The Museum of HP Calculators

HP Forum Archive 14

 Some 49G+ LimitationsMessage #1 Posted by John Smitherman on 9 Feb 2005, 12:06 a.m. Being a new 49G+ owner I have been trying some of the functions to see how well it handle "extremes". Needless to say, I am disappointed with what I have found thus far: o 1000! = 1e500 with no overflow warning. o 1.5! = 1.329 but in my opinion should give an improper (non-integer) input warning. o Comb(2000,1000) = 1e500 with no overflow warning. o asin(acos(atan(tan(cos(sin(7) = 7.000003185 which is no better than the 32sii and not up to TI 89 accuracy. Is anyone aware of a list of other 49G+ limitations, bugs, etc.? Regards, John

 Re: Some 49G+ LimitationsMessage #2 Posted by Vieira, Luiz C. (Brazil) on 9 Feb 2005, 1:21 a.m.,in response to message #1 by John Smitherman Hi, John; let's see:Quote:o 1000! = 1e500 with no overflow warning.Please, try:`[MODE][FLAGS][down-arrow] 6 times and [|/CHK]`You'll see that, now, overflow returns an error message. BTW, flag 20, underflow, may be checked for error as well. To maintain this setting, confirm with [OK] twice. Quote:o 1.5! = 1.329 but in my opinion should give an improper (non-integer) input warning. This is because the [!] function (and also FACT) computes GAMMA(x-1) when the argument is not an integer. Quote:o Comb(2000,1000) = 1e500 with no overflow warning.Same as first quote. Quote:o asin(acos(atan(tan(cos(sin(7) = 7.000003185 which is no better than the 32sii and not up to TI 89 accuracy.The actual value is 7.00000318497 (STD mode set). What is the value obtained with the TI89? Hope this gives you some new info. Best regards. Luiz (Brazil) Edited: 9 Feb 2005, 1:23 a.m.

 Re: Some 49G+ LimitationsMessage #3 Posted by Mike (Stgt) on 9 Feb 2005, 5:52 a.m.,in response to message #2 by Vieira, Luiz C. (Brazil) TI89 ROM Version 2.05 results 7 + 2.83714e-8 (at least in it's emulation). BTW, if you need an exact result for 1000! and COMB(2000;1000), the HP-49 has a CAS. Ciao.....Mike

 Re: Some 49G+ LimitationsMessage #4 Posted by Vieira, Luiz C. (Brazil) on 9 Feb 2005, 10:37 a.m.,in response to message #3 by Mike (Stgt) HI, Mike; Yeap! I actualy found a "many digits" result for 1000!... They were so too many I gave up reading. Thanks! Luiz (Brazil)

 That would be nice ;-) (was: Some 49G+ Limitations)Message #6 Posted by Vieira, Luiz C. (Brazil) on 9 Feb 2005, 5:56 p.m.,in response to message #5 by Mike (Stgt) Hi, Mike; I agree with you. You see, I take the HP49G as the closest scheme for this "custom" calculator. As you say, the HP49G already has all math tools you can dream of. If adding any particular procedure to the existing OS is an easy matter, then it is already one possible, not too expensive solution. The upgradable OS has been the major reason I bought two of them. Cheers. Luiz (Brazil)(somehow driffting the subject as well)

 The best users work with limitations and with pluses.Message #7 Posted by Eddie Shore on 9 Feb 2005, 9:15 p.m.,in response to message #5 by Mike (Stgt) No calcualtor will ever be perfect and every machine has its quirks. "That leads me to an idea: how about a bare calculator (with a screen as big as an HP200LX or a Sharp Zaurus) with an simple programming language (or two) to do your own solutions and enough storage to download ready to use application packs. That way this calculator will grow with your demands/tasks. In the end you'll have a very personal tool that fits _exactly_ *your* needs." That is why I love programming. Making the calculator (and Excel, Word, the VCR) fit the user's needs. Having memory is more important that too many extras.

 Re: Some 49G+ LimitationsMessage #8 Posted by John Smitherman on 9 Feb 2005, 10:11 p.m.,in response to message #3 by Mike (Stgt) Thanks Luiz and Mike for teaching me about some of the 49G+ features. A happier and less frustrated user, John

 Re: Some 49G+ LimitationsMessage #9 Posted by Ralph on 10 Feb 2005, 2:33 p.m.,in response to message #8 by John Smitherman After reading all these discursions about how calculators handle 'extremes' these are the results for the following functions on a HP65, HP41CX and HP48G: a)1000! HP65(9.999999999E99) wich is interpreted as an overflow HP41CX(OUT OF RANGE) HP48G the same as stated for the HP49G+ b)1.5! HP65 -display blinks HP41CX (DATA ERROR) HP48G the same as stated for the HP49G+ c)asin(acos(atan(tan(cos(sin(7)))))) HP65(7.003719916) worse than HP49G+, HP48G and HP41CX HP41CX(6.999519575) worse than HP49G+, HP48G HP48G the same as stated for the HP49G+ Certainly the results improve with the newer models, but a strange thing for case c), HP41CX gives a result that aproaches from under the correct value, instead of all others.

 Re: Some 49G+ LimitationsMessage #10 Posted by John Smitherman on 10 Feb 2005, 11:14 p.m.,in response to message #9 by Ralph Hi Ralph. You can find the results of more comparisons at: http://www.voidware.com/calcs/torturetest.htm Regards, John

 Re: Some 49G+ LimitationsMessage #11 Posted by Ralph on 11 Feb 2005, 4:16 a.m.,in response to message #10 by John Smitherman Thanks John, great site! Ralph

 less frustrationMessage #12 Posted by Mike (Stgt) on 11 Feb 2005, 6:04 a.m.,in response to message #8 by John Smitherman Nice to see my append was helpful. But I am a bit frustrated: An HP-49 or TI-89 has more math built in than I am ever able to understand. That's not a problem of those calculators, it's just an issue of capabilities: I will not in my lifetime understand such papers (alas PostScript only). No calculator may help me to see all coherencies that make up our world. But what I miss on machines like the HP-49 is a system that calculates with errors, internal errors too. So the answer of asin(acos(atan(tan(cos(sin(7)))))) would be 7 +/- 0.00xy% or an interval [7-e; 7+e]. There is (was?) a FORTRAN lib for that (cant remember neither its name nor source). I once had a system on the HP-41 to calculate with errors (no-no, not with internal errors of cause, and not with intervals). It was for estimating errors when calculating with measured values. I just had to program the formula, the derivation for each variable was done with the PPC-ROM. Ciao.....Mike (not having a full imagination of this world does not stop me dreaming...)

Go back to the main exhibit hall