Simplify with sci/eng on hp 50g

[Juraj O.]

When I change into engineering/scientific mode with a precision of e.g. 1 digit, '70E-9 *cos(0)' gets simplified into 1/e^16 (cas is set to exact). What is going on here?

[Crawl]

I am unable to reproduce your error. I get 7.0E-8

A couple of questions:

-In the answer, is that e the normal (2.71828...) e?

-Do you have E defined to be a variable with a value?

-Can you explain *exactly* what you're doing?

Example, I am in RPN mode, enter 70E-9 *cos(0) in single quotes, and hit EVAL (or simplify, whatever) and get 7.0E-8

[Juraj O.]

Set mode to engineering with precision 1, cas mode exact. Open equation editor, enter cos(0) * 0.01 (another example). Select whole expression and simplify it, or close the equation editor and invoke SIMPLIFY (EVAL does not exhibit this behaviour). It gets simplified to 1/e^5 (yes, that's e, the exponential function).

[Juraj O.]

I have made a recording of this in the emulator:

http://www.mediafire.com/?9cnml3z1c3yvhw8

[Bart (UK)]

Hi Juraj,

I confirm I get the same strange result when using SIMPLIFY and low display digits in Exact Mode. This seems to be a problem with version 2.15 of the calculator firmware. In version 2.09 it will request Approx. mode and display the correct answer.


To others,

Just entering the equation with EVAL it seems it always gets the right answer, irrespective of display digits setting.

To reproduce the error:
Have version 2.15 and set display to either FIX 1, SCI 1 or ENG 1
Enter the equation in Equation Writer, select it completely and use SIMP
or
enter the equation on the command line '70E-9·COS(0)' go to CONVERT -> REWRITE -> SIMPLIFY


EDIT:

Further to the above: it seems also to be a problem with the 50g association of integers and exact mode, and floating numbers and approx. mode.
Enter the equation as:
(70/1000000000)*COS(0)
and it will simplify it correctly.


Thus the v. 2.09 response is correct to request Approx. Mode because of the 70E-9. It does not request it when 70/1000000000 is entered. The 50g is wrong by just giving an answer that is incorrect. Something broke here between 2.09 & 2.15.

Edited: 7 Nov 2012, 6:48 a.m.

[Juraj O.]

In approx mode, 1•0.01 gets simplified right into 6.7E-3. This is really dangerous! How the hell did not anyone discover this earlier? That was on my 3rd or 4th day of owning a 50g....

Edited: 7 Nov 2012, 10:59 a.m.

[Gilles Carpentier]

How do you do ?

In RPN mode I get the right answer

[Bart (UK)]

Quote:

---

How do you do ?

In RPN mode I get the right answer

---

Hi Gilles,


v2.09 does get the right answer, but v2.15 gets the answer that Juraj noted. (we are talking about SIMPLIFY command with low digit ENG/SCI/FIX mode).

Edited: 7 Nov 2012, 11:52 a.m.

[Bart (UK)]

Quote:

---

How the hell did not anyone discover this earlier? That was on my 3rd or 4th day of owning a 50g....

---

I guess most users like lots of numbers and do not normally use the SIMPLIFY function with such low digit requirements.

Quote:

---

In approx mode, 1•0.01 gets simplified right into 6.7E-3.

---

It still seems to try and use the Exact techniques (in Exact mode it gives 1/e^5, which = 6.7E-3 with "-> NUM" in FIX 1). When entering it as '1·(1/100)' the correct answer is produced.

It seems to try and find the nearest answer representable with e. I tried several flag settings, but nothing seems to help. Perhaps others have better ideas. As I said, it seems something here broke between v2.09 and 2.15.

Edited: 7 Nov 2012, 12:06 p.m.

[Crawl]

Interesting. I tested this on a emulator, though, which would have been an older version. I left my physical HP50g at home, so I'll have to check this on that later.

[Crawl]

Quote:

---

Interesting. I tested this on a emulator, though, which would have been an older version. I left my physical HP50g at home, so I'll have to check this on that later.

---

It looks like my physical calculator does have this issue. Maybe I should try to install an older version?

[Han]

I can also confirm this behavior on the HP50G ROM 2.15

Han