nah « WordPress.com Tag Feed

Kenali dan Kunjungi Obyek Wisata di Pandeglang

bloggerinterest — Sat, 14 Nov 2009 02:41:54 +0000

Sedang jadi Blogger Interest emang mengasikkan, tambah asik lagi kalau saat ini sedang sibuk mensupport ; Kerja Keras Adalah Energi Kita,Kenali dan Kunjungi Obyek Wisata di Pandeglang,Kenali dan Kunjungi Obyek Wisata di Pandeglang,Kenali dan Kunjungi Obyek Wisata di Pandeglang,Kenali dan Kunjungi Obyek Wisata di Pandeglang,Kenali dan Kunjungi Obyek Wisata di Pandeglang.
Ada ada ah Ah air Akhirnya akhirnya aktifitas Aku Akulah alat aman ambil angka antara apa apalagi Artikel asam atau awalnya Awalnya bagaimana bagi bagiku bahkan Bahkan bahwa baik ban bangunan banyak barang baterai batu baut bayangkan beberapa bekas beli berangkat berapa Berarti berat berbeda berjalan berpendapat berpikir berserakan berubah berusaha beserakan besi biasanya biaya bisa Bu Bukankah bulan Bumi butuh cari Coba curiga dalam dan Dan dapat dari dengan Desember di diantaranya dibawah dibelinya diberikan diberikannya dibongkar didalam diinternet dijalan dikiranya diluar disaat disebabkan disekitar diselenggarkan ditawar ditempat ditengah Ditulis Dua duri enam es fotokopi gelar Gengsi gengsi gumamku guna Guru hal hampir hanya hanyalah harga harganya hari harus hasil hasilnya haus hausnya he hina hingga hitam ibadah Ibu Ide iklan ikut IM3 informasi ini iseng itu iyalah jadi Jadi Jadilah jalan Jalan jawab jerih juga kaget kaki kalau kantong kapan karena Karena kata kawat ke keberuntungan kecelakaan kecil kedalam Kegiatan kelancaran Keluarga keluarga Kemudian kemudian kenangan kenapa kendaraan kepada keperluan kesibukan ketahuan keteledoran khusus kilo kiloan kira Kita kliping kolong Kompas Kompetisi kulakukan kumasukkan lah lain Lain lainnya lakukan layak lebih lima lomba lumayan lupa lupakan Maaf maaf macam Malam malu Malu manfaat mantap Manusia masak masih mau Medoho melaksanakan melakukan melakukannya meletus melihat memang membahayakan memberikan memberikannya membongkar membuat membungkuk membutuhkan mempunyai menawar mencarinya mencoba mencolok mendapat mendapatkan mendekat mendekati mendengarkan Menemukan mengambil mengemasi mengembalikan mengganggu menghadapinya mengingat mengumpulkan mengumpulkannya mengurangi menjawab menjualnya menolong menunggu menyimpan Menyingkirkan menyingkirkan merasa mereka merupakan minggunya minta misalnya mobil motifasi motor MuDA mudah mudahan mulai Mungkin mungkin mur Nah naik Namun namun nasehat nasibku ngumpulin nyawa Oh Oleh oleh Orang orang pahala pakai paku paling payahku pecah pelebaran pemasangan pemilik Pemulung penasaran pengangkutan pengecatan pengepul perbuatan percakapan peristiwa Peristiwa perjalanan perjalanannya perkilo perkilonya Permisi Perosok petugas pinggir plastik Plewan pokoknya pompa pulang puluh rasa rata ratus ribu Rp rugi rumah rumahku Sahabat saja saku sampah sampai Santi satu Saya Sebagian sebelumnya Sebenarnya sebenarnya sebesar sebuah secara sedang segala segera sehari sehingga seingat sejenak sekali sekarang sekolah Sekolah sekolahku selain selalu selama Semarang sembuh Semua semua sendirinya sengaja seorang Seorang sepanjang sepeda seperti sepuluh seragam Seringkali serta sesama sesuatu setahun setara setengah setiap sewaktu Sewaktu Si sibuk sih sisa situ spanduk Sudah sudah Sulistiyowati tahu tahun tambahan tanah tanya tanyaku Tapi target tas tawar Teman teman tempat tentu Tentunya terdapat terdiam terhadapku Terjatuh terjatuh terkadang terkumpul termasuk ternyata tersebar tertancap tetangga Tetanggaku tiba tidak tidur tiga tinggal Toh total tua tugas turun tutup uang Untuk untuk Untung urat wah Walau Website www.susahnyaedittaghtmlalawebmaster.blogspot.com www.akulahsahabatbumi.blogspot.com www.mudaers.com Ya ya yaitu yang Yang 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 23 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 75 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 7768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 2007 3000 27000

Kerja Keras Adalah Energi Kita

bloggerinterest — Sat, 14 Nov 2009 02:32:57 +0000

Seelenfänger (V)

skriptum — Mon, 26 Oct 2009 12:00:36 +0000

Ach Seelenfänger, was tust du nur?
Tauchst meine Gedanken in blaue Blumen
und färbst meine Seele laut.

Nah und fern zugleich schmecke ich den Wind!
Fühle, wie er sich sittsam in meinen Augen niederlässt
um sich zugleich zu instrumentieren.

Hörst du das Rot, so klammheimlich?
Sieh, wie es sich weise mit dem Gelb verstrickt
und in schwarzer Unschuld von dannen schreitet.

Wir sollten uns berühren, mit Seelenfäden!
Sie sich verschlingen lassen in ihrer Ungeduld
und ihnen zeigen, dass es anders geht.

Seelenfänger, hörst Du es denn nicht?
Öffne Deine Arme, nimm die Klänge der Nacht
und tanze sie ins Grün, damit sie leben.

Problem 31 - Coins

sumidiot — Tue, 20 Oct 2009 03:01:04 +0000

In problem 31, we are given a list of coin values and a target value, and asked how many ways we can make that target value with coins of the given denominations. I think this is probably a pretty standard programming exercise, and I’m embarassed it took me as long as it did. Anyway, here’s what I came up with, nothing too special:

import sys, time

def worker(target, coins):
    """ assume coins are sorted decreasing """
    if target == 0:
        return 1 # there's only one way to not have money?

    # get rid of coins bigger than the target
    while(len(coins) and target < coins[0]):
        coins = coins[1:]

    # some easy cases
    if not len(coins): # all coins are > target
        return 0
    if len(coins) == 1:
        return (target % coins[0] == 0) and 1 or 0

    # there are >=2 types of coins, all <= target
    return sum([worker(target - v, coins[1:])
                for v in xrange(0, target + 1, coins[0])])

def solve(target, coins):
    t = time.time()
    coins.sort()
    coins.reverse()
    ret = worker(target, coins)
    t = time.time() - t
    return [ret, t]

if __name__ == "__main__":
    print solve(int(sys.argv[1]), map(int,sys.argv[2:]))

I also had fun trying this another way. Instead of breaking down the target amount and doing things recursively, I thought about having a huge amount of coins around. The example problem had a target of 200, with denominations 1, 2, 5, 10, 20, 50, 100, 200. Suppose I had 200/1=200 of the 1s, 200/2 = 100 of the 2s, 200/5 of the 5s, etc. I could then take any number of my 1s, along with any number of my 2s, etc, see how much I have, and see if the total is 200. So I’ve got a product space of possible combinations, and I want to pick out all those that have a value of 200. This presents a fun application of the map/reduce functions, which I can’t seem to get enough of.

Suppose I start with the list of coins available, [1,2,5,10,20,50,100,200], and my target value in mind. I want to replace each item in this list, say the value , with the list of values I could make using only $c$-valued coins (and only enough to get to 200). I’m going to do something a little silly, and store each value as it’s own list (this’ll make the next step easier, I think). So I first do

target = 200
coins = [1, 2, 5, 10, 20, 50, 100, 200]
var = map(lambda c: [ [v] for v in xrange(0,target+1,c)], coins)

Next, I want to make the cartesian product of all of these list of lists. I’ll build up my product, working my way through var. That is, I think about as , where each here is a list. This is the sort of thing reduce is for. Keeping in mind that all of my lists are lists of one-element lists (so “+” below is really “concatenate”), I can build the cartesian product via

var = reduce(lambda l,r: [ x+y for x in l for y in r ], var)

Now each element of var is a list with as many values as there are coins, e.g. [7, 6, 0, 40] corresponds to 7 in 1s, 6 in 2s, 0 in 5s, 40 in 10s (of course, in the running example, I need more values in the list, but I hope the point comes across). To find the combined value of the coins in each list, I just need the sum, so I do

var = map(sum, var)

Finally, I only want the values where the sum is my target value, so I filter. We’re asked for the number of ways to make the target value, so I take the length:

var = filter(lambda t: t == target, var)
print len(var)

This is not particularly efficient. In the stated problem, the product of my lists has 1.28 billion elements (significantly larger than the actual answer, which is less than 100,000). That’s rather a lot of space to be taking up. I tried running this code on a target value of 100, and gave up on it after about 20 minutes. For contrast, the first solution above, the recursive solution, takes about half a second on my computer. Still, I like this exercise of getting the cartesian product of a bunch of lists.

Monday's Blah (BRB)

NahM. — Mon, 19 Oct 2009 19:18:10 +0000

Hey everybody! Sorry about the lack of posts.. been busy and a big storm showed up here on saturday!

19:45

nene — Sun, 11 Oct 2009 03:40:33 +0000

“siento” las plantas crecer en la terraza, mientras escucho música china a un volumen mu

Mısır Piramitleri

topraksiz — Thu, 08 Oct 2009 18:59:32 +0000

Keşke şaka olsa.. Kötü bir eşek şakası bile olsa razıyım.. Bu kadar da dibe vurulmaz ki.. Ben bu ins

Whats Going on!

NahM. — Wed, 07 Oct 2009 21:42:24 +0000

Hey ppl, this is Nah! (no! realy?) lot of things happens so fast.. i know i know.. Kanye West! but g

Baum

Windowsbunny — Tue, 06 Oct 2009 06:41:42 +0000

Problem 30 - Sums of Digit Powers

sumidiot — Fri, 02 Oct 2009 02:58:54 +0000

In problem 30 we are asked to find all the numbers that are the sum of the -th powers of their digits. Well, the sum of those numbers. And . You can just about do it in one line…

Most of this problem is fairly easy: Given a number, compute the sum of the powers of its digits, and see if you get the number you started with. If you wanna find all such numbers, just loop for a while. The trouble is how long? If you’ve got some range in mind you want to look at, you can do the following:

def solve(pow, low, hi):
    """ Find the sum of all numbers that are the sum of their pow-th powers of digits in the interval [low,hi) """
    return sum(filter(lambda n:n==sum(map(lambda d:int(d)**pow,str(n))) xrange(low, hi)))

Ok, sure, it’s a longish line, but it’s all there. Python’s syntactic sugar for handling all the little looping is awesome.

The remaining trick is to figure out those bounds. The problem mentions not using 1 as one of the numbers, so we could start at 2. However, that means we could start at 2**pow, because that’ll be the lowest digit sum we’d make. On the high end, the biggest digit-power-sum will come from digits being 9. So I figure that to find an upper bound, you can think about digit sums for longer and longer strings of 9. If you think about 9s in a row, that’ll give a digit sum of (if is the specified power). Find the largest where this value is still a -digit number (or bigger), and you’ve got yourself an upper bound to loop to. Here’s some code to find the bound:

def bound(pow):
    """ Find bound on nums that can be sum of their pow-th powers of digits """
    if pow <= 1:
        return 9
    nine = 9**pow
    coeff = 1
    while(len(str(coeff*nine)) >= coeff):
        coeff += 1
    return (coeff - 1)*nine

Now, these bounds I’ve set up are pretty poor. In the case where the power is 4, as in the example problem, we’d loop from 2**4=16 to 32805, when the actual numbers that have the property we’re hoping for are between 1000 and 10000. We’ve overshot by something like 3 times too many values.

So, we could try to be more clever and come up with a smaller set to search in. But we could still use that one-line solver to pick from that set the appropriate values. And that makes me happy.

Problem 29 - One Liner

sumidiot — Wed, 30 Sep 2009 01:18:30 +0000

In problem 29 we are supposed to determine how many unique values there are for the value where , (all integers). I know there is clever counting one can do to determine this, but I thought for comparison I’d also just do up the brute force approach. Find all the values, get rid of duplicates, and count how many are left.

Python has a built-in type “set”, which will eliminate duplicates as I go. And you can ask for the length of the set. And list comprehension is awesome. So you can solve this in one line:

def solve(lowa, hia, lowb, hib):
    return len(set([a**b for a in xrange(lowa, hia+1) for b in xrange(lowb, hib+1)]))

Having wrapped up solution code into a function, I also like to use command line arguments to pass in… well, arguments. In this case, I’d like to allow two different options for command line options. The stated problem has the bounds 2 to 100 for both and . So if I take one number in, I want to assume it is the upper bound, and that 2 is the lower bound, for both variables. Alternatively, I could take all four bounds in as arguments. I know the following is crude, and doesn’t fail gracefully, but it’s got a lot of nice things going for it:

if __name__ == "__main__":
    if len(sys.argv) == 2:
        print solve(*(2*[2,int(sys.argv[1])]))
    else:
        print solve(*map(int, sys.argv[1:]))

I love this. Using * to change a list into the form my solve function expects, with 4 arguments… that’s awesome. Using [1:] to grab the arguments besides the called command… nice. 2*list to do a list concatenated with itself… delightful. And, of course, map. Who could forget map?

Anyway, I may return to the more mathematical, counting based solution for this. At the same time, though, this code is quick (to write and run (for the stated bounds)) and easily checked. Maybe I’ll leave the more mathematical solution to the other authors on this blog. I seem to recall that there were some…

Cesitli kiyafetler

Shera Shilova — Tue, 29 Sep 2009 10:24:33 +0000

Bayanlara cesitli kiyafetler. Bunlar sadece bir kismi, mutlaka ugramanizi tavsiye ederim.

Kiyafetler (outfits), 1 L$, Nah!

Sac (hair), 0 L$, Roge, grup hediyesi, cesitli renkler (group gift, many colors, scripted)

Problem 27 - Prime Generating Quadratics

sumidiot — Mon, 28 Sep 2009 01:48:31 +0000

With my freshly minted PrimeOracle, I’m ready to attack problem 27, in which we are asked to find polynomials that generate long strings of consecutive primes starting when you plug in 0.

Basically my idea for this problem was to iterate through all of the choices for the coefficients and in , and then for each polynomial plug in values until we hit a composite. Record maximums as appropriate.

There are a few easy optimizations to make. First, itself has to be prime, because we test . Next we test , so we need , that is . Those are the only optimizations I made, though I’m sure there are some more advanced ones out there.

So here’s my code:

# assume PrimeOracle.py is in the same directory as this file
from PrimeOracle import PrimeOracle

def solve(n):
    # seed our oracle to 2*n+2, since n=1 gives 1+a+b
    isprime = PrimeOracle(2*n + 2)
    maxp, maxn = 41, 40
    for b in [m for m in xrange(2, n) if isprime[m] ]:
        for a in xrange(1-b, n):
            idx = 1 # we already tesetd 0 with isprime[b]
            while(isprime[idx*idx+idx*a+b]):
                idx += 1
            if(idx - 1 > maxn):
                maxp, maxn = a*b, idx - 1
    return maxp

I had my oracle also tell me what the biggest number it was asked about was, and the answer was up around 12000. So we could change the starting seed in the oracle to be big enough to never have to extend, and probably save some time.

Prime Oracle

sumidiot — Mon, 28 Sep 2009 00:52:37 +0000

I sat down to work on problem 27, and realized it might be handy to have a class that’ll tell me when numbers are prime. Something that’ll keep some memory of tested values, and sieve when it needs to know about bigger numbers than have been tested. Here’s what I threw together:

""" A class that'll help you decide if things are prime

Make an object of this class, I'll call it oracle, and
oracle[n] is True or False depending on if n is prime or not

Assumes negatives are not prime
"""

class PrimeOracle:
    # Takes a starting maximum to sieve to
    def __init__(self, startmax = 2000):
        # self.isprime[n] = True if 2n+1 is prime
        self.isprime = [True for n in xrange(0,startmax//2 + 1)]
        self.isprime[0] = False
        # run the sieve
        for val in xrange(3, startmax + 1, 2):
            if(self.isprime[val // 2]):
                for mul in xrange(val*val, startmax + 1, 2*val):
                    self.isprime[mul // 2] = False

    # is n prime?
    def __getitem__(self, n):
        if n == 2:
            return True
        if n<2 or n % 2 == 0:
            return False
        nidx = n//2
        while(nidx >= len(self.isprime)):
            # extend the sieve
            # currently, we know the primality of 2*(len-1)+1=2*len-1
            # we can extend that to its square, 4*len^2 - 4len + 1
            # which would be index (4*len^2-4*len+1) // 2 = 4len(len-1)+1 // 2
            # so the new length should be one more than this
            # we already have length len, so we need to add (prev line) - len
            curlen = len(self.isprime)
            newlen = (4*curlen*(curlen-1) + 1) // 2
            self.isprime += [True for n in xrange(0, newlen+1-curlen)]
            maxknown = 2*curlen - 1
            maxtoknow = 2*newlen - 1
            for idx in xrange(1, curlen):
                if(self.isprime[idx]):
                    val = 2*idx + 1
                    # start crossing off multiples larger than maxknown
                    sfact = 1 + (maxknown // val)
                    if(sfact % 2 == 0):
                        sfact += 1 # otherwise our indexing is wrong
                    startfact = max(val, sfact)
                    for mul in xrange(startfact * val, maxtoknow + 1, 2*val):
                        self.isprime[mul // 2] = False
        return self.isprime[nidx]

    # the largest index we know about
    def __len__(self):
        return 2*len(self.isprime) - 1

    def __str__(self):
        return "\n".join(["%d: %s" % (2*n+1, self.isprime[n])
                          for n in xrange(0,len(self.isprime))])

It’s basically just the usual sieve, and then it extends itself whenever necessary.

There’s room for improvement. We could remember the highest value we’ve sieved to, and then take our array and just tease out the primes from it. Then we only store the list of primes. When we go to extend, we make a new array to hold whatever the new values would be, sieve appropriately, and then just append to our list of primes. Here’s my modified __getitem__ method:

def __getitem__(self, n):
    if n == 2:
        return True
    if n<2 or n % 2 == 0:
        return False
    while(n > self.sievedto):
        # we have sieved to sst = self.sievedto (assume odd)
        # we could sieve to sst^2 (will also be odd)
        # that'd be [sst+1 .. sst^2]
        # but really we might as well do [sst+2 .. sst^2]
        # sst=2s+1 => [2s+3 .. 4s^2+4s+1] //2 is [s+1 .. 2s(s+1)]
        # which has length 2s^2+s
        # and index i corresponds to 2(s+1+i)+1
        # so that an odd integer m is at m//2 - (s+1)
        s = self.sievedto // 2
        newsst = self.sievedto * self.sievedto
        isprime = [True for idx in xrange(0,2*s*s+s+1)]
        for pr in self.primes.keys():
            sfact = 1 + (self.sievedto // pr)
            if(sfact % 2 == 0):
                sfact += 1
            for mul in xrange(max(pr, sfact) * pr, newsst + 1, 2 * pr):
                isprime[mul//2 - (s+1)] = False
        for m in [2*(s+1+i)+1 for i in xrange(0,2*s*s+s+1) if isprime[i]]:
            self.primes[m] = True
        self.sievedto = newsst
    return self.primes.has_key(n)

My first guess was that the second version would be better, because it would be storing less. However, it seems to be a little slower. I guess filtering out the primes at each extension has its cost. Or my implementation could use some tweaking? I wish I knew a good way to compare the size requirements of these scripts, in addition to how long they take to run.

By the way, I would be a little shocked to find that I’m not off by one somewhere (many somewheres) in the above code. Little tests seem to work out. But use are your own peril.

Problem 28 - Spirals

sumidiot — Sun, 27 Sep 2009 19:18:42 +0000

In problem 28 we construct a spiral starting with 1 in the center, and build our way outwards clockwise. Then we are supposed to find the sum of the diagonals.

I figured there should be some formula for the sum of the diagonal entries, depending on the size of the grid you want to stop at. If we only compute diagonal elements when we have a square, the square will always have sides of odd length. So we should take a number , and compute the appropriate sum of the elements in the by grid. I’ll write down the sum for the case in the following manner:

  1^2 + (1^2+2) + (1^2+2*2) + (1^2+3*2)
+ 3^2 + (3^2+4) + (3^2+2*4) + (3^2+3*4)
+ 5^2 + (5^2+6) + (5^2+2*6) + (5^2+3*6)
+ 7^2

That seems to make a pattern pretty clear. We’re adding up odd squares (the rising diagonal coming off of the central 1) and then some offsets of those squares. The formula for the sum of squares is

and we can write the sum of odd squares as

Pushing some symbols around, I got down to the following formula for the sum of the diagonal elements in a by square:

Of course, those thirds make me nervous. The answer is clearly supposed to be an integer. But it’s ok. Gathering up those thirds we get

for the sum. Now, if is divisible by 3, we’ll get an integer in that first term. And if is not divisible by 3, it is either 1 or 2 mod 3, and so . So that third isn’t going to cause problems.

It’s pretty easy to code up the solution when you’ve got such a formula. By using an if-switch, I took into account the mod-3 thing above so that I was never working with floats. Here’s my code:

def solve(n):
    ret = 10*n*n + 1
    factor = 8*n*n+13
    if(n % 3 == 0):
        ret += 2*(n//3)*factor
    else:
        ret += 2*n*(factor//3)
    return ret

Then I decided it would be a fun little exercise to actually cook up some code to generate the spiral. I knew this would have slower run-time, and require more resources, but I wanted to write down the code anyway.

The way I thought about it, I’ll keep track of my current coordinates , starting at . To go to the next place, I look to my right, and see if that spot is filled. If it is, I just continue on in whichever direction I was going; otherwise, I move to that spot to my right. This means I need to keep track of which direction I’m facing. There are only 4 directions, and they cycle, so I’ll think of direction as an integer mod 4. If I make 0,1,2,3 correspond to Up, Right, Down, Left, then (my direction)+1 is the direction to my right. After writing down a little grid of input directions, and what they meant about where to move next, I came up some little formulas, and was good to go.

I had originally thought I’d wrap up an implementation into a class, so that you could index via , but then I realized that wouldn’t quite work, and you’d need to just index as . At that point, I figured I might as well just use a dictionary type. And away I went:

def dx(dir):
    """ interpret dir as 0,1,2,3 = N,E,S,W, return what dx is in that dir """
    return (dir%2) * (2-dir)

def dy(dir):
    """ dir as 0,1,2,3 = N,E,S,W return what dy is in that dir """
    return ((dir+1)%2) * (1-dir)

def solve(n):
    """ Build up a 2n+1 by 2n+1 grid, then sum diagonals """
    grid = {}
    posx, posy = 0, 0
    idx = 1
    dir = 0 # dir in 0,1,2,3 = N,E,S,W
    while(posx <= n):
        grid[(posx,posy)] = idx
        idx += 1
        # find the position to my right
        rsposx,rsposy = posx + dx((dir+1)%4), posy + dy((dir+1)%4)
        if(grid.has_key((rsposx,rsposy))):
            # keep going the direction I was going
            posx,posy = posx + dx(dir), posy + dy(dir)
        else:
            # turn right
            posx,posy = rsposx, rsposy
            dir = (dir+1)%4
    # sum the diagonals
    ret = sum([grid[(x,x)] for x in xrange(-n,n+1)])
    ret += sum([grid[(x,-x)] for x in xrange(-n,n+1)])
    ret -= 1 # the origin got counted twice
    return ret

Only two days behind my artificial deadline too.

Joban' ny rap gasy Clip Vidéo

hhdago — Fri, 25 Sep 2009 13:57:27 +0000

Problem 25 - More Fibonacci Numbers

sumidiot — Sat, 19 Sep 2009 13:15:53 +0000

In problem 25 we are asked to find the first Fibonacci number with a given number of digits. Actually, we are only asked to find which Fibonacci number it is (its index in the sequence of Fibonacci numbers), instead of the number itself. We’ve already visited Fibonacci numbers in problem 2, and had useful work there. For this problem, it will be helpful to note that the number of digits of can be found as , where denotes the “floor” function, equal to the greatest integer less than (on a not-particularly-related note, this isn’t the formula I would have used a week ago, glad I got that cleared up).

I decided to do some performance comparisons with this problem, with a few different versions of code. [GRR. WordPress seems to not be doing the sourcecode "tag" like I thought it should below. Or I'm messing something up...]

In version 1, I’ll compute each Fibonacci number successively, using the recurrence relation, then calculate the number of digits until I’ve found a big enough number. My function looks like:

def solve(n):
    ret = 1
    prev, cur = 0, 1
    while(cur &lt; 10**(n-1)):
        prev, cur = cur, prev + cur
        ret += 1
    return ret

In the second version, I’ll compute the -th Fibonacci number using the formula

that was used last time we saw Fibonacci numbers. In fact, I’ll wrap that up into its own function, “digs_fib” that takes in an “n” and gives the number of digits of the appropriate Fibonacci number. Now my code is basically:

def fib(n):
    """ The nth fibonacci number """
    alpha = (1 + 5**.5) / 2
    beta = (1 - 5**.5) / 2
    return (1/5**.5) * (alpha**n - beta**n)
def digs_fib(n):
    """ Number of digits of fib(n) """
    return int(math.log(fib(n), 10)) + 1
def solve(n):
    ret = 1
    while(digs_fib(ret) &lt; n):
        ret += 1
    return ret

Of course, that second term in the formula is smallish, and raised to big powers it gets smaller, so we could maybe ignore it. And then we’re taking the log of products and powers, so we can use log rules to re-write the expression for the number of digits, without directly calculating the Fibonacci number. My “digs_fib” function changes to the following:

def digs_fib(n):
    """ The approximate number of digits of the nth fibonum """
    alpha = (1+5**.5)/2
    return int(n*math.log(alpha,10)-.5*math.log(5, 10))+1

Finally, given a target number of digits , we could just about solve for in

and obtain

This means that given we can get an approximation to which Fibonacci number gives us that many digits in basically constant time. Now my “solve” is just

def solve(n):
    num = n - 1 + .5*math.log(5, 10)
    num /= math.log((1+5**.5)/2, 10)
    return int(num+1) # round up, approximately

I ran each of these through my little performance evaluation script and came up with a reasonable graph. I could only go up to asking for 300 digits, because after that the second version fails, since the numbers are too big. The other versions have no problem at all going up to 1000 digits (and well beyond). But anyway, here’s my graph:

You can barely see that constant time algorithm there along the -axis. The scale of the -axis is some linear change in actual time taken, but the shape of these graphs is what’s important anyway. Right?

Problem 26 - Recurring Decimals

sumidiot — Sat, 19 Sep 2009 13:06:56 +0000

In problem 26, we are asked to find the integer less than a given bound whose reciprocal has the longest decimal period among such integers. That is, among all fractions where is less than some bound, which fraction has the longest periodic part in its decimal representation?

According to Hardy and Wright, An Introduction to the Theory of Numbers (6th ed, Theorem 135),

The decimal for a rational number between 0 and 1 is terminating or recurring, and any terminating or recurring decimal is equal to a rational number. If , , and , then the decimal terminates after digits. If , , where 1' title='Q>1' class='latex' />, , and is the order of 10 (mod ), then the decimal contains non-recurring and recurring digits.

Let’s trim that down a bit. Our , so automatically (this notation is for the greatest common divisor, as usual). If the denominator, , is divisible by either 2 or 5, then either (1) the decimal terminates, or (2) the decimal repeats, with the same length recurring part as the reciprocal of a smaller integer. In either case, we will suppose that this is not the answer. Either (1) its recurring part has length 1, being a string of 0s, or (2) we could return the smaller integer with the same length recurring part.

So, suppose our is not divisible by 2 or 5. Then according to the theorem above, we may compute the length of the repeating part of the decimal for by finding the “order of 10 (mod )”. That is, find the smallest so that . This is easy enough to find, just keep raising 10 to bigger powers, and taking the answer mod , until you find 1.

Here’s my code

def solve(n):
    """ Return the d with largest decimal period, d&lt;n """
    max_d, max_p = 2, 1
    for d in xrange(3,n,2):
        if(d % 5 != 0):
            # now find the order of 10 mod Q
            # that is, the smallest p for which 10^p == 1 mod Q
            p = 1
            tmd = 10 % d # tmd = 10 mod d
            while(tmd != 1):
                p += 1
                tmd = (tmd * 10) % d
            if (p &gt; max_p):
                max_d, max_p = d, p
    return max_d

Pretty straightforward, I think. [Except, as of right now, WordPress doesn't seem to be handling < and > correctly in that code. Grr]

There’s almost certainly more math we could do here. But this seems to work well enough for now.

Nah! yukata 001

Onsen Akina — Fri, 11 Sep 2009 15:35:07 +0000

Nah! BUSH CLOVER

Onsen Akina — Fri, 11 Sep 2009 15:33:37 +0000

Nah! yukata sunflower

Onsen Akina — Fri, 11 Sep 2009 14:07:11 +0000

Nah! 003

Onsen Akina — Fri, 11 Sep 2009 14:05:55 +0000

Nah! 09_yukata_01

Onsen Akina — Fri, 11 Sep 2009 14:04:14 +0000

Nah! soar

Onsen Akina — Fri, 11 Sep 2009 13:17:00 +0000

Nah! burning

Onsen Akina — Fri, 11 Sep 2009 13:14:33 +0000