Is RPN still relevant?

[Matt Agajanian]

Althought we still have the HP-12C and HP-35S with the classic 4-register RPN, is RPN still a viable and marketable entry system from either HP's or consumers' pounts of view?

By the way, how fully-featured and functional is RPN on the HP-Prime?

[Dwight Sturrock]

It's only relevant as long as efficiency is important. In my casual observations, it's faster Matlab, Excel, TI and their ilk, and the app that came on your phone.

[ColinJDenman]

(10-05-2015 12:57 AM)Matt Agajanian Wrote:  ...
By the way, how fully-featured and functional is RPN on the HP-Prime?

Not very. Very much a work in progress compared to old school HP (I'm a 41 and 16 owner and started on the 70s National Semiconductor 4525 RPN). Given the target market (schools) it doesn't seem to get a high priority on firmware revisions and is clunky. Because the algebraic/textbook entry system and CAS system use things like polymorphic functions (same function can have different number and type of arguments) there can be some odd constructs in RPN. Sad really.

[cyrille de brébisson]

Hello,

There are a number of cases and places where RP* is still relevant, makes sense, allows to do things faster.

Educational math, is not, however, in my opinion, one of them.

RP* is very efficient, and also has a great advantage when you have intermediate, temporary results that you need to keep around.

HOWEVER, and this is the crux of the issue, in educational setting, you usually deal with cases where the 'student' will walk one small step at a time, do some more 'reading/learning', move the next step, etc.
This means that time will lapse between 2 steps on the calculator. During these 2 steps, he will have forgotten what is on each stack level and get stuck/make mistakes.
Algebraic does let you 'see' what you entered, modify it, review it...
Since algebraic does NOT have temporary, unnamed storage (the stack), it forces you to place stuff in variables (which takes longer and is less efficient), BUT is much more explicit, it is more adapted to this setting.

RP*, as stated above is GREAT for quick (in both sense of the term) calculations.
A RP* expert, who knows where he comes from and where he wants to go can even do very complex operation. However, after a while, even he will have to start writing stuff down.

Algebraic is more verbose, and as all verbose things, is both more explicit and longer.

As education is moving away from calculations toward higher level concepts, assuming that in the future more and more complex calculations will be done "by the tools", it does make sense to move way from RP* (sadly enough).

We see the same thing happening with computer languages with the move from Assembly toward C and now 'modern' C++ which is incredibly verbose and frustrating for a C 'old timer'.

This does not mean that RP* will completely disappear, but, like the hammer**, its use will probably slow down as other solutions, more adapted to the modern world, will appear, or as the world changes around it. The hammer is still used today, and so will RP* stay alive, BUT most a large number of 'old times' nails have been replaced by screws, and a large number of the leftover nails are put in using a pneumatic air-gun, not a hammer.

Cyrille, in a philosopher mood.

*N or L, pick your poison :-)
** probably not a good example, but I will use it anyway

[CR Haeger]

I expect the HP Team may not have time or $s to do this, but it seems to me that converting RPN to ALG (split screen?) might provide some useful feedback to newbie and experienced RPN users.

http://rosettacode.org/wiki/Parsing/RPN_...conversion

Even if this conversion only occurred for the first two stack levels I think it would be really useful. Certainly, it would keep both of my "RPN" and "ALG" neurons exercised.

[JimP]

(10-05-2015 12:57 AM)Matt Agajanian Wrote:  Althought we still have the HP-12C and HP-35S with the classic 4-register RPN, is RPN still a viable and marketable entry system from either HP's or consumers' pounts of view?

By the way, how fully-featured and functional is RPN on the HP-Prime?

Is RPN viable and marketable? I would say yes, because based on this monstrous "common core" analysis of how to subtract one number from another -- rather than enter one number, then the second, then hit the minus key, we have a convoluted mess. See: http://thetruthwins.com/archives/you-won...ubtraction (it's probably been pasted here before) But somewhere, someone in America decided to make it harder..... oh, what a silly act that was.

[brouhaha]

(10-05-2015 06:58 AM)cyrille de brébisson Wrote:  Since algebraic does NOT have temporary, unnamed storage (the stack),

Algebraic most certainly does have temporary, unnamed storage, which is usually implemented internally to the calculator as a stack. It's called parenthesis. The intermediate values of parenthesized subexpressions are on most calculators every bit as opaque as intermediate values in the RPN stack, if not more so, though the HP-71B "Calc Mode" is a notable counterexample. RPL machines are actually better about this, since they show you the stack contents.

That said, I agree that the education market doesn't need, and is in general not well served by, RPN or RPL calculators. As with all generalizations, there are some exceptions, but as a percentage of the calculator market, they are 0%. The only significant market for RPN calculators is the HP-12C, and that's only because it was so much better than the competing financial calculators of the time that it set a standard. If there was actually any non-trivial education market for RPN or RPL calculators, TI, Casio, and Sharp would be making them.

The reason Richard Ottosen and I are making our own RPN calculators is that we don't care that much about the education market, or about the mainstream calculator market in general. We're making the product primarily for ourselves, and if we can also sell a few to enthusiasts, so much the better. The enthusiast market is too small for giant multinational corporations to consider, except to the limited extent that they can at low incremental cost cram a few extra features with enthusiast appeal into an otherwise mass-market calculator.

I can't speak for the other people who are also making their own RPN or RPL calculators, but I suspect their motivations are similar to ours. Despite many complaints about how much our calculator is going to cost (estimated USD $299), we'll be very lucky if we don't lose our shirts selling it. $299 sounds like a lot of money, but the HP-65 cost $795 at introduction, which is $3,843 in today's dollars. While our calculator doesn't have the superb industrial design of the HP-65, I'd like to think that we beat it substantially in functionality, while being less than 1/12th the price. In fact, our price is under a third of the inflation-adjusted price of the HP-41C, and only 24% higher than the inflation-adjusted price of the HP-42S, the closest HP products in concept to what we're making.

I consider myself very lucky that I was introduced to HP calculators with RPN when I was in sixth grade. While I wasn't able to buy my own HP calculator until the HP-41CV was introduced, by which time I was in high school and had a part-time software development job, various people including some of my teachers let me use their HP calculators, starting with friends' HP-67 and HP-19C, and a chemistry teacher's HP-97.

[Brian D]

(10-06-2015 05:05 AM)JimP Wrote:  this monstrous "common core" analysis of how to subtract one number from another

Although to be honest there is some merit to the idea that 57-23 is ((57-20)-3), which can be a helpful way to think about things for two- or maybe three-digit numbers. They seem to take this too far, though.

[Harald]

(10-06-2015 05:05 AM)JimP Wrote:  Is RPN viable and marketable? I would say yes, because based on this monstrous "common core" analysis of how to subtract one number from another -- rather than enter one number, then the second, then hit the minus key, we have a convoluted mess. See: http://thetruthwins.com/archives/you-won...ubtraction (it's probably been pasted here before) But somewhere, someone in America decided to make it harder..... oh, what a silly act that was.

I have stared at this for some time now. This example is the first one I understood:



It is very similar to how I wold calculate it, but why use the difficult way?!?
I can't calculate 26+17 in one go either, but instead of doing something weird, I'd simply split 17 up in 10s and 1s: 26 + (10 + 7) = (26 + 10) + 7 = 36 + 7 = 43.
Ok, I admit, writing this down makes it look weird, too. But not as weird as splitting 17 into 13 + 4.

[Tugdual]

I still have faith in RPN to crunch calculations; I recently discovered RPL and found it extremely elegant and convenient for calculators. Now, even if it is possible, I don't think RPL is suited for very large apps to the point I wonder why the 50g is so big.
With the introduction of symbolic calculation, the use of RPN was naturally made obsolete. Still calculators have small keyboards and I simply don't think calculators are primarily designed for this usage, computers will do a much better job is all other cases. Calculators should remain simple tools used only for the sake of crunching quick calculations or coding some quick automation. For that reason, RPL and RPN have reasons to remain to the detriment of gigantic calculators which may disappears just like dinosaurs did.

[Namir]

In the late 70s RPN (and AOS) were the only game in town. BASIC found it's way on handheld pocket computers and of course PCs. Then came Turbo Pascal and many folks, like myself, went from programming legacy (spaghetti) BASIC to structured programming in Pascal (and some in C) and even structure Basic programming like QBasic. Then C++ came and many folks went OOP!

Today I write only short RPN programs. I like to code in Excel VBA, Matlab, PPL (HP Prime being a very capable machine) and other languages like R, Python, and Ruby.

Readability of code is really important. RPN (and RPL) do not get the Nobel Prize in code readability. We like algebraic code, especially the ones that support vector/matrix operations (like Matlab, and HP Prime, just to name a few) since they get loops out of the way and add more clarity and compatibility between matrix/vector mathematical expressions and comparable source code.

Namir

[SlideRule]

ALL

Perhaps the Alternative Calculator at http://www.waterlog.info/altcalc.htm is a POSITIVE indicator of the state of the art, with only an X - Y STACK & a temporary S register, written by a TEN YEAR OLD.

RPN is how I do ARITHMETIC, irrespective of how the Mathematical Expression is presented.

KUDOS KID!

[attachment=2644]

[John Colvin]

(10-06-2015 09:57 AM)Harald Wrote:  

(10-06-2015 05:05 AM)JimP Wrote:  Is RPN viable and marketable? I would say yes, because based on this monstrous "common core" analysis of how to subtract one number from another -- rather than enter one number, then the second, then hit the minus key, we have a convoluted mess. See: http://thetruthwins.com/archives/you-won...ubtraction (it's probably been pasted here before) But somewhere, someone in America decided to make it harder..... oh, what a silly act that was.

I have stared at this for some time now. This example is the first one I understood:



It is very similar to how I wold calculate it, but why use the difficult way?!?
I can't calculate 26+17 in one go either, but instead of doing something weird, I'd simply split 17 up in 10s and 1s: 26 + (10 + 7) = (26 + 10) + 7 = 36 + 7 = 43.
Ok, I admit, writing this down makes it look weird, too. But not as weird as splitting 17 into 13 + 4.

When I first looked at this method, I thought it was a way to avoid the carry, but in the following example, a carry is still required:
43 + 89
43 = 3 + 40
89 + 3 = 92 ; carry
92 + 40 = 132

So I don't see how this method simplifies anything.
" I can't calculate 26+17 in one go either" - Why not?

[brickviking]

(10-07-2015 08:13 PM)John Colvin Wrote:  " I can't calculate 26+17 in one go either" - Why not?

Often when I come across stuff like this, I (in my mind) compare the second digit to the nearest factor of ten larger and "keep" note of the difference. Then I add the factor, and take off the difference.

Code:


26 + 17:
26 + 20 (20 - 17 = 3) = 46-3=43

271+48:
271 + 50 (50 - 48 = 2) = 321 - 2 = 319

That's just in my memory, and I've got what I now consider grade school education up to about Year 11 which I've forgotten a good deal of. Multiplication is a bit more difficult, and division's even worse unless I have pen and paper or calculator. Amazingly, I'm still more competent than most of the people I meet for really simple "this is how much I need to pay you, and this is how much change I get from $30" stuff. Most of our population would be rather lost if the power went down, and they had no calculator available. Just don't ask me to binary-add with my fingers and toes.

(Post 40)

[Thomas Klemm]

(10-07-2015 08:13 PM)John Colvin Wrote:  43 + 89

It's just applying the associative law:
\[
\begin{align}
43 + 89 &= (42 + 1) + 89 \\
&= 42 + (1 + 89) \\
&= 42 + 90 \\
&= 132
\end{align}
\]

Or then you can go into the other direction:
\[
\begin{align}
43 + 89 &= 43 + (7 + 82) \\
&= (43 + 7) + 82 \\
&= 50 +82 \\
&= 132
\end{align}
\]

Why shouldn't children use this?

You can do the same with multiplication:
\[
\begin{align}
12 \times 37 &= (4 \times 3) \times 37 \\
&= 4 \times (3 \times 37) \\
&= 4 \times 111 \\
&= 444
\end{align}
\]

I just happen to know that \(111 = 3 \times 37\). So whenever I see \(37\) in a product I check if I can "borrow" \(3\) from another factor.

Cheers
Thomas

[JimP]

(10-07-2015 08:53 PM)Thomas Klemm Wrote:  You can do the same with multiplication:
\[
\begin{align}
12 \times 37 &= (4 \times 3) \times 37 \\
&= 4 \times (3 \times 37) \\
&= 4 \times 111 \\
&= 444
\end{align}
\]

It would feel easier to me, personally, to say 10x 37 = 370, then add 2x37 (=74) to that. But when written down, the old method of carry-the-number to the next column just seems less complicated than some of the elaborate breakdowns, appear to generate. Particularly with addition and subtraction. Long division and long multiplication was something my dad taught me when I was six (and I'll be eternally grateful to him for doing that!) -- but in the mid-1960s that was a more important art than it is today!

[ttw]

It's relevant for me anyway. Over the years I have broken one HP50g, lost one, and had one just die. I've got two now (I got one in today to replace a lost one.) I've got another that is partially damaged; the case has parts sticking out but it still works.

For quick implementation of ideas, the calculator is nice. If things work, then I'll go to the desktop and write a program, but that has lots of overhead (but runs a million times faster generally.)

[Sukiari]

(10-09-2015 08:10 PM)ttw Wrote:  It's relevant for me anyway. Over the years I have broken one HP50g, lost one, and had one just die. I've got two now (I got one in today to replace a lost one.) I've got another that is partially damaged; the case has parts sticking out but it still works.

For quick implementation of ideas, the calculator is nice. If things work, then I'll go to the desktop and write a program, but that has lots of overhead (but runs a million times faster generally.)

If you haven't tried Forth yet maybe give that a shot, or try Factor (like Forth on steroids) which is what I use sometimes too. RPL is very similar to Forth and that's not an accident.

[Jlouis]

For me, speaking of crunching numbers, RPN is, and will be, relavant as ever.

Algebraic is very prone to errors, with its parenthesis.

And please, don't talk about pretty print. I can give a result for an equation while one is still keystroking the equation in a pretty print calculator. (I know, not so humble, but still my opinion).

LONG LIVE RPN!

[Katie Wasserman]

(10-06-2015 06:07 AM)brouhaha Wrote:  The only significant market for RPN calculators is the HP-12C, and that's only because it was so much better than the competing financial calculators of the time that it set a standard. If there was actually any non-trivial education market for RPN or RPL calculators, TI, Casio, and Sharp would be making them.

I agree, but it's a serious market, maybe not a lot of money for HP but still a lot of unit sales. A high percentage of people in the financial world, including real estate, use these and would be lost without them. HP has tried several times to replace it with significantly better and easier to use models and has failed. The 19b, 17b, 20/30b have all come and gone and yet the 12c survives. In addition there are been several clones of the 12c. If it wasn't a very significant part of the overall calculator market those clones would not exist.

I love the fact that in this new smart phone every 1 or 2 years world that the humble 12c has managed to virtually never change as it goes into it's 35th year of continuous production.