:26, Jan Panteltje wrote:
>
>> Researchers show classical computers can keep up with, and surpass,
>> their quantum counterparts
>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>    Researchers adopt innovative method to boost speed and accuracy of
>> traditional computing
>
> It is a bold claim but not backed up by any convincing evidence.
>
> I can believe that classical computing and in particular NN type AI can
> be speeded up by making some gross heuristic approximations that are
> usually true. Ignoring almost irrelevant noisy information may work.
>
> Nothing can surpass an N bit quantum computer for factoring products of
> impossibly long primes. [snip...]
Entirely hypothetical. My expectation is that quantum computers
will never be able to factor numbers with prime factors much beyond
10^18 or so, if even that.

Jeroen Belleman

On 12/02/2024 10:41, Jeroen Belleman wrote:
> On 2/12/24 11:24, Martin Brown wrote:
>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>
>>> Researchers show classical computers can keep up with, and surpass,
>>> their quantum counterparts
>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>    Researchers adopt innovative method to boost speed and accuracy of
>>> traditional computing
>>
>> It is a bold claim but not backed up by any convincing evidence.
>>
>> I can believe that classical computing and in particular NN type AI
>> can be speeded up by making some gross heuristic approximations that
>> are usually true. Ignoring almost irrelevant noisy information may work.
>>
>> Nothing can surpass an N bit quantum computer for factoring products
>> of impossibly long primes. [snip...]
> Entirely hypothetical. My expectation is that quantum computers
> will never be able to factor numbers with prime factors much beyond
> 10^18 or so, if even that.

I remain unconvinced that quantum computers can be made reliable enough
to do anything remotely useful in the real world. OTOH they have been
making real progress and today's mobile phones look like magic compared
to the big iron of yesteryear. My first mainframe IBM 370/165 had a
whopping 4MB of main memory and you had to get a special ticket to use
more than 500k at once. Algebra systems wouldn't run in less than 2MB.

--
Martin Brown

On 2/12/2024 4:18 AM, Martin Brown wrote:
> On 12/02/2024 10:41, Jeroen Belleman wrote:
>> On 2/12/24 11:24, Martin Brown wrote:
>>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>>
>>>> Researchers show classical computers can keep up with, and surpass, their
>>>> quantum counterparts
>>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>>    Researchers adopt innovative method to boost speed and accuracy of
>>>> traditional computing
>>>
>>> It is a bold claim but not backed up by any convincing evidence.
>>>
>>> I can believe that classical computing and in particular NN type AI can be
>>> speeded up by making some gross heuristic approximations that are usually
>>> true. Ignoring almost irrelevant noisy information may work.
>>>
>>> Nothing can surpass an N bit quantum computer for factoring products of
>>> impossibly long primes. [snip...]
>> Entirely hypothetical. My expectation is that quantum computers
>> will never be able to factor numbers with prime factors much beyond
>> 10^18 or so, if even that.
>
> I remain unconvinced that quantum computers can be made reliable enough to do
> anything remotely useful in the real world. OTOH they have been making real
> progress and today's mobile phones look like magic compared to the big iron of
> yesteryear. My first mainframe IBM 370/165 had a whopping 4MB of main memory
> and you had to get a special ticket to use more than 500k at once. Algebra
> systems wouldn't run in less than 2MB.

Predicting that something "can't" be done/happen is usually folly.
Predicting that something WON'T (volition) be done is a safer bet.

My first AI course (mid 70's) predicted it would be "about 10 years"
for AI to be practical. That prediction was repeated "about every
10 years" and came to be a bit of a standing joke.

Now, almost "suddenly", it's "here" (in a particular form) and
folks are unprepared for it (both to exploit it and safeguard
against it).

Not only is it available but it is also ACCESSIBLE (early predictions
treated it as an "ivory tower" sort of technology; clearly not something
that Joe Average User could access!)

On 2/12/24 18:34, Don Y wrote:
> On 2/12/2024 4:18 AM, Martin Brown wrote:
>> On 12/02/2024 10:41, Jeroen Belleman wrote:
>>> On 2/12/24 11:24, Martin Brown wrote:
>>>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>>>
>>>>> Researchers show classical computers can keep up with, and surpass,
>>>>> their quantum counterparts
>>>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>>>    Researchers adopt innovative method to boost speed and accuracy
>>>>> of traditional computing
>>>>
>>>> It is a bold claim but not backed up by any convincing evidence.
>>>>
>>>> I can believe that classical computing and in particular NN type AI
>>>> can be speeded up by making some gross heuristic approximations that
>>>> are usually true. Ignoring almost irrelevant noisy information may
>>>> work.
>>>>
>>>> Nothing can surpass an N bit quantum computer for factoring products
>>>> of impossibly long primes. [snip...]
>>> Entirely hypothetical. My expectation is that quantum computers
>>> will never be able to factor numbers with prime factors much beyond
>>> 10^18 or so, if even that.
>>
>> I remain unconvinced that quantum computers can be made reliable
>> enough to do anything remotely useful in the real world. OTOH they
>> have been making real progress and today's mobile phones look like
>> magic compared to the big iron of yesteryear. My first mainframe IBM
>> 370/165 had a whopping 4MB of main memory and you had to get a special
>> ticket to use more than 500k at once. Algebra systems wouldn't run in
>> less than 2MB.
>
> Predicting that something "can't" be done/happen is usually folly.
> Predicting that something WON'T (volition) be done is a safer bet.
>
[...]

What will probably happen is that quantum computing will fail
to live up to expectations and at some point, funding will run
dry.

Jeroen Belleman

On 2/12/2024 12:26 PM, Jeroen Belleman wrote:
> On 2/12/24 18:34, Don Y wrote:
>> On 2/12/2024 4:18 AM, Martin Brown wrote:
>>> On 12/02/2024 10:41, Jeroen Belleman wrote:
>>>> On 2/12/24 11:24, Martin Brown wrote:
>>>>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>>>>
>>>>>> Researchers show classical computers can keep up with, and surpass, their
>>>>>> quantum counterparts
>>>>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>>>>    Researchers adopt innovative method to boost speed and accuracy of
>>>>>> traditional computing
>>>>>
>>>>> It is a bold claim but not backed up by any convincing evidence.
>>>>>
>>>>> I can believe that classical computing and in particular NN type AI can be
>>>>> speeded up by making some gross heuristic approximations that are usually
>>>>> true. Ignoring almost irrelevant noisy information may work.
>>>>>
>>>>> Nothing can surpass an N bit quantum computer for factoring products of
>>>>> impossibly long primes. [snip...]
>>>> Entirely hypothetical. My expectation is that quantum computers
>>>> will never be able to factor numbers with prime factors much beyond
>>>> 10^18 or so, if even that.
>>>
>>> I remain unconvinced that quantum computers can be made reliable enough to
>>> do anything remotely useful in the real world. OTOH they have been making
>>> real progress and today's mobile phones look like magic compared to the big
>>> iron of yesteryear. My first mainframe IBM 370/165 had a whopping 4MB of
>>> main memory and you had to get a special ticket to use more than 500k at
>>> once. Algebra systems wouldn't run in less than 2MB.
>>
>> Predicting that something "can't" be done/happen is usually folly.
>> Predicting that something WON'T (volition) be done is a safer bet.
>>
> [...]
>
> What will probably happen is that quantum computing will fail
> to live up to expectations and at some point, funding will run
> dry.

Possible. Or, some breakthrough will provide a way around
existing problems. Or, some OTHER technology will offer the
same promise with a different approach.

As long as there is the potential for a "desirable outcome",
folks will explore options in "that direction". (fusion
has been 10 years off for about as long as AI!)

[E.g., bubble memory was supposed to be a great, compact
nonvolatile memory... until solid state technologies
tipped the balance]

On a sunny day (Mon, 12 Feb 2024 10:34:52 -0700) it happened Don Y
<blockedofcourse@foo.invalid> wrote in <uqdkrv$1kq04$1@dont-email.me>:

>On 2/12/2024 4:18 AM, Martin Brown wrote:
>> On 12/02/2024 10:41, Jeroen Belleman wrote:
>>> On 2/12/24 11:24, Martin Brown wrote:
>>>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>>>
>>>>> Researchers show classical computers can keep up with, and surpass, their
>>>>> quantum counterparts
>>>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>>>    Researchers adopt innovative method to boost speed and accuracy of
>>>>> traditional computing
>>>>
>>>> It is a bold claim but not backed up by any convincing evidence.
>>>>
>>>> I can believe that classical computing and in particular NN type AI can be
>>>> speeded up by making some gross heuristic approximations that are usually
>>>> true. Ignoring almost irrelevant noisy information may work.
>>>>
>>>> Nothing can surpass an N bit quantum computer for factoring products of
>>>> impossibly long primes. [snip...]
>>> Entirely hypothetical. My expectation is that quantum computers
>>> will never be able to factor numbers with prime factors much beyond
>>> 10^18 or so, if even that.
>>
>> I remain unconvinced that quantum computers can be made reliable enough to do
>> anything remotely useful in the real world. OTOH they have been making real
>> progress and today's mobile phones look like magic compared to the big iron of
>> yesteryear. My first mainframe IBM 370/165 had a whopping 4MB of main memory
>> and you had to get a special ticket to use more than 500k at once. Algebra
>> systems wouldn't run in less than 2MB.
>
>Predicting that something "can't" be done/happen is usually folly.
>Predicting that something WON'T (volition) be done is a safer bet.
>
>My first AI course (mid 70's) predicted it would be "about 10 years"
>for AI to be practical. That prediction was repeated "about every
>10 years" and came to be a bit of a standing joke.
>
>Now, almost "suddenly", it's "here" (in a particular form) and
>folks are unprepared for it (both to exploit it and safeguard
>against it).
>
>Not only is it available but it is also ACCESSIBLE (early predictions
>treated it as an "ivory tower" sort of technology; clearly not something
>that Joe Average User could access!)

Now <when> would this breakthrough happen for fusion power?

On Tuesday, February 13, 2024 at 5:06:46 PM UTC+11, Jan Panteltje wrote:
> On a sunny day (Mon, 12 Feb 2024 10:34:52 -0700) it happened Don Y
> <blocked...@foo.invalid> wrote in <uqdkrv$1kq04$1...@dont-email.me>:
> >On 2/12/2024 4:18 AM, Martin Brown wrote:
> >> On 12/02/2024 10:41, Jeroen Belleman wrote:
> >>> On 2/12/24 11:24, Martin Brown wrote:
> >>>> On 12/02/2024 05:26, Jan Panteltje wrote:
> >>>>
> >>>>> Researchers show classical computers can keep up with, and surpass, their
> >>>>> quantum counterparts
> >>>>> https://www.sciencedaily.com/releases/2024/02/240209134402.htm
> >>>>> Researchers adopt innovative method to boost speed and accuracy of
> >>>>> traditional computing
> >>>>
> >>>> It is a bold claim but not backed up by any convincing evidence.
> >>>>
> >>>> I can believe that classical computing and in particular NN type AI can be
> >>>> speeded up by making some gross heuristic approximations that are usually
> >>>> true. Ignoring almost irrelevant noisy information may work.
> >>>>
> >>>> Nothing can surpass an N bit quantum computer for factoring products of
> >>>> impossibly long primes. [snip...]
> >>> Entirely hypothetical. My expectation is that quantum computers
> >>> will never be able to factor numbers with prime factors much beyond
> >>> 10^18 or so, if even that.
> >>
> >> I remain unconvinced that quantum computers can be made reliable enough to do
> >> anything remotely useful in the real world. OTOH they have been making real
> >> progress and today's mobile phones look like magic compared to the big iron of
> >> yesteryear. My first mainframe IBM 370/165 had a whopping 4MB of main memory
> >> and you had to get a special ticket to use more than 500k at once. Algebra
> >> systems wouldn't run in less than 2MB.
> >
> >Predicting that something "can't" be done/happen is usually folly.
> >Predicting that something WON'T (volition) be done is a safer bet.
> >
> >My first AI course (mid 70's) predicted it would be "about 10 years"
> >for AI to be practical. That prediction was repeated "about every
> >10 years" and came to be a bit of a standing joke.
> >
> >Now, almost "suddenly", it's "here" (in a particular form) and
> >folks are unprepared for it (both to exploit it and safeguard
> >against it).
> >
> >Not only is it available but it is also ACCESSIBLE (early predictions
> >treated it as an "ivory tower" sort of technology; clearly not something
> >that Joe Average User could access!)
>
> Now <when> would this breakthrough happen for fusion power?

It may have already happened. The Joint European Torus is a protoptype for a more or less practical fusion fueled generator, but there are private companies working on smaller machines which they think can be made to work rather sooner.

https://hb11.energy/

is one of them. They propose to fuse boron-11 and hydrogen which is a process that doesn't generate neutrons, which is helpful.

https://en.wikipedia.org/wiki/Aneutronic_fusion

--
Bill Sloman, Sydney

On 12/02/2024 19:26, Jeroen Belleman wrote:
> On 2/12/24 18:34, Don Y wrote:
>> On 2/12/2024 4:18 AM, Martin Brown wrote:
>>> On 12/02/2024 10:41, Jeroen Belleman wrote:
>>>> On 2/12/24 11:24, Martin Brown wrote:
>>>>> On 12/02/2024 05:26, Jan Panteltje wrote:
>>>>>
>>>>>> Researchers show classical computers can keep up with, and
>>>>>> surpass, their quantum counterparts
>>>>>>   https://www.sciencedaily.com/releases/2024/02/240209134402.htm
>>>>>>    Researchers adopt innovative method to boost speed and accuracy
>>>>>> of traditional computing
>>>>>
>>>>> It is a bold claim but not backed up by any convincing evidence.
>>>>>
>>>>> I can believe that classical computing and in particular NN type AI
>>>>> can be speeded up by making some gross heuristic approximations
>>>>> that are usually true. Ignoring almost irrelevant noisy information
>>>>> may work.
>>>>>
>>>>> Nothing can surpass an N bit quantum computer for factoring
>>>>> products of impossibly long primes. [snip...]
>>>> Entirely hypothetical. My expectation is that quantum computers
>>>> will never be able to factor numbers with prime factors much beyond
>>>> 10^18 or so, if even that.
>>>
>>> I remain unconvinced that quantum computers can be made reliable
>>> enough to do anything remotely useful in the real world. OTOH they
>>> have been making real progress and today's mobile phones look like
>>> magic compared to the big iron of yesteryear. My first mainframe IBM
>>> 370/165 had a whopping 4MB of main memory and you had to get a
>>> special ticket to use more than 500k at once. Algebra systems
>>> wouldn't run in less than 2MB.
>>
>> Predicting that something "can't" be done/happen is usually folly.
>> Predicting that something WON'T (volition) be done is a safer bet.
>>
> [...]
>
> What will probably happen is that quantum computing will fail
> to live up to expectations and at some point, funding will run
> dry.

A bit like AI in the late 70's then. Draughts(checkers) fell to it
almost immediately - chess and machine vision were much tougher nuts to
crack. Things in AI research only really hotted up again after Deep Blue
beat Kasparov and then again later when Alpha-Go beat the human world
champion at Go under match conditions.

The latter was a feat that no-one in the field really expected to come
so quickly. Not far behind that were the large language models.

The next iteration of quantum computing when it is next in fashion again
might well get over the line to being truly useful (but I'm not holding
my breath).

It's a bit like cheap fusion power - remaining elusively about 50 year
away and has done now for over 50 years.

--
Martin Brown