Posts: 65

Threads: 14

Joined: Jun 2013

While researching the Factorial Inconsistency, I discovered a small typo in the manual. On page 181 of the printed manual or 162 of the 3.1 online pdf manual under the description of "Domain Error" the manual states that logs of x <= 0 will return the "Domain Error".

This is not correct. Logs of x < 0 will return the "Domain Error", logs of x = 0 correctly return "- infinity error".

The "x <= 0" should read "x < 0" in this case.

Posts: 3,229

Threads: 42

Joined: Jul 2006

Almost correct, logs of -0 return domain error.

- Pauli

Posts: 65

Threads: 14

Joined: Jun 2013

How does one go about getting -0 into the x register? I tried and couldn't get it to ever show a -0 in the x register.

*Edited: 23 July 2013, 4:20 a.m. *

Posts: 3,229

Threads: 42

Joined: Jul 2006

Start by setting flag D. With this flag cleared, you can't get -0.

Ln(-0) returns a NaN result with flag D set which is equivalent to a domain error. To verify this: SF D 0 +/- CF D LN results in domain error not infinite.

- Pauli

Posts: 3,229

Threads: 42

Joined: Jul 2006

Well, in integer mode you can get -0 regardless of flag D but there LN(0) always gives a domain error.

- Pauli

*Edited: 23 July 2013, 4:37 a.m. *

Posts: 65

Threads: 14

Joined: Jun 2013

Perhaps the manual should read

Logs x < 0 or logs of -0 return a Domain Error. Then it would be accurate for all cases.

Posts: 3,229

Threads: 42

Joined: Jul 2006

Not in integer mode it wouldn't. No infinite results there.

- Pauli

Posts: 65

Threads: 14

Joined: Jun 2013

I combed through the manual, and couldn't find any special definition of what the "D" flag does. Could you enlighten me?

It obviously has an affect on how numbers are stored, thus allowing

the negative zero to be entered, but what is it's definition?

Posts: 3,229

Threads: 42

Joined: Jul 2006

Page 24 in my copy of the PDF describes it briefly. There are other references to the values it permits spread throughout.

Flag D does not change how numbers are stored, neither does it change how the various functions are computed. It prevents errors for infinite and NaN results and suppresses the display of the negative sign for negative zero.

- Pauli

Posts: 65

Threads: 14

Joined: Jun 2013

Perpetually Amazed.

It never ceases to amaze me how well written and perfect this calculator is. Every math function gives the most accurate results I have ever seen anywhere. Even better than most desktop computer math libraries.

I don't know how you managed to get this kind of precision everywhere.

Posts: 4,587

Threads: 105

Joined: Jul 2005

Sounds like a little challenge. I'll think about it this evening.

d:-)

Posts: 3,229

Threads: 42

Joined: Jul 2006

I think it will be quite a challenge to fully document this one :)

- Pauli

Posts: 4,587

Threads: 105

Joined: Jul 2005

About flag D: please see pp. 31, 87 (footnote 58), 116, 120, 125, 151, 180, 181 (footnote 92), and 183 of the printed manual.

d:-)