I am trying to use the following computation: 232582656(232582657-1) n = (2**32582656) * (2**32582657 - 1) I am getting the msg: perf.rb:5: warning: in a**b, b may be too big Infinity Any help will be appreciated. Thank you Victor

on 2007-05-04 21:12

on 2007-05-04 21:37

You do realize the magnitude of number you're trying to calculate? There IS a limit to how big even BigNum can go before deciding to crap out. I'd like to see someone try this in Lisp. If that interpreter can't handle it, then no language can. Jason

on 2007-05-04 22:45

On Sat, May 05, 2007 at 02:10:58AM +0900, Victor R. wrote: > I am trying to use the following computation: 232582656(232582657-1) > > n = (2**32582656) * (2**32582657 - 1) Homework question? Brute force probably isn't the way :-)

on 2007-05-04 23:34

Brian, Actually, NO. This is not a homework questions. It is, however and unfortunately, an issue of my deficiency and lack of knowledge of the language. For pure and simple intellectual curiosity, I am attempting to write a ruby pgm to find perfect numbers. The largest perfect number found to this date, if I believe website http://amicable.homepage.dk/perfect.htm is: (2**32582656) * (2**32582657 - 1). Been the neophyte that I am, I thought about picking it up from there and play with it, since I have at my disposal a very large IBM multiprocessor environment. I know that Brute force is not the way, but I don't have a nice algorithm to use. Thank you Victor

on 2007-05-04 23:35

On May 4, 2007, at 10:36 AM, Jason R. wrote: > You do realize the magnitude of number you're trying to calculate? > There IS > a limit to how big even BigNum can go before deciding to crap out. > I'd like > to see someone try this in Lisp. If that interpreter can't handle > it, then > no language can. Python 2.5 appears to have handled the expression easily enough (3-4 seconds on my iMac Intel Core Duo). Converting it to a string for display seems to be taking quite a long time, which isn't too surprising since it should have nearly 2 million decimal digits (according to log10(n)). -Mark

on 2007-05-05 00:10

On Sat, May 05, 2007 at 04:34:01AM +0900, Victor R. wrote: > The largest perfect number found to this date, if I believe website > http://amicable.homepage.dk/perfect.htm is: (2**32582656) * (2**32582657 - > 1). Then that's the number, represented in a compact form. > I know that Brute force is not the way, but I don't have a nice algorithm to > use. But what are you trying to achieve? Simply to display this number in its direct decimal form? If my rusty maths is correct, that number is a little under 2**(32582656 + 32582657) which is roughly 10 ** ( (32582656 + 32582657) * log(2)/log(10) ) ~= 10 ** 19616713 So there are nearly 20 million digits in the answer...

on 2007-05-05 00:26

For the record, I am running ruby 1.86 under AIX 5.3. If the answer is that it can be done on Ruby, that's an OK answer also. If this is the case, I start playing with Python, at least for this problem. Thanks Victor

on 2007-05-05 01:03

--- Victor R. <removed_email_address@domain.invalid> wrote: > For the record, I am running ruby 1.86 under AIX > 5.3. > If the answer is that it can be done on Ruby, that's > an OK answer also. If > this is the case, I start playing with Python, at > least for this problem. > > Thanks > > Victor p 2048 ** 31819 - 1024 ** 31819 or a = 1 32582656.times { a *= 2 } b = a 32582656.times { b *= 2 } p b - a

on 2007-05-05 01:13

```
I said:
> p 2048 ** 31819 - 1024 ** 31819
Oops. Ignore this one. I don't know what I was
thinking :)
```

on 2007-05-05 06:13

Todd, thank you for the example. I truly appreciate it. I'll give it a try. Victor

on 2007-05-05 18:34

--- Victor R. <removed_email_address@domain.invalid> wrote: > Todd, thank you for the example. I truly appreciate > it. I'll give it a try. Don't get too excited. On my system (2.4GHz Intel), it looks like it would take about 2.5 hours to compute (I didn't test it to see if it would even work). Also, it is most likely faster to multiply by larger numbers than 2. 2**32582656 == 4**16291328 == 16**8145664 == 256**4072832 == 65536**2036416 and so on Broken down, it's: 2**32582656 == 2**(2**10 * 3**2 * 3541) == ( (2**(2**10))) ** (3**2) ) ** 3541 I'm guessing now matter how you cut it up, you will get the same error, though :( Maybe ruby chokes at around a million digits.

on 2007-05-05 19:35

--- Todd B. <removed_email_address@domain.invalid> wrote: > > --- Victor R. <removed_email_address@domain.invalid> wrote: > > > Todd, thank you for the example. I truly > appreciate > > it. I'll give it a try. > > Don't get too excited. On my system (2.4GHz Intel), > it looks like it would take about 2.5 hours to > compute # The next 3 lines compute 2**32582656 in less than a minute on my system: a = 2 ** (2**10 * 9) b = 1 3541.times { b*=a } # a**3541, which equals 2**32582656, will fail to # execute in ruby, but the 3541.times works # this next line, however, s = b.to_s # takes a _very_ long time # similar (but much longer, I think) to Python's # rendering of b into a string.