MVSFORUMS.com

seekaysk · Beginner Joined: 31 Aug 2007 Posts: 49 Topics: 15

Hi guys,

I tried to find this on NET as well as MVSFORUMS search but could not get anything substantial.

I want to find

a) how does Cobol calculate a Negative base with fractional components ? or does it even calculate it or not ?

e.g. (-5)**(0.2593)

b) If expecting above answer is illogical (as what is suggested in the below MVS message), what should we do in practical terms ? Can anybody suggest me that not coding ON SIZE ERROR and putting negative sign in the result, is it logical ?

e.g. if Z = (5)**(0.2593), then actual result = -Z. is this logical ?

fyi. I found this one message on MVS which gives at run time:

******************************************************
IGZ0048W
IGZ0048W A negative base was raised to a fractional power in an exponentiation expression in program program-name at displacement displacement. The absolute value of the base was used.

Explanation: A negative number raised to a fractional power occurred in a library routine.

The value of a negative number raised to a fractional power is undefined in COBOL. If a SIZE ERROR clause had appeared on the statement in question, the SIZE ERROR imperative would have been used. However, no SIZE ERROR clause was present, so the absolute value of the base was used in the exponentiation.

Programmer Response: Ensure that the program variables in the failing statement have been set correctly.

System Action: No system action was taken.
******************************************************
_________________
Thanks.

dbzTHEdinosauer · Posted: Thu Oct 23, 2008 8:18 am Post subject:

I would think this is something fortran would handle better than COBOL, and that is were you should go. FORTRAN is provided by IBM as a programming language.

FORTRAN - formula translation

COBOL - common business oriented language.
_________________
Dick Brenholtz
American living in Varel, Germany

jsharon1248 · Posted: Thu Oct 23, 2008 10:41 am Post subject:

I'd do some research. I don't know that what you're trying to do is valid. I tried (-5**.2593) in a calculator and in Excel, and both returned errors indicating that this is not valid. I'm not a mathemetician, so I can't be sure, but it seems to me that what you're trying to do doesn't make sense. COBOL is just telling you that it's doing the best it can with what you provided, so if that's not good enough, you'll need to decide how you want to handle it. Probably checking for the negative base and issuing your own warning is the best option.

sriramla · Beginner Joined: 22 Feb 2003 Posts: 74 Topics: 1

As you know, (-5)**1 gives you a negative number as the result. But (-5)**2 gives you a positive number. In general, negative number raised to the power of odd number will give a negative number as the result; raised to even number will give a positive number as the result.

Having said that (-5)**0.5 (In other words, square root of -5) is a complex number and it makes very little sense in business applications. That is why you get this warning message. Regarding your question "if Z = (5)**(0.2593), then actual result = -Z. is this logical?" this is not correct.

May be you want to check how the result of this computation is being used and make a decision.

seekaysk · Beginner Joined: 31 Aug 2007 Posts: 49 Topics: 15

Thanks for the replies. I am at a financial shop. The equation I gave is used for Annualized formula calculation. Hence, having -5 as a Return is absolutely valid in a financial shop.
_________________
Thanks.

jsharon1248 · Posted: Thu Oct 23, 2008 1:09 pm Post subject:

If it's true that when the base is negative, the result will be a negative return, then you should use the absolute value of the base, raise it to the appropriate exponent, and then multiply the result by -1.

seekaysk · Beginner Joined: 31 Aug 2007 Posts: 49 Topics: 15

thanks jsharon1248 and all. I am still not sure whether multiplying the result by -1 is the exact logical thing to do...and that's the reason i had this thread so that I can know the experts views. I am still looking forward to if somebody has something else what the above guys could not post. Else I will close this thread concluding it's application specific to get the workaround of multiplying the result by -1 or just send warning to user.

thanks to all. As always, its a pleasure to know the views of the best in the industry.
_________________
Thanks.

Dibakar · Posted: Mon Oct 27, 2008 8:30 pm Post subject:

seekaysk,

I will go with what srirmala said. So there is no point in going for negative base unless you want to deal with imaginary amounts. This is what google calculator shows - http://www.google.com/search?sourceid=navclient&ie=UTF-8&rlz=1T4GPCK_enUS291US292&q=%28%2d5%29+%2a%2a%280%2e2593%29%3d

But still I will go with jsharon's logic. To me there is no negative amount, it is either credit or debit. So I will take the exponent of positive base and put the sign in front to show who owes the money. I hope this makes sense in your case too.

Diba.