I was bored and found a HP-27S

[Don Shepherd]

(06-11-2014 01:34 AM)rprosperi Wrote:  the source code makes even hardened RPL authors do a double-take and mutter "what the $*#& is that?!"

It's funny, but "what the $*#& is that?!" has always been my reaction when I have viewed RPL code. I admire what it can do, and I admire those who know and use it, but I've always appreciated the simplicity of good old RPN.

I've posted too many challenges or puzzles over the years, and people like Gerson (and many other smart people whose names escape me at the moment) have posted RPL solutions like "< dup dup drop dup 4 / >" will give you a list of all the atoms in the observable universe. OK, I'm exaggerating a little.

[Dave Britten]

(06-11-2014 02:13 AM)Don Shepherd Wrote:  I've posted too many challenges or puzzles over the years, and people like Gerson (and many other smart people whose names escape me at the moment) have posted RPL solutions like "< dup dup drop dup 4 / >" will give you a list of all the atoms in the observable universe. OK, I'm exaggerating a little.

<< OVER DUP DROP DUP 4 / >>, but yeah, basically.

[Thomas Radtke]

(06-10-2014 07:19 PM)Katie Wasserman Wrote:  

(06-10-2014 04:06 PM)Thomas Radtke Wrote:  It is programmable, the solver is just not turing complete. Calculators like the 20S play in the same league, allthough not as obvious.

"not Turing Complete", I think not!

I assumed some kind of array would be required, but I was wrong (at least on the 20S), sorry.

Not being an expert with the solver, I tried to construct an infinite loop this morning:

(1) A=sigma(I:1:10:0:I) gives 'solution not found'

(2) A=sigma(I:1:10:1:L(I:I-1)) gives 45

In (2), it appears, 'I' cannot be decremented and L(I:I-1) evaluates to I-1. Further investigating this:

(3) A=sigma(I:1:10:1:L(I:I-1)+I) gives 100 (apparently 45+55)

Sorry again if I've just shown my ignorance ;-/.

(feel free to fork this into another thread if it's of any interest)

[Dave Britten]

(06-11-2014 08:04 AM)Thomas Radtke Wrote:  I assumed some kind of array would be required, but I was wrong (at least on the 20S), sorry.

Not being an expert with the solver, I tried to construct an infinite loop this morning:

(1) A=sigma(I:1:10:0:I) gives 'solution not found'

(2) A=sigma(I:1:10:1:L(I:I-1)) gives 45

In (2), it appears, 'I' cannot be decremented and L(I:I-1) evaluates to I-1. Further investigating this:

(3) A=sigma(I:1:10:1:L(I:I-1)+I) gives 100 (apparently 45+55)

Sorry again if I've just shown my ignorance ;-/.

(feel free to fork this into another thread if it's of any interest)

The final paragraph on page 25 of the Technical Applications book mentioned earlier explains that the loop counter, start, and end values are copied to separate memory locations that can't be modified directly once the loop starts, so I think your conclusion that an infinite loop is impossible is probably correct. (Unless there's some other mechanism waiting to be abused. Forcing an iterative solution and constantly modifying the variable being solved for with L(), perhaps?)

[Thomas Klemm]

(06-10-2014 04:06 PM)Thomas Radtke Wrote:  It is programmable, the solver is just not turing complete.

We've already discussed that topic:
Message #41 Posted by x34 on 1 Feb 2011, 10:28 a.m.

Cheers
Thomas

[Thomas Radtke]

(06-11-2014 11:21 AM)Thomas Klemm Wrote:  We've already discussed that topic:
Message #41 Posted by x34 on 1 Feb 2011, 10:28 a.m.

Thank you, Thomas. I'm always impressed by your knowledge, btw.

I'm under the impression, the Turing-completeness of the solver hasn't been proven in this thread, but I might very well be wrong. Most of the contributions are far above my head ;-).

A note to Walter: Please make sure the internal variables in question can be user-modified in the WP43-solver. That shouldn't break compatibility to existing programs but offer some new methods :-).

[Don Shepherd]

(06-11-2014 10:50 AM)Dave Britten Wrote:  Unless there's some other mechanism waiting to be abused.

The ultimate HP-17b infinite loop:

Code:


A=SIGMA(I:-9.99999999999E499:9.99999999999E499:1E-499:0)

It won't let you do sigma(I:1:2:-1:0), but I imagine the example above would exhaust even the strongest set of batteries.

[Dave Britten]

(06-11-2014 12:18 PM)Don Shepherd Wrote:  

(06-11-2014 10:50 AM)Dave Britten Wrote:  Unless there's some other mechanism waiting to be abused.

The ultimate HP-17b infinite loop:

Code:


A=SIGMA(I:-9.99999999999E499:9.99999999999E499:1E-499:0)

It won't let you do sigma(I:1:2:-1:0), but I imagine the example above would exhaust even the strongest set of batteries.

Yeah, I imagine that would do the trick for all practical purposes.

[Thomas Klemm]

(06-11-2014 12:28 PM)Dave Britten Wrote:  Yeah, I imagine that would do the trick for all practical purposes.

Or else you can nest loops.

[Thomas Klemm]

(06-11-2014 11:55 AM)Thomas Radtke Wrote:  I'm under the impression, the Turing-completeness of the solver hasn't been proven in this thread, but I might very well be wrong.

You should have a look at Message #60. But be warned: it's a "vulgar proof of Turing completeness".

Cheers
Thomas

[rprosperi]

(06-11-2014 02:13 AM)Don Shepherd Wrote:  

(06-11-2014 01:34 AM)rprosperi Wrote:  the source code makes even hardened RPL authors do a double-take and mutter "what the $*#& is that?!"

It's funny, but "what the $*#& is that?!" has always been my reaction when I have viewed RPL code. I admire what it can do, and I admire those who know and use it, but I've always appreciated the simplicity of good old RPN.

I've posted too many challenges or puzzles over the years, and people like Gerson (and many other smart people whose names escape me at the moment) have posted RPL solutions like "< dup dup drop dup 4 / >" will give you a list of all the atoms in the observable universe. OK, I'm exaggerating a little.

Damn you Don! There is now coffee spit all over my desk and keyboard. I need to learn to be more careful when reading posts and drinking.

[Thomas Radtke]

(06-11-2014 01:08 PM)Thomas Klemm Wrote:  You should have a look at Message #60. But be warned: it's a "vulgar proof of Turing completeness".

I had:

Quote:Our goal is SUFFICIENT completeness, with finite number of steps and memory locations.

I'm still missing either the possibility of an infinite run within the solver or its exclusion in the above definition of a "vulgar proof". I can predict that any program eventually stops, and not only because batteries are empty or the hardware dies. I don't think the definition of a *sufficient*(1) turing completeness is met here.

I feel completely silly to argue with two experts of this forum :-D.

(1) Edited to insert this word

Edit 2: Yes, before someone mentions this - it might not be of any practical consequence, but that remains to show. I think you were about the same issue.

[Dave Britten]

(06-11-2014 02:34 PM)Thomas Radtke Wrote:  I'm still missing either the possibility of an infinite run within the solver or its exclusion in the above definition of a "vulgar proof". I can predict that any program eventually stops, and not only because batteries are empty or the hardware dies. I don't think the definition of a *sufficient*(1) turing completeness is met here.

I think it just depends on if you're talking computer science or computer engineering.

I'm the former, but since Turing completeness also has to assume essentially unlimited working memory and run time, I'm willing to make this affordance as well.

Horse shoes, hand grenades, and Turing completeness.

[Thomas Radtke]

(06-11-2014 03:03 PM)Dave Britten Wrote:  I'm the former, but since Turing completeness also has to assume essentially unlimited working memory and run time, I'm willing to make this affordance as well.

I accept and rest my case.

The finite loop thing was an attempt not to lose my entire case to Katie anyway. Guess she derserves to win this time ;-).

[Katie Wasserman]

(06-11-2014 03:19 PM)Thomas Radtke Wrote:  The finite loop thing was an attempt not to lose my entire case to Katie anyway. Guess she derserves to win this time ;-).

I'm doing a little victory dance -- a very little one -- actually, it's just my fingers dancing across the keyboard

[Don Shepherd]

(06-11-2014 01:22 PM)rprosperi Wrote:  

(06-11-2014 02:13 AM)Don Shepherd Wrote:  It's funny, but "what the $*#& is that?!" has always been my reaction when I have viewed RPL code. I admire what it can do, and I admire those who know and use it, but I've always appreciated the simplicity of good old RPN.

I've posted too many challenges or puzzles over the years, and people like Gerson (and many other smart people whose names escape me at the moment) have posted RPL solutions like "< dup dup drop dup 4 / >" will give you a list of all the atoms in the observable universe. OK, I'm exaggerating a little.

Damn you Don! There is now coffee spit all over my desk and keyboard. I need to learn to be more careful when reading posts and drinking.

Bob, you've been looking for an excuse to buy a new PC, now you have it! Just convince your wife that a coffee-laden keyboard requires a whole new system.

[jebem]

I'm back home (Lisbon district) from my travel from Africa, and had a package waiting for me at the Post Office.

Today I got my new acquisition: a HP-27S in a better condition than I was initially expecting.
Actually the photos enhance all the small imperfections and dust.
Made in Singapure, s/n:3122S04492 (Year 1991, week 22).

It is a fantastic good looking machine, excellent keyboard.
Even the display is very readable and acceptable contrast, considering that the included battery cells are on the weak side
(just 4.16Volt with the calculator powered on).

Thank you all for the above comments on this thread.
Meanwhile I'm discovering the wonderful world of the Pioneer machines.
A HP-42S almost new in the box is waiting for me to collect it from the original owner.
He has bought it in South Africa but he never really used it
(well, that's the story anyway, and I trust people, so I expect this to be true).

I'm publishing here some links to my HP-27S.




Clean battery compartment. IR LED for printing looks good as well.





Power On message. I can tell the same in a bad day...





Some shots of the mandatory Self Test sequence. Pressing CRL + LN...
SPD stands for Speed, right? "1050320" ...
This calculator sports a precise timer to maintain the internal Date and Time.
So it needs some sort of precise timing.











URAM Test... what is it?







Full PI precision in the HP universe.





Almost 7 KByte of memory!




http://3.bp.blogspot.com/-HRqhnwBVhmM/U6...M_7769.JPG

[rprosperi]

(06-25-2014 06:49 PM)jebem Wrote:  I'm back home (Lisbon district) from my travel from Africa, and had a package waiting for me at the Post Office.

Today I got my new acquisition: a HP-27S in a better condition than I was initially expecting.
Actually the photos enhance all the small imperfections and dust.
Made in Singapure, s/n:3122S04492 (Year 1991, week 22).

Nice machine Jebem, I think you will enjoy it, despite it's Algebraic nature.

Pull up a copy of HP-27S/19B Technical Applications (and of course the 27S UG) from the Museum DVD and learn the Solver.

Then search the Forum (OLD & NEW) for great articles from Don Shepherd and then, when you think you have it mastered, search for an article w/title something like "Long Equation for HP17B Solver" from Gerson IIRC. Astounding stuff, implementing the Trig functions and much more.

Keep in mind the Solver is the same for the 19B/19BII, 27S, 17B and 17BII, so you have to widen your search to get all the interesting Solver articles.

[Don Shepherd]

(06-26-2014 01:23 PM)rprosperi Wrote:  

(06-25-2014 06:49 PM)jebem Wrote:  I'm back home (Lisbon district) from my travel from Africa, and had a package waiting for me at the Post Office.

Today I got my new acquisition: a HP-27S in a better condition than I was initially expecting.
Actually the photos enhance all the small imperfections and dust.
Made in Singapure, s/n:3122S04492 (Year 1991, week 22).

Nice machine Jebem, I think you will enjoy it, despite it's Algebraic nature.

Pull up a copy of HP-27S/19B Technical Applications (and of course the 27S UG) from the Museum DVD and learn the Solver.

Then search the Forum (OLD & NEW) for great articles from Don Shepherd and then, when you think you have it mastered, search for an article w/title something like "Long Equation for HP17B Solver" from Gerson IIRC. Astounding stuff, implementing the Trig functions and much more.

Keep in mind the Solver is the same for the 19B/19BII, 27S, 17B and 17BII, so you have to widen your search to get all the interesting Solver articles.

Don't forget Thomas Klemm, a true genius. See his article on the eight queens problem for an example of some brilliant solver code.

[jebem]

(06-26-2014 01:23 PM)rprosperi Wrote:  Nice machine Jebem, I think you will enjoy it, despite it's Algebraic nature.

Pull up a copy of HP-27S/19B Technical Applications (and of course the 27S UG) from the Museum DVD and learn the Solver.

Then search the Forum (OLD & NEW) for great articles from Don Shepherd and then, when you think you have it mastered, search for an article w/title something like "Long Equation for HP17B Solver" from Gerson IIRC. Astounding stuff, implementing the Trig functions and much more.

Keep in mind the Solver is the same for the 19B/19BII, 27S, 17B and 17BII, so you have to widen your search to get all the interesting Solver articles.

Thanks for your excellent hints, Bob!
I don't have the MoHPC DVD yet, that will be one of my next acquisitions.
And that Solver from the 80's is just amazing!