Description
A supermarket in Tehran is open 24 hours a day every day and needs a number of cashiers to fit its need. The supermarket manager has hired you to help him, solve his problem. The problem is that the supermarket needs different number of cashiers at different times of each day (for example, a few cashiers after midnight, and many in the afternoon) to provide good service to its customers, and he wants to hire the least number of cashiers for this job.
The manager has provided you with the least number of cashiers needed for every one-hour slot of the day. This data is given as R(0), R(1), …, R(23): R(0) represents the least number of cashiers needed from midnight to 1:00 A.M., R(1) shows this number for duration of 1:00 A.M. to 2:00 A.M., and so on. Note that these numbers are the same every day. There are N qualified applicants for this job. Each applicant i works non-stop once each 24 hours in a shift of exactly 8 hours starting from a specified hour, say ti (0 <= ti <= 23), exactly from the start of the hour mentioned. That is, if the ith applicant is hired, he/she will work starting from ti o’clock sharp for 8 hours. Cashiers do not replace one another and work exactly as scheduled, and there are enough cash registers and counters for those who are hired.
You are to write a program to read the R(i) ‘s for i=0..23 and ti ‘s for i=1..N that are all, non-negative integer numbers and compute the least number of cashiers needed to be employed to meet the mentioned constraints. Note that there can be more cashiers than the least number needed for a specific slot.
Input
The first line of input is the number of test cases for this problem (at most 20). Each test case starts with 24 integer numbers representing the R(0), R(1), …, R(23) in one line (R(i) can be at most 1000). Then there is N, number of applicants in another line (0 <= N <= 1000), after which come N lines each containing one ti (0 <= ti <= 23). There are no blank lines between test cases.
Output
For each test case, the output should be written in one line, which is the least number of cashiers needed.
If there is no solution for the test case, you should write No Solution for that case.
Sample Input
1
1 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
5
0
23
22
1
10
Sample Output
1
Source
Tehran 2000
分析：
分析：
我们按刚才所讲到的方法对此题进行处理。这题很容易想到如下的不等式模型：
设num[ i ]为i时刻能够开始工作的人数，x[ i ]为实际雇佣的人数，那么x[ I ]<=num[ I ]。
设r[ i ]为i时刻至少需要工作的人数，于是有如下关系：
x[ I-7 ]+x[ I-6 ]+x[ I-5 ]+x[ I-4 ]+x[ I-3 ]+x[ I-2 ]+x[ I-1 ]+x[ I ]>=r[ I ]
设s[ I ]=x[ 1 ]+x[ 2 ]…+x[ I ]，得到
0<=s[ I ]-s[ I-1 ]<=num[ I ]， 0<=I<=23
s[ I ]-s[ I-8 ]>=r[ I ]， 8<=I<=23
s[ 23 ]+s[ I ]-s[ I+16 ]>=r[ I ]， 0<=I<=7
对于以上的几组不等式，我们采用一种非常笨拙的办法处理这一系列的不等式(其实也是让零乱的式子变得更加整齐，易于处理)。首先我们要明白差分约束系统的应用对象（它通常针对多个二项相减的不等式的）于是我们将上面的所有式子都转化成两项未知项在左边，另外的常数项在右边，且中间用>=连接的式子，即:
s[ I ]-s[ I-1 ]>=0 (0<=I<=23)
s[ I-1 ]-s[ I ]>=-num[ I ] (0<=I<=23)
s[ I ]-s[ I-8 ]>=r[ I ] (8<=I<=23)
s[ I ]-s[ I+16 ]>=r[ I ]-s[ 23 ] (0<=I<= 7)
这里出现了小的困难，我们发现以上式子并不是标准的差分约束系统，因为在最后一个式子中出现了三个未知单位。但是注意到其中跟随 I变化的只有两个，于是s[23]就变得特殊起来，看来是需要我们单独处理，于是我们把 s[ 23 ]当作已知量放在右边。
经过这样的整理，整个图就很容易创建了，将所有形如 A-B>=C 的式子 我们从节点B 引出一条有向边指向 A 边的权值为C （这里注意由于左右确定，式子又是统一的>=的不等式，所以A和B是相对确定的，边是一定是指向A的） ，图就建成了 。
最后枚举所有s[ 23 ]的可能值，对于每一个s[23]，我们都进行一次常规差分约束系统问题的求解，判断这种情况是否可行，如果可行求出需要的最优值，记录到Ans中，最后的Ans的值即为所求。
代码：
Source Code
Problem: 1275
Memory: 892K Time: 63MS
Language: Pascal Result: Accepted
Source Code
const
maxn=1000;
type
node=record
u,v,w:longint;
end;
var
n,m,casen,casei:longint;
e:array[1..maxn] of node;
t,r,d:array[-1..maxn] of longint;
procedure add(x,y,z:longint);
begin
inc(m);
with e[m] do
begin
u:=x; v:=y; w:=z;
end;
end;
procedure init;
var
i,x:longint;
begin
for i:=0 to 23 do read(r[i]);
readln;
readln(n);
fillchar(t,sizeof(t),0);
for i:=1 to n do
begin readln(x); inc(t[x]); end;
end;
procedure main;
var
i,j,min:longint;
check:boolean;
begin
for min:=1 to n do
begin
m:=0;
for i:=-1 to 23 do
d[i]:=100000;
add(23,-1,-min);
for i:=0 to 23 do
begin
add(i,i-1,0); add(i-1,i,t[i]);
if i>7 then add(i,i-8,-r[i]) else add(i,i+16,-r[i]+min);
end;
for j:=0 to 23 do
begin
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then
begin
d[v]:=d[u]+w;
check:=false;
end;
if check then break;
end;
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then begin check:=false; break; end;
if check then begin writeln(min); exit; end;
end;
writeln('No Solution');
end;
begin
readln(casen);
for casei:=1 to casen do
begin
init;
main;
end;
end.
const
maxn=1000;
type
node=record
u,v,w:longint;
end;
var
n,m,casen,casei:longint;
e:array[1..maxn] of node;
t,r,d:array[-1..maxn] of longint;
procedure add(x,y,z:longint);
begin
inc(m);
with e[m] do
begin
u:=x; v:=y; w:=z;
end;
end;
procedure init;
var
i,x:longint;
begin
for i:=0 to 23 do read(r[i]);
readln;
readln(n);
fillchar(t,sizeof(t),0);
for i:=1 to n do
begin readln(x); inc(t[x]); end;
end;
procedure main;
var
i,j,min:longint;
check:boolean;
begin
for min:=1 to n do
begin
m:=0;
for i:=-1 to 23 do
d[i]:=100000;
add(23,-1,-min);
for i:=0 to 23 do
begin
add(i,i-1,0); add(i-1,i,t[i]);
if i>7 then add(i,i-8,-r[i]) else add(i,i+16,-r[i]+min);
end;
for j:=0 to 23 do
begin
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then
begin
d[v]:=d[u]+w;
check:=false;
end;
if check then break;
end;
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then begin check:=false; break; end;
if check then begin writeln(min); exit; end;
end;
writeln('No Solution');
end;
begin
readln(casen);
for casei:=1 to casen do
begin
init;
main;
end;
end.
const
maxn=1000;
type
node=record
u,v,w:longint;
end;
var
n,m,casen,casei:longint;
e:array[1..maxn] of node;
t,r,d:array[-1..maxn] of longint;
procedure add(x,y,z:longint);
begin
inc(m);
with e[m] do
begin
u:=x; v:=y; w:=z;
end;
end;
procedure init;
var
i,x:longint;
begin
for i:=0 to 23 do read(r[i]);
readln;
readln(n);
fillchar(t,sizeof(t),0);
for i:=1 to n do
begin readln(x); inc(t[x]); end;
end;
procedure main;
var
i,j,min:longint;
check:boolean;
begin
for min:=1 to n do
begin
m:=0;
for i:=-1 to 23 do
d[i]:=100000;
add(23,-1,-min);
for i:=0 to 23 do
begin
add(i,i-1,0); add(i-1,i,t[i]);
if i>7 then add(i,i-8,-r[i]) else add(i,i+16,-r[i]+min);
end;
for j:=0 to 23 do
begin
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then
begin
d[v]:=d[u]+w;
check:=false;
end;
if check then break;
end;
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then begin check:=false; break; end;
if check then begin writeln(min); exit; end;
end;
writeln('No Solution');
end;
begin
readln(casen);
for casei:=1 to casen do
begin
init;
main;
end;
end.
const
maxn=1000;
type
node=record
u,v,w:longint;
end;
var
n,m,casen,casei:longint;
e:array[1..maxn] of node;
t,r,d:array[-1..maxn] of longint;
procedure add(x,y,z:longint);
begin
inc(m);
with e[m] do
begin
u:=x; v:=y; w:=z;
end;
end;
procedure init;
var
i,x:longint;
begin
for i:=0 to 23 do read(r[i]);
readln;
readln(n);
fillchar(t,sizeof(t),0);
for i:=1 to n do
begin readln(x); inc(t[x]); end;
end;
procedure main;
var
i,j,min:longint;
check:boolean;
begin
for min:=1 to n do
begin
m:=0;
for i:=-1 to 23 do
d[i]:=100000;
add(23,-1,-min);
for i:=0 to 23 do
begin
add(i,i-1,0); add(i-1,i,t[i]);
if i>7 then add(i,i-8,-r[i]) else add(i,i+16,-r[i]+min);
end;
for j:=0 to 23 do
begin
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then
begin
d[v]:=d[u]+w;
check:=false;
end;
if check then break;
end;
check:=true;
for i:=1 to m do
with e[i] do
if d[v]>d[u]+w then begin check:=false; break; end;
if check then begin writeln(min); exit; end;
end;
writeln('No Solution');
end;
begin
readln(casen);
for casei:=1 to casen do
begin
init;
main;
end;
end.
扫一扫
344
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
