Natural display V.S. RPN, which is better?



#2

I am a high school student and I collect calculators. I am used to the "natural" way of input and output that CASIO uses. (On fx-es and es plus series) Then, when I come to RPN, I really do not think it's efficiency, or, it is efficiency only with numeric calculating. And the biggest problem of RPN is that I can't check my input after I do the calculate! LAST X or Undo helps little, and RPN means no replay function.


And hp doesn't have its own natural display engine on scientific calculators. The one used on smartmath 300 is completely a copy of CASIO. IMHO, only CASIO, SHARP and TI has own natural display engines.

Another question is: why does EquationWriter on hp48 run so slow? It's based on a 2/4 MHz CPU. fx-991es has a 500 kHZ CPU but it runs smoothly.


#3

"Another question is: why does EquationWriter on hp48 run so slow? It's based on a 2/4 MHz CPU."

Oh my god ! The 48 is 23 years old !Try the HP50G, eqw is instantaneous. I agree that the first equation write in 48SX is unusable because too slow. With the same processor (but faster 2X 4x?) EQW on HP49 is fine (I suppose it's rewrite). On the 50 it's perfect.

Edited: 4 May 2013, 8:03 a.m.


#4

There exist EquationWriter alternatives for the HP 48 which run very fast even on the SX, see RainEQ.

The HP 48 was one of the first (if not even the first one) to feature something like an EquationWriter. The built-in HP-written engine is slow, but in contrast to most of the alternatives it supports units in expressions.

The 49 has a fast EquationWriter alternative built-in which is derived from some software for the HP 48.

#5

When I look at a complicated formula I don't read it from left to right. I see it as a collection of separate parts which fit together. So, if I'm working out

(x-vt)/sqrt(1-(v/c)^2)

I first focus on v/c; then on the size of the square root; then on x and vt and how they compare to each other; and then on the final answer. Each separate part has its own physical meaning. Seeing the value of each separate part, by working out the formula step-by-step, gives me a better understanding of the complete expression.

I'll let other people defend RPN over step-by-step algebraic evaluation (!) but the above is why I prefer not to enter a complex formula as a single expression.

Nigel (UK)

#6

Algebraic shows what you want to get. RPN shows how you get there. Real-life engineering is not like school. In real-life engineering, you don't usually have an equation in front of you to enter, but instead you think through the process and steps of what you have to figure out, making RPN a good, efficient choice.


#7

Quote:
Algebraic shows what you want to get. RPN shows how you get there. Real-life engineering is not like school. In real-life engineering, you don't usually have an equation in front of you to enter, but instead you think through the process and steps of what you have to figure out, making RPN a good, efficient choice.





Very well said! Succinct and to the point. For exploratory calculations, which are so common in real-life engineering problems, nothing beats a hand-held RPN/RPL calculator. I use computers to write code, design FPFAs, create circuitry and lay out circuit boards but I still find my HP-48SX invaluable for many things.

As for "Real-life engineering is not like school. ", that is the understatement of the year.

#8

And don't forget that HP48SX build in 1989 has a CAS System inside that also handles units, complex numbers and matrix operations, ... On hpcalc.org you can see screenshots from what's possible with user written system level programs on HP48. Up to fast stack display replacements with full equation support per stack entry! (Java, HPSauce, EQSTK, ...) With some of these "updates" installed an 20 years old HP48SX could feel like new. ;-)

Back to your question :) I agree with Nigel. If I deal with larger more complex formulas I separate them in logical parts and or store these parts in different variables. The CAS rebuild them for me if I need a numerical solution.

Let me give you an easy example:

Store 'x-vt' in A,

'(v/c)^2' in B,

'A/sqrt(1-B)' in C.

Put C on stack and press EVAL and the HP48 CAS expands magically :) to (x-vt)/sqrt(1-(v/c)^2). Now imagine to build up your personal stored formula lib for later use...

Oh - it's exact why I love my 48SX so much. ;-)

Back in the old days of studying EE I create a folder for each homework in the { HOME } directory. Let's say {HOME HW4.5 }. Then create another folder for each part A_n { HOME HW4.5 A_1 } in it. Now I store all needed formulas, results, ... in these folders when I work on it. If I come back on the next day I can easily go thru all my work.

I think for every calc you need a workflow. For most HP RPN calcs you need more dive into it to feel the power ;-).


Edited: 4 May 2013, 5:14 p.m.

#9

Realistically, the machine is initially pretty irrelevant as it's a shortcut to cut out the boring repetitive bits of arithmetic.

You can get perfect natural display with symbolic manipulation (!) with one of those old fashioned graphite sticks, some flattened and bleached cellulose fibers, possibly a chunk of tree sap if it's a bad day and that grey stuff between your ears.

When the problem is solved, checking your inputs and replay is a case of looking at the paper in front of you.

Now from a subjective point of view, I use an RPL machine (50g) as it uses variadic functions with arguments on the stack for all operations. That allows me to abstract, store and compose the above paper work many times in different forms instantly.

Effectively, it allows knowledge to be reused efficiently. An example is:

Can you solve a resistor network with 20 resistors accurately in under a minute on a 991ES / 9850g / nSpire CAS / TI-83?

Nope - because the machine's user abstraction gets in the way and they lack the efficient protocol-less storage of knowledge.


Edited: 5 May 2013, 2:20 a.m.


#10

Quote:
Can you solve a resistor network with 20 resistors accurately in under a minute on a 991ES / 9850g / nSpire CAS / TI-83?

Yes, provided they are all connected in series :-)
However, if in parallel I would reach for my HP-48GX then do
{ R1 R2 ...  R20} 1/x MTH LIST SigmaLIST 1/x

#11

Or my LISPy way:

{ R1 .. R20 } RPAR FOLD

FOLD being fold/apply function left. I will post this program when I get a min.

RPAR --> << INV SWAP INV + INV >>

EDIT: RPAR above should be quoted in master expression.

EDIT 2: I think I actually prefer your method :)


Edited: 5 May 2013, 4:48 a.m.


#12

Actually, I use

{ R1 R2 ... R20 } PRLL
Where PRLL is this program:
%%HP: T(3)A(D)F(,);
\<< DUP TYPE NOT { 2
\->LIST } IFT INV
\GSLIST INV
\>>

I've been using my HP-48GX lately (the clock is never right on my HP 50g - but that's not the only reason).

Cheers,

Gerson.


#13

Nice solution. Will play with that one for a bit.

Why the 48GX over the 50G? (I have been considering a 48GX for a number of reasons myself).


#14

Quote:
Why the 48GX over the 50G?

1) Battery life.

I am using rechargeable Ni-MH 900 mAh SONY batteries on my 50g. Too often I have to recharge them before using the calculator (after not using it for a while). There is a new SANYO low self-discharge battery, but I haven't found it yet. The HP-48GX, on the other hand, will run for more than a year on a new set of three alkaline batteries, under moderate use.

2) Better keyboard.

I've had no problem with the 50g keyboard (the back arrow key has a lower profile, only slightly taller than the ON key, but I don't remember if it's always been so), but it doesn't compare to the HP-48 keyboard. Also, I prefer the 48 keyboard layout (perhaps because I used a 48GX for more than three years in college).

3) Reliable clock

Not this is really important, I could just not display it on the 50g. It is not an accuracy problem, it appears the clock just stops when certain calculations are performed (I haven't checked this yet). Right now my 50g is displaying 23:58 04:MAY while my 48GX shows 18:12 05:MAY, only two minutes away from the correct time (both were set about three months ago). I just quit setting the 50g clock.

Quote:
I have been considering a 48GX for a number of reasons myself

It's slower, but not too slower when considering battery consumption. Also, it doesn't have CAS (but you can install Erable, better on a RAM card). If the HP-50g (or HP-49g+) was your first RPL calculator, you may not like it. MetaKernel is nice too (an improved version of it is built into the HP-49G/G+ and 50g), but it requires an exclusive 128K RAM card. Perhaps you could try it on Emu48 first (select actual speed) before getting yourself one.


#15

Sounds pretty spot on to me.

I admit battery life on my 50g is pretty bad (~1 month under my usage) but I bought 2 packs of those Sanyo Eneloops and a charger and always keep a set of charged ones lying around. The cost over 2 years was much better. It is annoying having to worry about it though considering my Casio FX4000P has the second set of batteries in it since I bought it with in 1993...

Keyboard is fine for me after arguing with ->KEYTIME for a bit. Better than any other calc I've used. Probably could be better though as I do mis-key occasionally.

Yes I've noticed the clock is pretty rubbish. My £5 Casio F91W watch is responsible for all timing duties these days though as it keeps better time than anything I've had even up to 100x the cost, plus you can see it in the dark and it takes a kicking.

I don't use CAS and to be honest don't push it hard. The main features I use are units, basic functions, numeric solver, small-medium RPL progs.

How does the 48 manual compare to the 50g one? The 50g one is terrible.

Will fire up Emu48 and play.

Thanks for your time.


#16

Quote:
How does the 48 manual compare to the 50g one?

Not nearly as complete. See for yourself:

http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c00442267/c00442267.pdf

http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c00442266/c00442266.pdf

http://www.hpcalc.org/hp48/docs/misc/hp48gaur.zip

P.S.: You'll need the lastest 48GX ROM image, in case you don't have it yet:

http://www.hpcalc.org/details.php?id=4368

Edited: 5 May 2013, 6:36 p.m.

#17

Quote:


It's slower, but not too slower when considering battery consumption.


Did you consider trying SpeedUI for the HP 48GX ?


Ray


#18

Yes, I did consider SpeedUI, but then I decided to install MetaKernel first because I am more familiar with it (through the HP-49G and 50g version of it). As I am waiting for a new 1-MB RAM card, I am thinking of getting another 48GX as a backup and use my old 512K RAM card and SpeedUI on it (Here one of these days someone I know was selling a black-LCD 48GX for quite a good price, but I missed it for a couple of hours).

Unlike MetaKernel, SpeedUI is still being upgraded, which is very nice :-)
Thank you very much for your suggestion and continued efforts on improving SpeedUI!

Gerson.

#19

Hi Gerson. Have you tried using your 50G with a USB adapter? This could help extend the life of your batteries.

Regards,

John


#20

Hello John,

Yes, I have, especially when running programs that take too long to run. Most of times I connect it to a USB port which makes my desktop even more messy :-)

Regards,

Gerson.


Edited: 6 May 2013, 8:34 p.m.


Possibly Related Threads...
Thread Author Replies Views Last Post
  [PRIME] RPN: another attempt at returning more than one value to the RPN stack Marcus von Cube, Germany 5 338 11-05-2013, 02:44 AM
Last Post: Marcus von Cube, Germany
  Re: Natural display V.S. RPN, which is better? Waon Shinyoe (China) 2 138 05-05-2013, 10:47 AM
Last Post: Bunuel66
  Can the display for a HP32Sii be replaced with the display from another model? Bruce Larrabee 8 383 10-10-2012, 10:05 PM
Last Post: Luiz C. Vieira (Brazil)
  Calculator natural language user interface Donald Williams 12 407 01-07-2012, 09:45 AM
Last Post: Vince (Italy)
  Unusual non-HP RPN (red led display) to restore... Cristian Arezzini 11 391 10-24-2011, 02:58 PM
Last Post: Martin Pinckney
  16 segment Led display would mimic a HP-41 display? Giuseppe Marullo 3 204 11-12-2009, 08:08 PM
Last Post: db (martinez, ca.)
  Say with me: "Natural Complex Entry" Ángel Martin 5 224 10-02-2009, 08:57 AM
Last Post: Ángel Martin
  New Iphone & Itouch calculator - Access RPN and Active RPN Nigel Bamber 1 145 06-10-2009, 04:13 PM
Last Post: Jean-Michel
  Another calculator with RPN (Corvus 500 RPN) Saile (Brazil) 7 256 03-31-2009, 12:14 AM
Last Post: Michael de Estrada
  Base of Natural Logarithms Vincze 20 503 08-14-2007, 08:00 PM
Last Post: Miguel Toro

Forum Jump: