Bug HP39GII....or feature?

[Bunuel66]

I found a funny (weird) behavior of the 39GII,just try to compute
1/3EE5 by using 3 then 10⁵ or 3. then 10⁵. In the late cases the string will be interpreted literally without any consideration for the semantic of 10⁵. This is very error prone, and other calculators don't behave that way. It is perfectly clear that, with the current implementation each key delivers a set of characters, building progressively a string which is late interpreted before computation. Why not? but it would be nice that 'special' strings will be separated by a blank in order to identify them. In that particular case it could be implicit that the power of ten relies to the number immediately to its left separated by a blank.
Just my two cents. Maybe in some future prom...

[Tim Wessman]

Hello,

We are aware of this issue and looking into it. You are correct that at this point it essentially behaves like a typing aid. By changing it to a special function type of argument it should improve things to work more as expected. I envision that it will use a small caps 10 to provide some distinctiveness. On the 50g it puts in the ALOG() function, but even then it doesn't do what is asked for here.

Just out of curiosity when I was investigating this issue, I noticed the nspire has the same problem as reported here. It seems to be that if you are familar with or use Casios, this behavior is more expected. Since it doesn't break any other behavior, I personally agree it should behave more like the EEX in that it is an extension of a number, rather then typing [1] [0] [^] which is what it does now essentially.

TW

Edited: 10 Aug 2012, 1:32 p.m.

[fhub]

A short question, Tim:

Your download link for the HP39gII emulator doesn't work anymore (since quite a few days):

http://hpcalculating.com/yn5_s4jdSXa

Has it any meaning that the emulator has been removed from there?

Maybe you're working on a new version? ;-)

Franz

[Gilles Carpentier]

Quote:

---

On the 50g it puts in the ALOG() function, but even then it doesn't do what is asked for here.

---

Tim, I think we have here not one but 2 questions ! it is also a 'problem' of priority of calculation.

If I understand well (I only see 10&#8309 in the message of Bunuel but I suppose it is about the 10^ function)

In algebraic on HP50G the * is implicit (like on 39gII):

3LN(20)results 3*LN(20)

3/5LOG(20) results 3/5*LN(20) => (3/5)*LN(20) (wich is logic becaus there is absolutly no reason that * has more priority that /

same with ALOG

3ALOG(10) results 3*ALOG(10)

1/3ALOG(10) results 1/3*ALOG(10) wich is perfectly logic and is (1/3)*ALOG(10)

In my opinion :

1/3EE5 is not the same that 1/3*10^5 wich must be interpreted as (1/3)*10^5. There is absolutly no reason that 10^x works in a different way that others functions

ALOG or special 10^ characters is a good idea (to avoid 3KEY_10^_5 to be interpreted as 310^5) but 1/3*10^5 results 1/3EE5 would be a bug (and not a feature ;) in my opinion.

I would prefer ALOG to avoid confusion between 10^ ( 1 0 followed by ^ function) and 10^ (ALOG function, with one parameter)

Another example where RPN is more logic and without ambiguity

PS: I also remark that '1 shift EEX 3 ' is the same as ' 1 ALPHA E 3'

Edited: 10 Aug 2012, 3:19 p.m. after one or more responses were posted

[Tim Wessman]

The issue is that the only people that really use the 10^ function are using it more like EEX then ALOG. So you end with this issue. I personally have never used it in my life, but in certain places in the world it gets used in lieu of the EEX syntax.

[Bunuel66]

Well, the point is that when you do actual calculations with a pen, you never use EE but 10^x. It is even the way it is written in textbooks of math or physics. I'm perfectly comfortable with your interpretation, but it's not real life, at least not current real life where calculators are supposed to be used as you write. EE is reminiscent of the first calculators where you didn't had 10^x function, and later on when it was strictly the reciprocal of log(x).
We can discuss for ever about the way you enter an expression in a calculator. Usually, the term 10^x applies to the number immediately to its left. The example your giving can be interpreted as you mention but, by hand, the ambiguity is solved either by the power being written on the lower or upper line, or by using braces. For the HP 39GII there is only one line...
My point is that if HP wants to be back as a serious competitor to TI or Casio that kind of point can be devastating. Think to a kid who will get a wrong result due to the current behavior while his friend will get it right....Not that good for HP....
What they want at school is a 'cool' calculator with a fancy color screen which provides results right when they enter them as they write. Just, as a test, ask kids if they prefer RPN or algebraic interface....
And please don't misunderstand me. I'm really an HP fan for years, and yes RPN was, and still is, damn cool. And yes I'm an old school engineer with pens in the pocket of the white shirt and still knowing how to use a slide rule.... Unfortunately, this is not the target for HP with this calculator. If they want to sell to schools or educational public, that kind of point is important.

Regards

[Bunuel66]

BTW, I apologize for the strange signs in my previous post...I'll check better my message next time ;-(

[Nick R]

If calculators are designed so that all students have to do is enter a mathematical problem exactly as it appears in the textbook, what are they actually learning?

[Gilles Carpentier]

Hi Bunuel,

no problem, I understand. But I never use 10^x instead of E

_For me_ 123E12 is a number and 123*10^12 is a calcul. Perhaps it's not the same things for news generation.

-> I tried with XCAS (in use in many french scool) and it works in the same way.

-> Same things with Open Office, E is not 10^x

-> But Wolfram (mathematica) works the way you explain in a very very curious way :

1/3*10^5 results 1/(3.10^5)

but (with just a point after 5):

1/3*10^5. results (1/3)*10^5 (!!!)

And

1/2*10^(2+3) result (1/3)*10^5 (!!!)

Curious conclusion for Wolfram :

1/3*10^5 is not equal to 1/3*10^(3+2) :~/

I found this behavior absolutly illogical. I am curious about something as 1/3*10^n :O I think it is a dangerous behavior wich can give unpredictable result and bugs

Edited: 10 Aug 2012, 6:36 p.m.

[Thomas Radtke]

To avoid confusion, a good textbook writes a fraction different, i.e. numerator top, nominator bottom. I've never seen it any other way in educational material (I'm in Germany).

Edited: 11 Aug 2012, 8:06 a.m.

[Bunuel66]

Well, I'm probably not clear enough.
-First, forget about the fraction. Just try 310^5 and 3.110^5, did you get the expected result?
-Then, for fraction, we (in France) use also a line separator between numerator and denominator, but as you only have a line for input on the HP39GII this leads to some additional ambiguities.
-Finally, on all the textbook I have seen, both in US and France specially in physics and chemistry, powers of 10 are not written with E. May be we didn't use the same library....;-)
About Xcas, it is not that used and I wouldn't describe it as kids friendly....If Xcas is used as an underlying engine it is not that surprising to get the very same behavior.
I haven't check Mathematica.

To summarize it seems that the main problem is that the interpreter reads strings of characters and doesn't parse the input correctly for some operators. Actually 10^x is not seen as an operator but as a set of characters if not separated by a space. And when separated by a space it applies not to the immediate left number but to the immediate left result.

That said, it's a consumers market. I'm very comfortable with the idea that a company makes its product as it likes. If the consumer doesn't like he has just to change of producer. Good news for the competitors ;-)

[Patrice]

Hi Bunuel,

As far as I understand it, for everybody,

310^5 means (310)^5

3.110^5 means (3.110)^5

If you think that writing 310^5 is 3*10^5, how should you write what will lead to (310)^5 ?

If you think that writing 3.110^5 is 3.1*10^5, how should you write what will lead to (3.110)^5 ?

if you expect something else, you are forgetting that the 39GII is an AMERICAN tool which respect AMERICAN way to write a number and that the decimal separator is a point!

You can't expect the point to act otherwise!

This calc is using the convention that EEX or E says that a power of 10 is part of a number, why not using it?

The layout of an equation/formula include some implicit things (hidden parenthesis), you can't expect to write the same formula in linear form without making explicit what was implicit.

By the way, I am French, I know that 3,1.10^5 means 3.1*10^5 or 3.1E5 in France, but it is not a French tool.

Patrice

“Everything should be made as simple as possible, but no simpler.” Albert Einstein

[Bunuel66]

Sorry, you haven't got the point. 310^5 is what you got after typing '3' and then 'shift' then '10^x' and '5', same for the other expression with the point.
I know perfectly that the point is a separator, but it is not the point of discussion. And yes, HP is an american trade mark. And yes EE is to write the exponent of ten for a number. The fact that it is an american product is fully irrelevant with the comments.

Let's stop this discussion. The ones who have got the point doesn't need more and the ones who haven't will not get it.

Keep It Simple Stupid....(who said that?)

Regards

[Tim Wessman]

Tried sending you an email through the forum but not sure if you are receiving it. Send me a message at timwessman at hp.com so I can show you something.

TW

[Gilles Carpentier]

Hi Bunuel

I agree with you on this point...

As the multiplication is implicit (is this in english word ?) on the 39gII 3_10^x_5 (clic 3 then 10^x key then 5 ) must return 3E5 and not 310^5

Edited: 11 Aug 2012, 3:00 p.m.