Related
There is a well known problem called "Triple Step" that states:
"A child is running up a staircase with n steps and can hop either 1 step, 2 steps, or 3 steps at a time. Implement a method to count how many possible ways the child can run up the stairs"
The algorithm below is a version without memoization:
int countWays(int n) {
if (n < 0) {
return 0;
} else if (n == 0) {
return 1;
} else {
return countWays(n-1) + countWays(n-2) + countWays(n-3);
}
}
I know that its runtime can be improved from the exponential time complexity.
But I really would like to know how to build a dynamic programming table over this problem,
for example I tried the table below for n being 4 steps:
0 | 1 | 2 | 3 | 4 <= staircase size
1 1 | 1 | 1 | 1 | 1
2 1 | 1 | 2 | 2 | 3
3 1 | 1 | 2 | 3 | 4 <=** There's something wrong because for n=4 the output should be 7
Could someone give me a hint about how this table could be built for the problem above? (or maybe the table is fine and I'm not able to interpret it right)
Thanks!
As you mentioned that countWays(N) can be solved by taking sum of countWays(N-1), countWays(N-2) and countWays(N-3).
Since we know the answer for n<=0, we can start constructing our solution from n=0 to n=N and at any point of time we will always have N-1, N-2 and N-3 values ready to be used.
In the process of constructing solution from n=0 to n=N at any point of time we should have results of our earlier calculations stored somewhere.
you can take 3 variables to store these values and keep updating these 3 variables at each iteration to store the last 3 calculations.
int countWays(int n) {
int last = 1; // for n = 0
int secondLast = 0; // for n = -1
int thridLast = 0; // for n = -2
for(int i = 1 ; i <= n ; i++) {
int current = last + secondLast + thirdLast;
thirdLast = secondLast;
secondLast = last;
last = current;
}
return last;
}
instead of taking 3 variables you can store all the earlier calculations in an array and the code will look like this,
int countWays(int n) {
if(n<0) return 0;
int[] a = new int[n+3];
a[0] = 0;
a[1] = 0;
a[2] = 1; // stores the result for N=0
for(int i = 3 ; i < n+3 ; i++) {
a[i] = a[i-1] + a[i-2] + a[i-3];
}
return a[n+2];
}
and array will look like,
Answer -> [0, 0, 1, 1, 2, 4, 7]
Value Of N -> -2, -1,0 ,1, 2, 3, 4
Array created in this solutions is known is dynamic programming table also known as memoization or bottom up approach to DP
Run time complexity of above solution is O(N)
There is another way to solve these type of problems in O(Log N) time complexity, where solution can be described in terms of a linear recurrence relation.
The solution is known as Matrix Exponentiation, follow this link for more explanation - https://discuss.codechef.com/t/building-up-the-recurrence-matrix-to-compute-recurrences-in-o-logn-time/570
The table for this is 1d which is the staircase size, on every step x you add x-1, x-2, and x-3 if possible, for example:
0 | 1 | 2 | 3 | 4 <= staircase size
1st step 1 | 1 | 0 | 0 | 0 only x-1 is possible
2nd step 1 | 1 | 2 | 0 | 0 x-1 + x-2 are possible
3rd step 1 | 1 | 2 | 4 | 0 x-1 + x-2 + x-3 are possible
4th step 1 | 1 | 2 | 4 | 7 x-1 + x-2 + x-3 are possible
More explanation:
Step 1:
only reachable by 1-step
Step 2:
1-step + 1-step
2-steps
Step 3:
1-step + 1-step + 1-step
1-step + 2-steps
2-steps + 1-step
3-steps
Step 4:
1-step + 1-step + 1-step + 1-step
2-steps + 1-step + 1-step
1-step + 2-steps + 1-step
1-step + 1-step + 2-steps
1-step + 3-steps
3-steps + 1-step
2-steps + 2-steps
Assume that you are using dynamic programming to solve the problem.
Let a be the name of the variable of your table.
The formulae for a[n] with n = 0, 1, 2, .... are as you mentioned:
a[0] = 1
a[n] = a[n-1] + a[n-2] + a[n-3]
Be sure that a[n] for n < 0 is 0 always.
The answer for staircase size = 4 can be solved only if all the answers for 0 <= staircase size < 4 are given. i.e., a[4] can be calculated only if a[0], a[1], ..., a[3] are calculated.
The answer for staircase size = 3 can be solved only if all the answers for 0 <= staircase size < 3 are given. i.e., a[3] can be calculated only if a[0], ..., a[2] are calculated.
The answer for staircase size = 2 can be solved only if all the answers for 0 <= staircase size < 2 are given. i.e., a[2] can be calculated only if a[0], a[1] are calculated.
The answer for staircase size = 1 can be solved only if all the answers for 0 <= staircase size < 1 are given. i.e., a[1] can be calculated only if a[0] is calculated.
a[0] is the first formula.
Here, you can start.
a[0] = 1 // Initialization
a[1] = a[0] + a[-1] + a[-2] = a[0] + 0 + 0 // calculated at 1st loop (a[1] = 1)
a[2] = a[1] + a[0] + a[-1] = a[1] + a[0] + 0 // calculated at 2nd loop (a[2] = 1 + 1)
a[3] = a[2] + a[1] + a[0] // calculated at 3rd loop (a[3] = 2 + 1 + 1)
a[4] = a[3] + a[2] + a[1] // calculated at 4th loop (a[4] = 4 + 2 + 1)
...
a[n] = a[n-1] + a[n-2] + a[n-3] // calculated at nth loop
I have a 5x-5 maze specified as follows.
r = [1 0 1 1 1
1 1 1 0 1
0 1 0 0 1
1 1 1 0 1
1 0 1 0 1];
Where 1's are the paths and 0's are the walls.
Assume I have a function foo(policy_vector, r) that maps the elements of the policy vector to the elements in r. For example 1=UP, 2=Right, 3=Down, 4=Left. The MDP is set up such that the wall states are never realized so policies for those states are ignored in the plot.
policy_vector' = [3 2 2 2 3 2 2 1 2 3 1 1 1 2 3 2 1 4 2 3 1 1 1 2 2]
symbols' = [v > > > v > > ^ > v ^ ^ ^ > v > ^ < > v ^ ^ ^ > >]
I am trying to display my policy decision for a Markov Decision Process in the context of solving a maze. How would I plot something that looks like this? Matlab is preferable but Python is fine.
Even if some body could show me how to make a plot like this I would be able to figure it out from there.
function[] = policy_plot(policy,r)
[row,col] = size(r);
symbols = {'^', '>', 'v', '<'};
policy_symbolic = get_policy_symbols(policy, symbols);
figure()
hold on
axis([0, row, 0, col])
grid on
cnt = 1;
fill([0,0,col,col],[row,0,0,row],'k')
for rr = row:-1:1
for cc = 1:col
if r(row+1 - rr,cc) ~= 0 && ~(row == row+1 - rr && col == cc)
fill([cc-1,cc-1,cc,cc],[rr,rr-1,rr-1,rr],'g')
text(cc - 0.55,rr - 0.5,policy_symbolic{cnt})
end
cnt = cnt + 1;
end
end
fill([cc-1,cc-1,cc,cc],[rr,rr-1,rr-1,rr],'b')
text(cc - 0.70,rr - 0.5,'Goal')
function [policy_symbolic] = get_policy_symbols(policy, symbols)
policy_symbolic = cell(size(policy));
for ii = 1:length(policy)
policy_symbolic{ii} = symbols{policy(ii)};
end
I have a string S which consists of a's and b's. Perform the below operation once. Objective is to obtain the lexicographically smallest string.
Operation: Reverse exactly one substring of S
e.g.
if S = abab then Output = aabb (reverse ba of string S)
if S = abba then Output = aabb (reverse bba of string S)
My approach
Case 1: If all characters of the input string are same then output will be the string itself.
Case 2: if S is of the form aaaaaaa....bbbbbb.... then answer will be S itself.
otherwise: Find the first occurence of b in S say the position is i. String S will look like
aa...bbb...aaaa...bbbb....aaaa....bbbb....aaaaa...
|
i
In order to obtain the lexicographically smallest string the substring that will be reversed starts from index i. See below for possible ending j.
aa...bbb...aaaa...bbbb....aaaa....bbbb....aaaaa...
| | | |
i j j j
Reverse substring S[i:j] for every j and find the smallest string.
The complexity of the algorithm will be O(|S|*|S|) where |S| is the length of the string.
Is there a better way to solve this problem? Probably O(|S|) solution.
What I am thinking if we can pick the correct j in linear time then we are done. We will pick that j where number of a's is maximum. If there is one maximum then we solved the problem but what if it's not the case? I have tried a lot. Please help.
So, I came up with an algorithm, that seems to be more efficient that O(|S|^2), but I'm not quite sure of it's complexity. Here's a rough outline:
Strip of the leading a's, storing in variable start.
Group the rest of the string into letter chunks.
Find the indices of the groups with the longest sequences of a's.
If only one index remains, proceed to 10.
Filter these indices so that the length of the [first] group of b's after reversal is at a minimum.
If only one index remains, proceed to 10.
Filter these indices so that the length of the [first] group of a's (not including the leading a's) after reversal is at a minimum.
If only one index remains, proceed to 10.
Go back to 5, except inspect the [second/third/...] groups of a's and b's this time.
Return start, plus the reversed groups up to index, plus the remaining groups.
Since any substring that is being reversed begins with a b and ends in an a, no two hypothesized reversals are palindromes and thus two reversals will not result in the same output, guaranteeing that there is a unique optimal solution and that the algorithm will terminate.
My intuition says this approach of probably O(log(|S|)*|S|), but I'm not too sure. An example implementation (not a very good one albeit) in Python is provided below.
from itertools import groupby
def get_next_bs(i, groups, off):
d = 1 + 2*off
before_bs = len(groups[i-d]) if i >= d else 0
after_bs = len(groups[i+d]) if i <= d and len(groups) > i + d else 0
return before_bs + after_bs
def get_next_as(i, groups, off):
d = 2*(off + 1)
return len(groups[d+1]) if i < d else len(groups[i-d])
def maximal_reversal(s):
# example input: 'aabaababbaababbaabbbaa'
first_b = s.find('b')
start, rest = s[:first_b], s[first_b:]
# 'aa', 'baababbaababbaabbbaa'
groups = [''.join(g) for _, g in groupby(rest)]
# ['b', 'aa', 'b', 'a', 'bb', 'aa', 'b', 'a', 'bb', 'aa', 'bbb', 'aa']
try:
max_length = max(len(g) for g in groups if g[0] == 'a')
except ValueError:
return s # no a's after the start, no reversal needed
indices = [i for i, g in enumerate(groups) if g[0] == 'a' and len(g) == max_length]
# [1, 5, 9, 11]
off = 0
while len(indices) > 1:
min_bs = min(get_next_bs(i, groups, off) for i in indices)
indices = [i for i in indices if get_next_bs(i, groups, off) == min_bs]
# off 0: [1, 5, 9], off 1: [5, 9], off 2: [9]
if len(indices) == 1:
break
max_as = max(get_next_as(i, groups, off) for i in indices)
indices = [i for i in indices if get_next_as(i, groups, off) == max_as]
# off 0: [1, 5, 9], off 1: [5, 9]
off += 1
i = indices[0]
groups[:i+1] = groups[:i+1][::-1]
return start + ''.join(groups)
# 'aaaabbabaabbabaabbbbaa'
TL;DR: Here's an algorithm that only iterates over the string once (with O(|S|)-ish complexity for limited string lengths). The example with which I explain it below is a bit long-winded, but the algorithm is really quite simple:
Iterate over the string, and update its value interpreted as a reverse (lsb-to-msb) binary number.
If you find the last zero of a sequence of zeros that is longer than the current maximum, store the current position, and the current reverse value. From then on, also update this value, interpreting the rest of the string as a forward (msb-to-lsb) binary number.
If you find the last zero of a sequence of zeros that is as long as the current maximum, compare the current reverse value with the current value of the stored end-point; if it is smaller, replace the end-point with the current position.
So you're basically comparing the value of the string if it were reversed up to the current point, with the value of the string if it were only reversed up to a (so-far) optimal point, and updating this optimal point on-the-fly.
Here's a quick code example; it could undoubtedly be coded more elegantly:
function reverseSubsequence(str) {
var reverse = 0, max = 0, first, last, value, len = 0, unit = 1;
for (var pos = 0; pos < str.length; pos++) {
var digit = str.charCodeAt(pos) - 97; // read next digit
if (digit == 0) {
if (first == undefined) continue; // skip leading zeros
if (++len > max || len == max && reverse < value) { // better endpoint found
max = len;
last = pos;
value = reverse;
}
} else {
if (first == undefined) first = pos; // end of leading zeros
len = 0;
}
reverse += unit * digit; // update reverse value
unit <<= 1;
value = value * 2 + digit; // update endpoint value
}
return {from: first || 0, to: last || 0};
}
var result = reverseSubsequence("aaabbaabaaabbabaaabaaab");
document.write(result.from + "→" + result.to);
(The code could be simplified by comparing reverse and value whenever a zero is found, and not just when the end of a maximally long sequence of zeros is encountered.)
You can create an algorithm that only iterates over the input once, and can process an incoming stream of unknown length, by keeping track of two values: the value of the whole string interpreted as a reverse (lsb-to-msb) binary number, and the value of the string with one part reversed. Whenever the reverse value goes below the value of the stored best end-point, a better end-point has been found.
Consider this string as an example:
aaabbaabaaabbabaaabaaab
or, written with zeros and ones for simplicity:
00011001000110100010001
We iterate over the leading zeros until we find the first one:
0001
^
This is the start of the sequence we'll want to reverse. We will start interpreting the stream of zeros and ones as a reversed (lsb-to-msb) binary number and update this number after every step:
reverse = 1, unit = 1
Then at every step, we double the unit and update the reverse number:
0001 reverse = 1
00011 unit = 2; reverse = 1 + 1 * 2 = 3
000110 unit = 4; reverse = 3 + 0 * 4 = 3
0001100 unit = 8; reverse = 3 + 0 * 8 = 3
At this point we find a one, and the sequence of zeros comes to an end. It contains 2 zeros, which is currently the maximum, so we store the current position as a possible end-point, and also store the current reverse value:
endpoint = {position = 6, value = 3}
Then we go on iterating over the string, but at every step, we update the value of the possible endpoint, but now as a normal (msb-to-lsb) binary number:
00011001 unit = 16; reverse = 3 + 1 * 16 = 19
endpoint.value *= 2 + 1 = 7
000110010 unit = 32; reverse = 19 + 0 * 32 = 19
endpoint.value *= 2 + 0 = 14
0001100100 unit = 64; reverse = 19 + 0 * 64 = 19
endpoint.value *= 2 + 0 = 28
00011001000 unit = 128; reverse = 19 + 0 * 128 = 19
endpoint.value *= 2 + 0 = 56
At this point we find that we have a sequence of 3 zeros, which is longer that the current maximum of 2, so we throw away the end-point we had so far and replace it with the current position and reverse value:
endpoint = {position = 10, value = 19}
And then we go on iterating over the string:
000110010001 unit = 256; reverse = 19 + 1 * 256 = 275
endpoint.value *= 2 + 1 = 39
0001100100011 unit = 512; reverse = 275 + 1 * 512 = 778
endpoint.value *= 2 + 1 = 79
00011001000110 unit = 1024; reverse = 778 + 0 * 1024 = 778
endpoint.value *= 2 + 0 = 158
000110010001101 unit = 2048; reverse = 778 + 1 * 2048 = 2826
endpoint.value *= 2 + 1 = 317
0001100100011010 unit = 4096; reverse = 2826 + 0 * 4096 = 2826
endpoint.value *= 2 + 0 = 634
00011001000110100 unit = 8192; reverse = 2826 + 0 * 8192 = 2826
endpoint.value *= 2 + 0 = 1268
000110010001101000 unit = 16384; reverse = 2826 + 0 * 16384 = 2826
endpoint.value *= 2 + 0 = 2536
Here we find that we have another sequence with 3 zeros, so we compare the current reverse value with the end-point's value, and find that the stored endpoint has a lower value:
endpoint.value = 2536 < reverse = 2826
so we keep the end-point set to position 10 and we go on iterating over the string:
0001100100011010001 unit = 32768; reverse = 2826 + 1 * 32768 = 35594
endpoint.value *= 2 + 1 = 5073
00011001000110100010 unit = 65536; reverse = 35594 + 0 * 65536 = 35594
endpoint.value *= 2 + 0 = 10146
000110010001101000100 unit = 131072; reverse = 35594 + 0 * 131072 = 35594
endpoint.value *= 2 + 0 = 20292
0001100100011010001000 unit = 262144; reverse = 35594 + 0 * 262144 = 35594
endpoint.value *= 2 + 0 = 40584
And we find another sequence of 3 zeros, so we compare this position to the stored end-point:
endpoint.value = 40584 > reverse = 35594
and we find it has a smaller value, so we replace the possible end-point with the current position:
endpoint = {position = 21, value = 35594}
And then we iterate over the final digit:
00011001000110100010001 unit = 524288; reverse = 35594 + 1 * 524288 = 559882
endpoint.value *= 2 + 1 = 71189
So at the end we find that position 21 gives us the lowest value, so it is the optimal solution:
00011001000110100010001 -> 00000010001011000100111
^ ^
start = 3 end = 21
Here's a C++ version that uses a vector of bool instead of integers. It can parse strings longer than 64 characters, but the complexity is probably quadratic.
#include <vector>
struct range {unsigned int first; unsigned int last;};
range lexiLeastRev(std::string const &str) {
unsigned int len = str.length(), first = 0, last = 0, run = 0, max_run = 0;
std::vector<bool> forward(0), reverse(0);
bool leading_zeros = true;
for (unsigned int pos = 0; pos < len; pos++) {
bool digit = str[pos] - 'a';
if (!digit) {
if (leading_zeros) continue;
if (++run > max_run || run == max_run && reverse < forward) {
max_run = run;
last = pos;
forward = reverse;
}
}
else {
if (leading_zeros) {
leading_zeros = false;
first = pos;
}
run = 0;
}
forward.push_back(digit);
reverse.insert(reverse.begin(), digit);
}
return range {first, last};
}
Given n = 1 to 10^5, stored as a string in decimal format.
Example: If n = 968, then out of all subsequences i.e 9, 6, 8, 96, 68, 98, 968 there are 3 sub-sequences of it, i.e 968, 96 and 8, that are divisible by 8. So, the answer is 3.
Since the answer can be very large, print the answer modulo (10^9 + 7).
You can use dynamic programming. Let f(len, sum) be the number of subsequences of the prefix of length len such that their sum is sum modulo 8 (sum ranges from 0 to 7).
The value of f for len = 1 is obvious. The transitions go as follows:
We can start a new subsequence in the new position: f(len, a[i] % 8) += 1.
We can continue any subsequence from the shorter prefix:
for old_sum = 0..7
f(len, (old_sum * 10 + a[i]) % 8) += f(len - 1, old_sum) // take the new element
f(len, old_sum) += f(len - 1, old_sum) // ignore the new element
Of course, you can perform all computations module 10^9 + 7 and use a standard integer type.
The answer is f(n, 0) (all elements are taken into account and the sum modulo 8 is 0).
The time complexity of this solution is O(n) (as there are O(n) states and 2 transition from each of them).
Note: if the numbers can't have leading zeros, you can just one more parameter to the state: a flag that indicates whether the first element of the subsequence is zero (this sequences should never be extended). The rest of the solution stays the same.
Note: This answer assumes you mean contiguous subsequences.
The divisibility rule for a number to be divisible by 8 is if the last three digits of the number are divisible by 8. Using this, a simple O(n) algorithm can be obtained where n is the number of digits in the number.
Let N=a_0a_1...a_(n-1) be the decimal representation of N with n digits.
Let the number of sequences so far be s = 0
For each set of three digits, a_i a_(i+1) a_(i+2), check if the number is divisible by 8. If so, add i + 1 to the number of sequences, i.e., s = s + i. This is because all strings a_k..a_(i+2) will be divisible by 8 for k ranging from 0..i.
Loop i from 0 to n-2-1 and continue.
So, if you have 1424968, the subsequences divisible are at:
i=1 (424 yielding i+1 = 2 numbers: 424 and 1424)
i=3 (496 yielding i+1 = 4 numbers: 496, 2496, 42496, 142496)
i=4 (968 yielding i+1 = 5 numbers: 968, 4968, 24968, 424968, 1424968)
Note that some small modifications will be needed to consider numbers lesser than three digits in length.
Hence the total number of sequences = 2 + 4 + 5 = 11. Total complexity = O(n) where n is the number of digits.
One can use the fact that for any three-digit number abc the following holds:
abc % 8 = ((ab % 8) * 10 + c) % 8
Or in other words: the test for a number with a fixed start-index can be cascaded:
int div8(String s){
int total = 0, mod = 0;
for(int i = 0; i < s.length(); i++)
{
mod = (mod * 10 + s.charAt(i) - '0') % 8
if(mod == 0)
total++;
}
return total;
}
But we don't have fixed start-indices!
Well, that's pretty easy to fix:
Suppose two sequences a and b, such that int(a) % 8 = int(b) % 8 and b is a suffix of a. No matter what how the sequence continues, the modulos of a and b will always remain equal. Thus it's sufficient to keep track of the number of sequences that share the property of having an equal value modulo 8.
final int RESULTMOD = 1000000000 + 7;
int div8(String s){
int total = 0;
//modtable[i] is the number of subsequences with int(sequence) % 8 = i
int[] modTable = new int[8];
for(int i = 0; i < s.length(); i++){
int[] nextTable = new int[8];
//transform table from last loop-run (shared modulo)
for(int j = 0; j < 8; j++){
nextTable[(j * 10 + s.charAt(i) - '0') % 8] = modTable[j] % RESULTMOD;
}
//add the sequence that starts at this index to the appropriate bucket
nextTable[(s.charAt(i) - '0') % 8]++;
//add the count of all sequences with int(sequence) % 8 = 0 to the result
total += nextTable[0];
total %= RESULTMOD;
//table for next run
modTable = nextTable;
}
return total;
}
Runtime is O(n).
There are 10 possible states a subsequence can be in. The first is empty. The second is that there was a leading 0. And the other 8 are a ongoing number that is 0-7 mod 8. You start at the beginning of the string with 1 way of being empty, no way to be anything else. At the end of the string your answer is the number of ways to have a leading 0 plus an ongoing number that is 0 mod 8.
The transition table should be obvious. The rest is just normal dynamic programming.
Where I'm at
I'm trying to figure out how many beers I can buy with 10 RMB after recycling every bottle I get. It's obvious to me that I'm doing something wrong, procedurally, but it's not occurring to me what that is. I'm currently reading "How To Think Like a Computer Scientist: Think Python" on chapter 9. I feel like this should be an easy program for me, but I'm not sure how to loop in the recycling portion of the app. What would be the most concise way to rinse and repeat beer purchases?
The question
Basically, one beer costs 2 RMB. 2 bins gets 1 RMB. 4 caps gets 1 RMB. I'm starting out with 10 RMB. How many beers can I buy (recycling all the bins and caps)?
#5 bottles 5 caps
#= 3 rmb + 1 caps 1 bottles
#6th bottle bought
#= 2rmb + 2 caps
#7th bottle bought
#= 0rmb + 3 caps 1 bottles.
import math
def countbeers(rmb):
beers = 0;
caps = 0;
bins = 0;
bcost = 2;
for i in range (0,rmb):
beers += 1/2
for i in range (0,math.floor(beers)):
caps += 1
bins += 1
rmb = rmb - bcost
for i in range (0,caps):
rmb += 1/4
for i in range (0,bins):
rmb += 1/2
# if rmb > 2 what goes here, trying to loop back through
return beers
print(countbeers(10))
Second attempt
#5 bottles 5 caps
#= 3 wallet + 1 caps 1 bottles
#6th bottle bought
#= 2wallet + 2 caps
#7th bottle bought
#= 0wallet + 3 caps 1 bottles.
import math
global beers
global caps
global bins
global bcost
beers = 0
caps = 0
bins = 0
bcost = 2
def buybeers(wallet):
beers = 0
for i in range (0,wallet):
beers += 1/2
wallet -= 2
return beers
def drinkbeers(beers):
for i in range (0,math.floor(beers)):
caps += 1
bins += 1
wallet = wallet - bcost
return wallet, caps, bins
def recycle(caps, bins):
for i in range (0,caps):
wallet += 1/4
for i in range (0,bins):
wallet += 1/2
return wallet
def maxbeers(wallet):
if wallet > 2:
buybeers(wallet)
if math.floor(beers) > 1:
drinkbeers(beers)
if caps > 4 | bins > 2:
recycle(caps, bins)
return wallet
wallet = int(input("How many wallet do you have?"))
maxbeers(wallet)
if wallet >= 2:
maxbeers(wallet)
elif wallet < 2:
print(beers)
Your main problem is that you are not looping. Every beer you bought from rmb gives you one more bottle, and one more cap. This new bottle and cap might be enough to earn you another rmb, which might be enough for another beer. Your implementation handles this to a limited extent, since you call maxbeers multiple times, but it will not give the correct answer if you give it a truckload of beers, i.e. 25656 bottles.
If you know the number of rmb you have, you can do the calculation by hand on paper and write this:
def maxbeers(rmb):
return 7 # totally correct, I promise. Checked this by hand.
but that's no fun. What if rmb is 25656?
Assuming we can exchange:
2 bottles -> 1 rmb
4 caps -> 1 rmb
2 rmb -> 1 beer + 1 cap + 1 bottle
we calculate it like this, through simulation:
def q(rmb):
beers = 0
caps = 0
bottles = 0
while rmb > 0:
# buy a beer with rmb
rmb -= 2
beers += 1
caps += 1
bottles += 1
# exchange all caps for rmb
while caps >= 4:
rmb += 1
caps -= 4
# exchange all bottles for rmb
while bottles >= 2:
rmb += 1
bottles -= 2
return beers
for a in range(20):
print("rmb:", a, "beers:", q(a))
Then we can buy 20525 beers.