• <ins id="pjuwb"></ins>
    <blockquote id="pjuwb"><pre id="pjuwb"></pre></blockquote>
    <noscript id="pjuwb"></noscript>
          <sup id="pjuwb"><pre id="pjuwb"></pre></sup>
            <dd id="pjuwb"></dd>
            <abbr id="pjuwb"></abbr>

            The Fourth Dimension Space

            枯葉北風(fēng)寒,忽然年以殘,念往昔,語默心酸。二十光陰無一物,韶光賤,寐難安; 不畏形影單,道途阻且慢,哪曲折,如渡飛湍。斬浪劈波酬壯志,同把酒,共言歡! -如夢令

            POJ 1150 ——The Last Non-zero Digit(數(shù)論)

            題意很簡單,要求你求出一個排列數(shù)P(n,m)中最后一個非0的數(shù)字.
            由于n的數(shù)值巨大,想直接求出來恐怕是不可行的。
            在網(wǎng)上有這樣一個英文的解題報告,由于缺少中文的資料,硬著頭皮把它看完了:-)

            The first program

            Consider the number N! factored into product of powers of prime numbers. It means N!=2i * 3j * 5k * 7l * ... Note, that for each N>1 the power i is greater than k. It means, that the last non-zero digit of N! is the same as the last digit of N! / (2k * 5k). Therefore we can compute the result using the equation:

            (N! / (2k * 5k)) mod 10 = ((N! / (2i * 5k)) mod 10 * 2i-k mod 10) mod 10

            Number i can be obtained easily - we will divide each a=1,2,...,N by 2 until the resulting number is not divisible by 2 and after each division we will add one to the variable i. Number k can be obtained in the same manner. Let f(i) denotes the number which we obtain by dividing i by the 2a * 5b where a and b are the highest numbers such that the i is divisible by this product. Number (N! / (2i * 5k)) mod 10 is equal to f(N!) mod 10 and can be computed as f(1) * f(2) * ... * f(N) mod 10. We will perform operation mod 10 after each multiplication in order to keep the resulting number as small as possible.

            The advantege of this approach is that we do not need to implement arithmetics of large numbers. Some ideas used here are used in the second, more efficient program, as well.

            The second program

            The second program also computes the result as (2i-k mod 10 * f(N!) ) mod 10. Numbers i and k are computed much more efficiently. More precisely

            i=N div 2 + (N div 2) div 2 + ((N div 2) div 2) div 2 + ...

            (We get zero after finite number of divisions.) Number k can be computed in the same way. After that we can compute i-k and we need to find 2i-k mod 10. Observe, that

            21 mod 10 = 2, 22 mod 10 = 4, 23 mod 10 = 8, 24 mod 10 = 6, 25 mod 10 = 2, 26 mod 10 = 4, ...

            i.e. the period is 4 and we need only to compute (i-k) mod 4 and then to find corresponding last digit. This observation can help us to simplify computation of i and k - we do not need their exact values (that can be long) but we need only (i-k) mod 4.

            We have shown how to compute 2i-k mod 10. Now let us consider f(N!) mod 10 = ((f(1) mod 10) * (f(2) mod 10) * ... * (f(N) mod 10)) mod 10. Note, that f(i) mod 10 is always 1,3,7 or 9. If we knew, how many 3,7,9 are among (f(1) mod 10), (f(2) mod 10), ..., (f(N) mod 10), we could compute 3a mod 10, 7b mod 10, 9c mod 10 in the similar way as we did for 2i-k (last digits of powers of 3,7,9 are also periodical).

            To compute the number of 3,7,9 among (f(1) mod 10), (f(2) mod 10), ..., (f(N) mod 10) is not so easy. We will divide numbers 1,2,...,N into groups so, that in each group are numbers with same quotient i/f(i) and we will compute number of 3,7,9 among (f(i) mod 10) for each such group separatelly (there are O(N2) such groups). First, let us consider a group in which i/f(i)=1. This is the group of all numbers not divisible by 2 and 5. The number of 3,7,9 in this group is the same as number of 3,7,9 among 1 mod 10, 2 mod 10, ..., N mod 10. This number can be counted easily - it is N div 10 + a where a is 1 if the last digit of N is at least 3 (resp. at least 7 or at least 9). Now let us consider a group in which i/f(i)=L (where L=2a * 5b). We obtain this group by taking each L-th number from the sequence 1,2,3,... and dividing it by L. It means that number of 3,7,9 for this group will be the same as the number of 3,7,9 among 1 mod 10, 2 mod 10, ..., (N div L) mod 10.

            Now we know everything we needed for construction of a program. Since numbers in the input file are long, we need to implement arithmetics for long numbers. However, by careful implementation we can achieve that only division of a long number by small integer is necessary.



            這個題怎么來做呢?先別急,我們先來討論一下下面幾個子問題:
            1.如何求出n階乘中質(zhì)因數(shù)x(比如說5)出現(xiàn)的次數(shù)?
            比如說15的階乘 :1*2*3*4*5*6*7*8*9*10*11*12*13*14*15
            由于5這個質(zhì)因數(shù)只在5的倍數(shù)里面才出現(xiàn),所以我從5,10,15中各抽出一個5,這相當(dāng)于有15/5個質(zhì)因數(shù)5,之后5,10,15就變成了1,2,3;
            由于非5的倍數(shù)顯然不在考慮范圍之內(nèi),所以我們只需繼續(xù)討論它的子問題3!即可。
            這樣,我們可以用遞歸來解決這個問題。有了這個方法,我們是不是能夠輕易地解決n!末尾有多少個0呢?想想看...n!后0的個數(shù)是不是就和某個質(zhì)因數(shù)的個數(shù)有關(guān)呢?^_^
            比如說,我們要求5出現(xiàn)的次數(shù),可以這樣寫:

            int get5(int n)//計算n!中質(zhì)因子5的出現(xiàn)次數(shù)
            {
                
            if(n==0)
                    
            return 0;
                
            return n/5+get5(n/5);
            }


             2.如何求出n!階乘最后非0位?
            比如說我們要找10!最后非0位,由于質(zhì)因數(shù)2和5組合之后會使得末尾產(chǎn)生0.那么我們不妨把10!中2,5質(zhì)因數(shù)全部去掉,(但是請注意2的數(shù)目其實比5的要多,所以我們要在最后考慮多余的2對末位的影響)
            如 1*2*3*4*5*6*7*8*9*10 去掉2 ,5 因子后 就是1*1*3*1*1*3*7*1*9*1,由于2,5因子已經(jīng)被去除,那么剩下的數(shù)字末尾一定是3,7,9,1中四者之一。然后我們再求出這么一串?dāng)?shù)相乘以后末尾的數(shù)是幾.最后再補上2對末位的影響即可!

            總結(jié)一下,求10!最后一個非0位的步驟如下:
            step1:首先將10!中所有2,5因子去掉;
            step2:然后求出剩下的一串?dāng)?shù)字相乘后末尾的那個數(shù)。
            step3:由于去掉的2比5多,最后還要考慮多余的那部分2對結(jié)果的影響。
            step4:output your answer!

            正如上面文章里所說的“To compute the number of 3,7,9 among (f(1) mod 10), (f(2) mod 10), ..., (f(N) mod 10) is not so easy”,這里面步驟2是個難點。如何求出剩下的這串?dāng)?shù)字相乘后最后一位是幾呢?這可以轉(zhuǎn)化成求出這些數(shù)里面末尾是3,7,9的數(shù)字出現(xiàn)的次數(shù)(為啥?因為這些數(shù)的n次方是有規(guī)律的,周期為4,不信你可以推一下)
            好,現(xiàn)在問題就是如何求出這串?dāng)?shù)字中末尾3,7,9各自出現(xiàn)的次數(shù)了;

            一個數(shù)列實際上可以分成偶數(shù)列和奇數(shù)列,以1*2*3*4*5*6*7*8*9*10為例

            分成1 3 5 7 9,   2 4 6 8 10

            這樣我們嘗試分別進(jìn)行統(tǒng)計,可以發(fā)現(xiàn),實際上2,4,6,8,10中的個數(shù)也就是1 2 3 4 5中的個數(shù),也就是說我們又把這個問題劃分成了一個原來問題的子問題。

            f(n) = f(n/2) + g(n),g(n)表示奇數(shù)列中的數(shù)目,所以我們需要解決g(n)

            再次觀察g(n)

            實際上又分成了兩部分1 3 7 9 11 13 17 19 21。。。以及5的奇倍數(shù)5,15,25。。。說明又出現(xiàn)了子問題,如果要統(tǒng)計這個數(shù)列中末尾為x(1,3,7,9)的個數(shù)可以這樣寫:g(n,x) = n/10+(n%10 >= x)+g(n/5,x) 

            這樣利用了兩個遞歸方程,我們就可以在lgn的時間內(nèi)計算出末尾為1,3,7,9的數(shù)的個數(shù)了

            好了,現(xiàn)在我們得到了這串?dāng)?shù)字中末尾是3,7,9的數(shù)字的個數(shù),我們利用循環(huán)節(jié)的性質(zhì)可以快速地算出這串?dāng)?shù)字相乘后mod 10的結(jié)果,在考慮下當(dāng)時多除的2(其實也可以用循環(huán)節(jié)來處理),便可求出答案!




            解決了上面兩個子問題,我想求P(n,m)最后一個非0位就變得十分容易了。
            P(n,m)實際上等于 n! / (n-m)!
            我們可以求出n! 和(n-m)!中質(zhì)因數(shù)2,5,3,7,9分別出現(xiàn)的次數(shù),然后再各自相減。
            然后再用循環(huán)節(jié)處理,即可!
            BTW,這里還要注意一個trick,就是2的出現(xiàn)次數(shù)如果小于5,(這對于排列數(shù)來說是可能的)我們可以直接輸出5,如果2的數(shù)目等于5,那么2的循環(huán)節(jié)不需要考慮。至于3,7,9的循環(huán)節(jié),由于這些數(shù)的4次方末位剛好是1,所以就不需要特殊考慮了。


            附代碼:
            #include<iostream>
            #include
            <cstring>
            #include
            <cmath>
            using namespace std;

            int get2(int n)//計算n!中質(zhì)因子2的出現(xiàn)次數(shù)
            {
                
            if(n==0)
                    
            return 0;
                
            return n/2+get2(n/2);
            }


            int get5(int n)//計算n!中質(zhì)因子5的出現(xiàn)次數(shù)
            {
                
            if(n==0)
                    
            return 0;
                
            return n/5+get5(n/5);
            }


            //////////////////////////////////////////////////////////////////////////
            int g(int n,int x)//計算f(1) to f(n) 中,奇數(shù)數(shù)列中末尾為x的數(shù)出現(xiàn)的次數(shù)
            {
                
            if(n==0)
                    
            return 0;
                
            return n/10+(n%10>=x)+g(n/5,x);
            }


            int getx(int n,int x)//計算f(1) to f(n)中,末尾為x的數(shù)的出現(xiàn)次數(shù)
            {
                
            if(n==0)
                    
            return 0;
                
            return getx(n/2,x)+g(n,x);
            }

            //////////////////////////////////////////////////////////////////////////

            int table[4][4=
            {
                    
            6,2,4,8,//2^n%10的循環(huán)節(jié),注意如果2的個數(shù)為0時候,結(jié)果應(yīng)該是1,要特殊處理。 
                    1,3,9,7,//3
                    1,7,9,3,//7
                    1,9,1,9,//9    
            }
            ;//3,7,9的循環(huán)節(jié)中第一位,剛好是1,故不需要考慮這些數(shù)字出現(xiàn)次數(shù)為0的情況。


            int main()
            {

                
            int n,m;
                
            int num2;
                
            int num3;
                
            int num5;
                
            int num7;
                
            int num9;
                
            while(scanf("%d%d",&n,&m)!=EOF)
                
            {
                    num2
            =get2(n)-get2(n-m);
                    num5
            =get5(n)-get5(n-m);
                    num3
            =getx(n,3)-getx(n-m,3);
                    num7
            =getx(n,7)-getx(n-m,7);
                    num9
            =getx(n,9)-getx(n-m,9);
                    
            int res=1;
                    
            if(num5>num2)
                    
            {
                        printf(
            "5\n");
                        
            continue;
                    }

                    
            else 
                    
            {
                        
            if(num2!=num5)
                        
            {
                            res
            *=table[0][(num2-num5)%4];
                            res
            %=10;
                        }
            //如果num2==num5,那么2^0次方mod 10應(yīng)該為1 ,而不是table中的6,所以要特殊處理。
                        
                        res
            *=table[1][num3%4];
                        res
            %=10;
                        res
            *=table[2][num7%4];
                        res
            %=10;
                        res
            *=table[3][num9%4];
                        res
            %=10;
                    }

                    printf(
            "%d\n",res);
                }

                
            return 0;
            }


                                                                                                                                                                                                                     special thanks to 星星 

            posted on 2009-10-31 19:07 abilitytao 閱讀(3603) 評論(8)  編輯 收藏 引用

            評論

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2009-11-01 13:20 凡客誠品網(wǎng)

            似懂非懂  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論)[未登錄] 2009-11-01 17:14 abilitytao

            @凡客誠品網(wǎng)
            請不要使用帶有廣告信息的署名  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2010-03-04 15:52 foreverlin

            寫的很好很易懂,受教了  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2010-07-22 16:59 ACMer

            贊一個  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2010-07-29 11:20 inowfordream

            寫得很好,受教了!感謝萬分!  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2010-08-11 21:55 slp

            您好,能問你一個問題嗎?這個函數(shù)的子函數(shù)
            int odd_getX(int n,int x){
            if(n == 0) return 0;
            return n/10+(n%10 >= x)+ odd_getX(n/5,x);
            }

            odd_getX(n/5,x)不是又生成了一個1 2 3...的序列,應(yīng)該要在進(jìn)行偶序列和奇序列的劃分,為什么直接判斷他為奇序列處理?如果你知道原理的話麻煩發(fā)郵件給我1032327983@qq.com 再次感謝  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2010-08-11 22:09 slp

            哦 我發(fā)現(xiàn)我錯了,比如奇序列1 3 5 7 9 11 13 15 17 19 ......
            二分為: 1 3 7 9 11 13 17 19 .....
            五倍數(shù): {5 15 25 35 ....}/5 = {1 3 5 7} 還是奇序列 我錯了!
            Orz!  回復(fù)  更多評論   

            # re: POJ 1150 ——The Last Non-zero Digit(數(shù)論) 2012-07-14 09:52 pictureyong

            step2:然后求出剩下的一串?dāng)?shù)字相乘后末尾的那個數(shù)
            對這步的理解不對吧,  回復(fù)  更多評論   


            只有注冊用戶登錄后才能發(fā)表評論。
            網(wǎng)站導(dǎo)航: 博客園   IT新聞   BlogJava   博問   Chat2DB   管理


            狠狠色丁香久久婷婷综合_中| 久久亚洲春色中文字幕久久久| 国产精品99久久久久久宅男| 日本久久中文字幕| 久久久国产乱子伦精品作者| 久久久久亚洲AV综合波多野结衣| 伊人久久大香线蕉综合Av| 国产亚洲精午夜久久久久久| 中文精品久久久久人妻不卡| 国产精品成人无码久久久久久| 亚洲欧美伊人久久综合一区二区| 国产福利电影一区二区三区久久久久成人精品综合 | 久久精品国产亚洲5555| 久久精品中文字幕无码绿巨人| 精品久久久久久无码中文野结衣 | 久久成人精品视频| 久久精品国产久精国产一老狼| 国内精品久久久久久不卡影院 | 亚洲国产精品成人AV无码久久综合影院| 亚洲成色www久久网站夜月| 久久综合视频网站| 久久久久黑人强伦姧人妻| 久久r热这里有精品视频| 久久精品午夜一区二区福利| 人人妻久久人人澡人人爽人人精品| 99久久精品国产综合一区| 欧美激情精品久久久久| 国产精品久久久久久福利69堂| 日日躁夜夜躁狠狠久久AV| 亚洲乱码中文字幕久久孕妇黑人| 午夜视频久久久久一区| 性高朝久久久久久久久久| 一本大道久久香蕉成人网| 香蕉久久夜色精品国产2020| 久久久久无码国产精品不卡| 久久青青草原精品国产不卡| 欧美午夜A∨大片久久| 久久亚洲AV成人无码| 一本色道久久88—综合亚洲精品| 亚洲精品国产美女久久久| 久久国产精品77777|