HP Prime emulator: Gamma function



#2

I played around with the HP Prime emulator today & also tried the Gamma function. For positive numbers everything seems to be OK, but for negative numbers the Prime emulation gives different results compared to my 15C, 15C LE, 34C, 35s & HP49g+ (which all give the same result, e.g. -5.5! => -0.0600; unfortunately I forgot the result of the Prime emulator & my virtual W7 machine is not up at the moment...)

Might there be a problem with the Gamma function of the HP Prime? Or is it only an emulator problem?


#3

Did you look at the help for the function? Graph both Gamma(X) and X! and I think you'll see what is happening. Also compare the graph at http://en.wikipedia.org/wiki/Gamma_function to what you see on the calc.

TW

Edited: 21 Aug 2013, 2:54 p.m.


#4

On the HP-49G+ and on the Prime, Gamma and ! are two different functions. From the manual:

---------------------------------------

Factorial: Factorial of a positive integer. For non-integers, ! = |-- (x + 1).
This calculates the gamma function.

value!
.
.
.
Gamma: Returns the value of the gamma function (|--) for a number a.
Gamma(a)
---------------------------------------

Thus Gamma(-4.5) = (-5.5)! = -0.0600196013005

Gerson.


-----

This was meant to Stephan. Somehow it has ended up here.

Edited: 21 Aug 2013, 3:05 p.m.


#5

Thanks, Gerson.
Actually I know the difference between ! & Gamma. I just was astonished that my Prime emulator gave me some unexpected result, but as Joe proposed I'll check my settings first.


#6

You might also want to check your emulator version:



#7

WolframAlpha disagrees with the precedence of - and !: -5.5! = -287.885....

Cheers

Thomas

#8

Gerson

are you sure about -5.5! ?

I get this for (-5.5)! with my emulator -5.5! -> -287.885...


#9

Gilles and Thomas,

I did enter (-5.5)!, but the history showed -5.5! (I hadn't noticed that). Looks like we have another bug here:

Cheers,

Gerson.

#10

I checked the thing again this morning & got the incorrect results again:



:-o



Edit: It seems I have problems with the images (if it's not only our proxy server) ...

The values are:

Gamma(-4.5) = -.245736133286

(-5.5)! = -.245736133286



However, after resetting the calculator it shows the correct result:



Thus it might have been some weird settings preventing the calculator to return the correct results:

Gamma(-4.5) = -.0600196013005

(-5.5)! = -.0600196013005

The version of my emulator is:



Version 2013 8 5. Rev: 4980
The emulator tells me that there's no newer version available.



I apologise for all inconveniences...

Thanks for your support!


Edited: 22 Aug 2013, 2:16 a.m.


#11

[img:http://img96.imageshack.us/img96/1623/yaui.png]



[img:http://img580.imageshack.us/img580/8314/dkpr.png]



[img:http://img839.imageshack.us/img839/6941/nv0h.png]



Not a problem with a proxy but with markup. And I used the direct URL.

Cheers

Thomas

#12

I got it !

CAS mode
Set <\ in degree

(-5.5)! = -.245736133286

Rev 4980


#13

Good find! This means the formula involving sine is being used for negative arguments. The bug persists on Rev. 5106.


#14

Quote:
This means the formula involving sine is being used for negative arguments.

= -0.24573613328569488790394658885834332520579131620384...

Quote:
The values are:

Gamma(-4.5) = -.245736133286

Gerson, I think you're correct! But then why is -4.5 used instead of 5.5 in sin?

Cheers

Thomas

Edited: 22 Aug 2013, 10:23 a.m.


#15

Perhaps they're doing

(-z)!(z - 1)! = pi/(sin(pi*z))
and letting
z = -4.5
Thus,
Gamma(-4.5) = (-5.5)! = pi/(sin(-pi*4.5)*4.5!) = -0.0600196013005 
Considering the argument has been unduly converted into degrees, we convert it back to radians, that is,
z = -4.5*pi/180
and we get the same result the Prime is giving, -0.245736133286, which of course is completely wrong, as you know.

Cheers,

Gerson.

Edited to fix a typo.


Edited: 22 Aug 2013, 2:53 p.m.


#16

Oh, now I see:

Thanks for pointing that out!

Best regards

Thomas


#17

Actually I only intended to write it in an easier-to-remember and less prone to confusion form:

Best regards,

Gerson.


#18

This formula reminds me of Valentin's HP-15C Mini-challenge.

#19

I tested with XCAS and same problem...

#20

Quote:
However, after resetting the calculator it shows the correct result:



Thus it might have been some weird settings preventing the calculator to return the correct results:

Gamma(-4.5) = -.0600196013005

(-5.5)! = -.0600196013005


Not your fault, but HP's. They've forgotten to check the user's angle mode setting when evaluating Gamma . See Gilles's post below. You've indeed found a new bug. Thanks for reporting!

Regards,

Gerson.


#21

You're welcome. Actually it was by mistake... ;-)

Does anyone know if the bug also occurs on the real Prime or just in the emulator?

Edited: 22 Aug 2013, 11:24 a.m.

#22

Hi,

So, CAS mode does not show the expression the way you entered (CAS mode, Radians):


What is the difference between these two ?


Well, I copy the first one :


Then press [Enter] key:



Then, I copy the second one :

And there you have it. Of course, pressing [Enter] key gives:


#23

So they look alike, but are actually different :-)
Perhaps more difficult to fix than the other bug, but I think HP should address this one as well. Otherwise screen prints won't be a reliable information anymore.

Regards,

Gerson.

#24

If it's a bug, it must involve a setting of some kind, because after Calculator Clear on the emulator, all four of the following return the same correct result (-.0600196013005) both in Home and in CAS:

(-5.5)! (using the [-] key)

(-5.5)! (using the [+/-] key)

Gamma(-4.5) (using the [-] key)

Gamma(-4.5) (using the [+/-] key)

If you see the bug again, please let us know what your non-default settings are. Thanks!

-Joe-


#25

Thanks, Joe.
I'll have to check that tomorrow.

#26

Hi Joe

The bug is in CAS mode and ° setting :

Set angle in Radian
CAS mode

(-5.5)! -> -.6...

now set angle in ° (always CAS mode)

(-5.5)! -> -.2457...

(Rev 4980)


Edited: 22 Aug 2013, 3:14 a.m.


#27

The setting (Degree or radian) affects both environments (goto settings in CAS or Home and see they are the same). The error only occurs from the CAS screen, not from the Home screen.

#28

Hello prime researchers,

the gamma function is normally defined for complex numbers except the negative integers. Maybe there is connection between the DEG and RAD mode and the inconsistent results in CAS mode?

Greeting
peacecalc

#29

Quote:
More important to get things right, with a smaller subset of functions that are fully tested and well documented, before starting the "lets lump in everyone's favorite function" game...

Couldn't agree more.

Cheers

Thomas

#30

I compared a few negative values (-5.5, -4.5, -3.5, -2.25, -2.45, and -22.45) for the Gamma function using the emulator AND using Wolfram Alpha. I did not see any difference in the results.

Namir

Edited: 21 Aug 2013, 4:53 p.m.


Possibly Related Threads...
Thread Author Replies Views Last Post
  Prime Emulator Connection Problem John Colvin 3 711 12-14-2013, 11:00 PM
Last Post: Han
  HP Prime Emulator download John Colvin 2 535 12-14-2013, 05:54 PM
Last Post: John Colvin
  HP50g: Writing a function that returns a function Chris de Castro 2 584 12-10-2013, 06:49 PM
Last Post: Han
  IFERR function on HP Prime Mic 2 514 12-02-2013, 01:33 AM
Last Post: cyrille de Brébisson
  HP Prime: Dirichlet's eta function recognized but not numerically evaluated Helge Gabert 0 292 11-16-2013, 03:41 PM
Last Post: Helge Gabert
  Possible bug with sqrt function in the HP prime Michael de Estrada 6 678 11-15-2013, 12:49 PM
Last Post: Michael de Estrada
  HP-41 MCODE: The Last Function - at last! Ángel Martin 0 302 11-08-2013, 05:11 AM
Last Post: Ángel Martin
  [Prime] any ideas for a undo function? Stefan Dröge (Germany) 8 831 11-04-2013, 04:37 PM
Last Post: Damien
  HP Prime 'where' function bluesun08 11 1,093 10-29-2013, 06:56 PM
Last Post: Joe Horn
  HP Prime - Defining a function bluesun08 5 640 10-23-2013, 02:43 PM
Last Post: Han

Forum Jump: