▼
Posts: 1,619
Threads: 147
Joined: May 2006
Just in case you missed it:
http://groups.google.com/group/comp.sys.hp48/browse_frm/thread/5e7ef329b61c0497/d08b731b023d922b#d08b731b023d922b
Joe Horn said,
Quote:
It would *seem* that all integers raised to the 4th power contain at
least one digit less than 5. For example, 26 to the 4th power is
456976, which contains a 4. However, there does exist a positive
integer X such that X^4 contains no digits less than 5.
Mini-challenge: Write a User-RPL program that finds the only (?)
positive integer X such that X^4 contains no digits less than 5.
Winner: Fastest solver that doesn't cheat.
-Joe-
▼
Posts: 412
Threads: 40
Joined: Mar 2006
Found it! Unique solution, at least up to 1E6.
Using Free42 (will not work on regular HP42):
00 { 60-Byte Prgm }
01>LBL "MC"
02 1
03 STO 00
04>LBL 01
05 RCL 00
06 ENTER
07 ×
08 ENTER
09 ×
10>LBL 02
11 RCL ST X
12 10
13 MOD
14 5
15 X>Y?
16 GTO 03
17 Rv
18 -
19 10
20 ÷
21 X#0?
22 GTO 02
23 "FOUND "
24 ARCL 00
25 AVIEW
26 STOP
27>LBL 03
28 RCL 00
29 1
30 +
31 STO 00
32 1E6
33 X>Y?
34 GTO 01
35 .END.
J-F
▼
Posts: 2,761
Threads: 100
Joined: Jul 2005
Since n^4 ends in 1 for n ending in 1, 3, 7 and 9 and final 5 appear to always introduce either 0 or 2 in the answer, you can do it in half the time by replacing the '1' in lines 02 and 29 with '2'.
Gerson.
Posts: 2,761
Threads: 100
Joined: Jul 2005
Is ENTER * faster than X^2?
Even numbers ending in 0 can be handled by inserting X=0? and GTO 03 between lines 13 and 14, but the improvement (if any) is hard to notice on Free42.
Gerson.
▼
Posts: 412
Threads: 40
Joined: Mar 2006
You're right X^2 is just simpler...
Free42 on PC, especially the binary version, is fast enough to find the solution in less than one second, but the decimal version is needed to check all numbers up to 1e6 thanks to the 25-digit accuracy.
This is what I mostly appreciate with Free42: fast speed and enhanced accuracy versus the HP42S, and I like to be able to choose between speed (binary version) or accuracy (decimal version) depending on the problem.
If only there was a Free71...
J-F
Posts: 556
Threads: 9
Joined: Jul 2007
Also, why not:
1
STO+ 00
RCL 00
42S on Iphone solves it almost instantly.
Today I noticed 42S showing:
00 { 31-Line Prgm }
rather than bytes, I guess it makes no difference using different methods of calculating x ^ 4 would change program running time.
Posts: 63
Threads: 8
Joined: Dec 2009
May I offer a program for the 50g as a "newb", not a fast, efficient or clever one, but as a service to others like me just
new to RPL. I'm using the fact as given in the original post that the number can't end in 1,3,7 or 9, and the hint that it can be
proven that 5 is excluded too. If this program wouldn't have found a solution in a reasonable time (and the real answer did end in 5), it could have been easily re-done to search for multiples of 5 (and more quickly). So this program churns through the even numbers in just under 20 minutes. Just store 0 in variable 'N' and then the answer is left there at the end. Again, pretty much a brute force program just to demonstrate some functions for my fellows wondering how one might begin to solve this challenge!
<< DO 2 'N' STO+ 'N^4'
EVAL ->STR DUP SIZE
-> s
<< 1 CF 1 s FOR i
DUP HEAD SWAP TAIL SWAP
OBJ-> IF 4 <= THEN
1 SF END
NEXT DROP >>
UNTIL 1 FC?
END >>
Happy programming! Regards, Glenn
▼
Posts: 3,283
Threads: 104
Joined: Jul 2005
You can of course exclude the even multiples of 5 but why the numbers simply ending in 5? I'd like to see a proof.
▼
Posts: 63
Threads: 8
Joined: Dec 2009
Looking at the beginnings of a proof for excluding 5, the "10a + b"
term is suggested. If you multiply out (10a + b)^4 and simplify,
and substitute b=5, you get the following arrangement:
10000a^4 + 10000a^3 + 25000a^2 + 5000a + 625. Notice if a = 1, you
get 50625 (15^4). The "proof" comes in showing that the fourth-
largest place is always 0. This is because the last two terms in the expansion have a 5 in the thousands place, and when a is even
the two 5's become 0's in the sum, and the other terms all involve higher places. When a is odd, a^2 is also odd, and those two terms
keep a 5 so then the two 5's add up to 0 again. Is this a formal
proof? It seems to hold up. Anyway, a great challenge and it's
what this forum is all about. Regards, Glenn
▼
Posts: 63
Threads: 8
Joined: Dec 2009
Sorry, please see msg.#13 :-)
Posts: 260
Threads: 0
Joined: Oct 2008
Hello.
This challenge triggers my curiosity. I am searching for a clever mathematical way of this interesting problem, but still clue less.
Meanwhile, my attempt of a quite "brute force" search program lead me to the following RPL program that I run on my HP-28S (it is so simple that it will run on all RPL from 28C to 50g):
« DO @ Do main loop
2 + @ looking only for even numbers
DUP 1 DISP @ Optional display current number – slows research
DUP SQ SQ @ Compute x^4
->STR @ Convert to string
0 48 52
FOR c @ 48 to 52 are ASCII code for "0" to "52"
OVER c CHR
POS @ test if digit less than 5 present in "x^4"
+ @ equivalent to OR
NEXT
SWAP DROP @ remove "x^4" after tests
UNTIL
NOT @ sum of POS is 0 if none of the digits were found
END
»
'PRO1' STO
Usage: Enter starting number 0, the program simply stop leaving the answer in stack. Continue searching by simply re-running it, the previous value is kept in the stack.
As every one may have notice, it is really close to the one Glenn Shields post previously.
But, as same one may have notice, it is unable to search for number greater than 1000 !
Because of the 12 digit precision of HP-28S, the x^4 value is rounded and digits can’t be tested to the missing unit position !
1000^4 = 1000000000000 which is 1.E+12
So I suggest that HP49/50g users use full precision arithmetic capabilities, and HP28/48 users binary number which may be efficient up to 65536 since largest binary of 64 bits is 2^64 which quadric root is 2^16.
« 64 STWS DEC @ switch to decimal binary word full length
DO
#2d + @ looking only for even binary numbers
DUP 1 DISP @ Optional display current number – slows research
DUP
DUP * DUP * @ Compute x^4: SQ not available with binary numbers
->STR
0 48 52
FOR c @ Search for 0 to 4 digit in "x^4"
OVER c CHR
POS +
NEXT
SWAP DROP @
UNTIL
NOT @ sum is 0 for any solution
END
»
My HP-28S found the solution in 7 min (~6min without DUP 1 DISP).
No other solution was found before reaching 65536 after a 2h30 run.
Edited: 28 Mar 2010, 3:59 a.m. after one or more responses were posted
▼
Posts: 63
Threads: 8
Joined: Dec 2009
Wow, C.Ret, I tried your program on the 50g (but without the second DUP in the third line) and got an amazing 1:59 time. Then I removed the DUP 1 DISP and got an even more amazing 1:48 (but it was more fun with the display). This is the cleverness I was looking for, great use of POS, CHR and NOT, these just blitz the other tests and string manipulations. This type of programming was
given in examples in the user guides but it's great to see them applied in other ways. Great!!
▼
Posts: 63
Threads: 8
Joined: Dec 2009
DUH!!!! Didn't that just prove that our number will always end in
0625? Now there are two nails for that coffin. Yup.
▼
Posts: 3,283
Threads: 104
Joined: Jul 2005
After a quick look at your formula I just decided to ignore the rest of your reasoning. Enough zeros in the factors to see the obvious. ;)
Posts: 320
Threads: 59
Joined: Dec 2006
Okay, I know this is an HP calculator challenge, but I had to give it a try with Mathematica. Here's the code:
i = 2; While[Min[IntegerDigits[i^4]] < 5, i = i + 2]
It found the solution in 0.009491 seconds. Darn quick. I also let it run until i was about 26,000,000; no more solutions found.
CHUCK
Edited: 28 Mar 2010, 1:20 p.m.
▼
Posts: 2,761
Threads: 100
Joined: Jul 2005
Quote:
I also let it run until i was about 26,000,000; no more solutions found.
Isn't this just a matter of probability? I think it is very unlikely new solutions are found as i increases although this is not impossible. For instance, let's investigate 5-digit fourth powers: We expect to find about eight of them, sqrt(sqrt(99999)) - sqrt(sqrt(10000)) = 7.78, of which only 40% will be candidates to a solution (the ones whose bases end in 2, 4, 6 and 8), that is, about three candidates ( 7.78 * 0.4 = 3.11). In fact, for this case, we have eight powers, 10^4 through 17^4, but only 12^4, 14^4 and 16^4 might be a solution. The probability of an individual digit is equal or greater than five is 0.5 and the probability a five-digit power is a solution is 0.5^5 = 1/32. So, the probability a solution is found among our three candidates is 3/32 or about one chance in eleven.
Likewise, if we check all 14-digit fourth powers we'll find the chances are smaller: about one in seventeen. But we've been lucky this time, as we know.
((sqrt(sqrt(1e15)) - sqrt(sqrt(1e14)))*0.4)/(0.5^-14) = 16.64
Of course, if new properties concerning these powers are discovered, this result should be reconsidered and they might even lead to a different conclusion.
Gerson.
▼
Posts: 320
Threads: 59
Joined: Dec 2006
Definitely a probability problem. 26000000^4 has 30 digits in it. The probability that none are less than 5 is P(none are<5)=0.5^30 =9.3x10^-10, and it only gets worse from there. But again, it's not impossible, but highly unlikely. :)
Edited: 28 Mar 2010, 8:55 p.m. after one or more responses were posted
▼
Posts: 2,761
Threads: 100
Joined: Jul 2005
Considering there's a lot more squares of squares with 30 digits in them, the overall probability to find a solution in this group is 9.17e-3 or about 1/109 (if the distribution of the digits follows a random order, which is not exactly the case). But you're right, that's still a dim probability and it dwindles as we go further.
Edited: 28 Mar 2010, 7:28 p.m.
Posts: 709
Threads: 104
Joined: Nov 2005
Quote:
Definitely a probability problem. 26000000^4 has 30 digits in it. The probability that none are less than 5 is P(none are<5)=0.5^30 =9.3x10^-10, and it only gets worse from there. But again, it's not impossible, but highly unlikely. :)
That would only be true if the digits were completely random and independent of the number being raised to the fourth power. However, any number ending in an odd digit has a probability of 1 for having at least one small digit when raised to the power of four.
It is certainly true that a RANDOM 30 digit number among all possible 30 digit numbers is unlikely to have no small digit.
Posts: 564
Threads: 72
Joined: Sep 2005
Having a couple of hours free today I thought I give it a try with MCODE for the 41. Alas, we don't have enough digits precision (10) in the normal format and I'm too lazy right now to see if 13 digits would be enough (as noone has posted the actual solution number). The below MCODE program tests until X=316 before it crashes out without having found a solution. Is that correct? Maybe I get around to rewriting it for 13 digits but looking at the code, its not trivial and run time wise, this is going to be slow as well. But would be great to know how high the solution number is and if it can be found with 13 digit accuracy at all...
For those enjoying MCODE, here is the code:
To speed up the expensive part - calculating x^4, I first calculate which power of 2 is smaller than x (and then smaller than x^2) as the NUT can do multiplication by 2 very quickly so that should speed up the power calculation. It also checks for the starting number not ending in 1,3,5,7,9,0. All of them seem obvious, although it took a little while for me to make sure 5 is correct as well.
Cheers,
Peter
;=============================================================================================
; X4 - Find a number x who's 4th power has no digit less than 5
;
; Takes a starting number from x
; moves into a mantissa only format, right justified
; checks that does not end in 1,3,5,7,9,0
; calcs x^2
; calcs x^4
; Checks if any digits <5
; If no, putputs result to X-reg
; If yes, read x, increase by one, loop
;©pplatzer 2010
;=============================================================================================
.NAME "X4" ;Main Code
*0012 0B4 #0B4 ; "4"
*0013 018 #018 ; "X"
0014 1A0 [X4] A=B=C=0 ;tabula rasa
0015 2A0 SETDEC ;all calcs in decmode
0016 35C R= 12 ;spot of first digit in C[M]
0017 0F8 READ 3(x) ;get start number
0018 2FA ?C#0 M ;check that not 0
0019 06B JNC [StartW1] +13 0026 ;no -> start with 1
001A 2F6 ?C#0 XS ;check that no negative exponent
001B 05F JC [StartW1] +11 0026 ;no -> start with 1
001C 31C R= 1
001D 2E2 ?C#0 @R ;check that exp < 10
001E 047 JC [StartW1] +8 0026 ;no -> start with 1
001F 05E C=0 MS ;just in case
0020 39C R= 0
0021 2A2 C=-C-1 @R ;complement exponent
0022 262 [RJus] C=C-1 @R ;right justify number
0023 02F JC [RJusD] +5 0028 ;if underflow, we are done
0024 3DA RSHFC M
0025 3EB JNC [RJus] -3 0022
0026 04E [StartW1] C=0 ALL ;invalid start number
0027 23A C=C+1 M ;starting point = 1
0028 046 [RJusD] C=0 S&X ;clean up from shifting loop
0029 27A C=C-1 M ;as [NxtTry] does a C=C+1 first
002A 0E8 WRIT 3(x) ;write start to X
002B 1A0 [NxtTry] A=B=C=0 ;tabula rasa
002C 270 RAMSLCT ;select first ram chip
002D 0F8 READ 3(x) ;read last try #
002E 23A C=C+1 M ;next digit
002F 289003 ?CGO [ERROF] 00A2
0031 0E8 WRIT 3(x) ;
0032 01C [CheckDigs] R= 3 ;check that last dig is not 1,3,5,7,9,0
0033 10E A=C ALL ;save number into A
0034 2E2 ?C#0 @R ;check for 0
0035 3B3 JNC [NxtTry] -10 002B
0036 222 C=C+1 @R ;check for 9
0037 3A7 JC [NxtTry] -12 002B
0038 222 C=C+1 @R
0039 222 C=C+1 @R ;check for 7
003A 38F JC [NxtTry] -15 002B
003B 222 C=C+1 @R
003C 222 C=C+1 @R ;check for 5
003D 377 JC [NxtTry] -18 002B
003E 222 C=C+1 @R
003F 222 C=C+1 @R ;check for 3
0040 35F JC [NxtTry] -21 002B
0041 222 C=C+1 @R
0042 222 C=C+1 @R ;check for 1
0043 347 JC [NxtTry] -24 002B
0044 04E [StartX2] C=0 ALL ;good start number in A
0045 09A B=A M ;save number into B
0046 23A C=C+1 M ;C:= 1
0047 1A6 A=A-1 S&X
0048 070 [Get2Pwr] N=C ;save this power to N
0049 166 A=A+1 S&X ;find which power of 2 < #
004A 1FA C=C+C M ;2, 4, 8, 16, ...
004B 31A ?A<C M ;still # smaller than 2^n
004C 3E3 JNC [Get2Pwr] -4 0048 ;yes, keep looping
004D 1A6 A=A-1 S&X ;so that we can jump on 0 underflow
004E 2EF JC [NxtTry] -35 002B ;if underflow, we need to try next number
004F 0B0 C=N ;get back last power of 2 that was smaller thanb #
0050 1DA A=A-C M ;those are the multiplication left after the 2^n
0051 0DA C=B M ;get back #
0052 1FA [DoPwr2] C=C+C M ;start multi
0053 289003 ?CGO [ERROF] 00A2 ;if overflow -> out of range
0055 1A6 A=A-1 S&X ;count down n from 2^n
0056 3E3 JNC [DoPwr2] -4 0052
0057 07A [DoRestM] A<>B M ;get # back
0058 21A C=A+C M ;do rest multi
0059 289003 ?CGO [ERROF] 00A2
005B 07A A<>B M
005C 1BA A=A-1 M
005D 3D3 JNC [DoRestM] -6 0057
005E 07A A<>B M ;correct one to many additions
005F 0BA C<>A M
0060 25A C=A-C M ;correct result for x^2
;------------------------------------ Stepping Stone -----------------------------
0061 013 JNC [StartX4] +2 0063
0062 24B [NxtTryJP] JNC [NxtTry] -55 002B
;------------------------------------ Stepping Stone -----------------------------
0063 046 [StartX4] C=0 S&X ;clean up
0064 10E A=C ALL
0065 08E B=A ALL
0066 04E C=0 ALL ;find power of 2
0067 23A C=C+1 M
0068 1A6 A=A-1 S&X
0069 070 [4Get2Pwr] N=C ;save this power to N
006A 166 A=A+1 S&X ;find which power of 2 < #
006B 1FA C=C+C M ;2, 4, 8, 16, ...
006C 31A ?A<C M ;still # smaller than 2^n
006D 3E3 JNC [4Get2Pwr] -4 0069 ;yes, keep looping
006E 1A6 A=A-1 S&X ;so that we can jump on 0 underflow
006F 39F JC [NxtTryJP] -13 0062 ;if underflow, we need to try next number
0070 0B0 C=N ;get back last power of 2 that was smaller thanb #
0071 1DA A=A-C M ;those are the multiplication left after the 2^n
0072 0DA C=B M ;get back #
0073 1FA [4DoPwr2] C=C+C M ;start multi
0074 289003 ?CGO [ERROF] 00A2 ;if overflow -> out of range
0076 1A6 A=A-1 S&X ;count down n from 2^n
0077 3E3 JNC [4DoPwr2] -4 0073
0078 07A [4DoRestM] A<>B M ;get # back
0079 21A C=A+C M ;do rest multi
007A 289003 ?CGO [ERROF] 00A2
007C 07A A<>B M
007D 1BA A=A-1 M
007E 3D3 JNC [4DoRestM] -6 0078
007F 07A A<>B M ;correct one to many additions
0080 0BA C<>A M
0081 25A C=A-C M ;correct result for x^4
0082 00E [CheckR] A=0 ALL ;now check if good result
0083 0AE A<>C ALL ;# into A
0084 130105 LDIS&X 105 ;check for 5
0086 33C RCR 1 ;5 into C[MS]
0087 056 C=0 XS ; 008 into C[S&X]
0088 3EE [ShftNr] LSHFA ALL ;move # into A[MS]
0089 266 C=C-1 S&X
008A 35E ?A#0 MS ;are we there yet?
008B 3EB JNC [ShftNr] -3 0088
008C 31E [TstDig] ?A<C MS ;is digit < 5
008D 2AF JC [NxtTryJP] -43 0062 ;if yes, try next number
008E 3EE LSHFA ALL ;shift in next digit
008F 266 C=C-1 S&X ;count down number of digits
0090 3E3 JNC [TstDig] -4 008C ;loop to test next digit
0091 1A0 [FoundIt] A=B=C=0 ;if getting here, we found the number
0092 0F8 READ 3(x) ;now move to correct number formnat
0093 35C R= 12
0094 2E2 [FrmtNr] ?C#0 @R
0095 027 JC [X4Done] +4 0099
0096 2FC RCR 13 ; =-1
0097 166 A=A+1 S&X
0098 3E3 JNC [FrmtNr] -4 0094
0099 0A6 [X4Done] A<>C S&X
009A 39C R= 0 ;only exp up to 9 possible anyway
009B 2A2 C=-C-1 @R
009C 0E8 WRIT 3(x)
009D 3E0 RTN
▼
Posts: 564
Threads: 72
Joined: Sep 2005
okay,I'm such a sucker for these things, it is pathetic. But so much fun...
Anyway, I could not rest but wanted to see if we can entice the 41 to also spit out an answer in non-geological times. The code below uses all 14 internal digits we have to represent numbers, better multiplication logic and takes advantage of the fact that we can exclude from the interim result x^2 all numbers that start with 1, 4, 5, 6 as they will yield a first number <5. On V41 and normal speed setting (which feels close to a 41cx or maybe a bit faster) it gives us the answer in about 45 seconds. If we remove the tests for X^2 it takes about 1minute. Not bad for the little guy...
Cheers
Peter
;=============================================================================================
; X43 - Find a number x who's 4th power has no digit less than 5
;
; Takes a starting number from x
; uses right justified format in C[S&X & M]
; only uses odd numbers (ending in ) 2,4,6,8 (as 0,1,3,5,7,9 are out).
; calcs x^2
; Exclude x^2 that start in 1,4,5,6
; calcs x^4
; Checks if any digits <5
; If no, outputs result to X-reg
; If yes, read x, increase by two, loop
;©pplatzer 2010
;=============================================================================================
.NAME "X43" ;Main Code
*0103 0B3 #0B3 ; "3"
*0104 034 #034 ; "4"
*0105 018 #018 ; "X"
0106 1A0 [X43] A=B=C=0 ;tabula rasa
0107 2A0 SETDEC ;all calcs in decmode
0108 35C R= 12 ;spot of first digit in C[M]
0109 0F8 READ 3(x) ;get start number
010A 2FA ?C#0 M ;check that not 0
010B 06B JNC [StartW23] +13 0118 ;no -> start with 2
010C 2F6 ?C#0 XS ;check that no negative exponent
010D 05F JC [StartW23] +11 0118 ;no -> start with 2
010E 31C R= 1
010F 2E2 ?C#0 @R ;check that exp < 10
0110 047 JC [StartW23] +8 0118 ;no -> start with 2
0111 05E C=0 MS ;just in case
0112 39C R= 0
0113 2A2 C=-C-1 @R ;complement exponent
0114 262 [RJus3] C=C-1 @R ;right justify number
0115 037 JC [RJusD3] +6 011B ;if underflow, we are done
0116 3DA RSHFC M
0117 3EB JNC [RJus3] -3 0114
0118 04E [StartW23] C=0 ALL ;invalid start number
0119 23A C=C+1 M ;starting point = 2
011A 23A C=C+1 M ;starting point = 2
011B 046 [RJusD3] C=0 S&X ;clean up from shifting loop
011C 03C RCR 3 ;now check that even number
011D 3D8 C<>ST ;shift C[01] into flags
011E 38C ?FSET 0 ;odd number?
011F 027 JC [OddNr3] +4 0123
0120 3D8 C<>ST ;bring back the number
0121 26E C=C-1 ALL ;as [NxtTry] does 2x C=C+1 first
0122 013 JNC (conEv3) +2 0124
0123 3D8 [OddNr3] C<>ST ;bring back the number
0124 26E (conEv3) C=C-1 ALL ;as [NxtTry] does 2x C=C+1 first
0125 0E8 WRIT 3(x) ;write start to X
0126 1A0 [NxtTry3] A=B=C=0 ;tabula rasa
0127 270 RAMSLCT ;select first ram chip
0128 0F8 READ 3(x) ;read last try #
0129 22E C=C+1 ALL ;next even number
012A 22E C=C+1 ALL
012B 0E8 WRIT 3(x) ;
012C 39C [Check03] R= 0 ;check if it ends in 0
012D 2E2 ?C#0 @R ;is last digit <> 0
012E 3C3 JNC [NxtTry3] -8 0126 ;no -> get next number
012F 00E [StartX23] A=0 ALL ;Clean A (accumulator for multi)
0130 0EE0CE B=C ALL ;put number into B as well
0132 2DC R= 13 ;C[MS]
0133 3D4 [MLoop3] R=R-1 ;move pointer
0134 2D4 ?R= 13 ;are we done?
0135 037 JC [DoneX23] +6 013B ;Yes, we have done all digits
0136 3EE LSHFA ALL ;=x10
0137 262 [MCnt3] C=C-1 @R ;count down digits of number in C
0138 3DF JC [MLoop3] -5 0133 ;if 0, no multiplication
0139 12E A=A+B ALL ;if <>0, do multiplication
013A 3EB JNC [MCnt3] -3 0137
013B 08E [DoneX23] B=A ALL ;result into B
013C 0CE C=B ALL ;result into C
013D 00E A=0 ALL ;clear A
013E 2DC R= 13 ;check if we have a start in 1,4,5,6
013F 3D4 [Fnd1stX2] R=R-1 ;find first dig
0140 2E2 ?C#0 @R ;is this the first dig?
0141 3F3 JNC [Fnd1stX2] -2 013F ;no, keep looking
0142 262 C=C-1 @R ;Check for 1
0143 2E2 ?C#0 @R ;
0144 313 JNC [NxtTry3] -30 0126
0145 262 C=C-1 @R
0146 262 C=C-1 @R
0147 262 C=C-1 @R ;Check for 4
0148 2E2 ?C#0 @R ;
0149 2EB JNC [NxtTry3] -35 0126
014A 262 C=C-1 @R ;Check for 5
014B 2E2 ?C#0 @R ;
014C 2D3 JNC [NxtTry3] -38 0126
014D 262 C=C-1 @R ;Check for 6
014E 2E2 ?C#0 @R ;
014F 2BB JNC [NxtTry3] -41 0126
0150 0CE C=B ALL ;all good, do X^4
0151 2DC R= 13 ;prep for next multi
0152 3D4 [MLoop23] R=R-1 ;move pointer
0153 2D4 ?R= 13 ;are we done?
0154 037 JC [DoneX43] +6 015A ;yes, we have done all digits
0155 3EE LSHFA ALL
0156 262 [MCnt23] C=C-1 @R
0157 3DF JC [MLoop23] -5 0152
0158 12E A=A+B ALL
0159 3EB JNC [MCnt23] -3 0156
015A 00B [DoneX43] JNC [CheckR3] +1 015B ;result in A:= x^4
015B 04E [CheckR3] C=0 ALL ;now check if good result
015C 130135 LDIS&X 135 ;check for 5 & counter for 13 digits
015E 33C RCR 1 ;5 into C[MS]
015F 01B JNC [Chk1st3] +3 0162
0160 3EE [ShftNr3] LSHFA ALL ;move # into A[MS]
0161 266 C=C-1 S&X
0162 35E [Chk1st3] ?A#0 MS ;are we there yet?
0163 3EB JNC [ShftNr3] -3 0160
0164 31E [TstDig3] ?A<C MS ;is digit < 5
0165 20F JC [NxtTry3] -63 0126 ;if yes, try next number
0166 3EE LSHFA ALL ;shift in next digit
0167 266 C=C-1 S&X ;count down number of digits
0168 3E3 JNC [TstDig3] -4 0164 ;loop to test next digit
0169 1A0 [FoundIt3] A=B=C=0 ;if getting here, we found the number
016A 0F8 READ 3(x) ;now move to correct number formnat
016B 1BC RCR 11 ;move it into C[M]
016C 35C R= 12
016D 2E2 [FrmtNr3] ?C#0 @R
016E 027 JC [X4Done3] +4 0172
016F 2FC RCR 13 ; =-1
0170 166 A=A+1 S&X
0171 3E3 JNC [FrmtNr3] -4 016D
0172 0A6 [X4Done3] A<>C S&X
0173 39C R= 0 ;only exp up to 9 possible anyway
0174 2A2 C=-C-1 @R
0175 0E8 WRIT 3(x)
0176 3E0 RTN
▼
Posts: 1,619
Threads: 147
Joined: May 2006
Nice! I've been so swamped I have been unable to work out any solution except for a small python script because I wanted to see the answer:
import re;
i=0;
while(1):
i+=2;
if not i % 10:
next;
if not re.search("[0-4]",str(i**4)):
print i, i**4;
break;
|