Coding Challenge - Morris Sequence

[PeejayAdams]

Yes, these strings get huge, don't they? The whole thing really slows down in a very visible way once the lengths start to reach the 100,000s (40th iteration on) and gets ultra slow from there on in. I'll leave mine running for a while and see exactly where it crashes and burns. Max chars for a StringBuilder are 2,147,483,647, I believe and it's clearly going to hit that at some point so I guess the true answer as you suggest involves hiving off the repeating elements. My code (spectacularly Q&D C# - written on a whim in "do ... while" loops because someone recently said "for" loops are for dinosaurs) looks like this:

    static void Main(string\[\] args)
    {
        string result = ("1");

        int i = 1;

        do
        {
            Console.WriteLine(i.ToString() + "\\t" + result.Length.ToString());

            result = GetNextResult(result);
            i += 1;
        }
        while (i < 101);


        Console.ReadLine();

    }

    static string GetNextResult(string origin)
    {
        char character;
        int occurences;
        StringBuilder result = new StringBuilder();

        do
        {
            if (origin.Length == 0)
                break;

            ProcessNextGroup(origin, out character, out occurences);

            origin = origin.Substring(occurences);

            result.Append(occurences.ToString());
            result.Append(character.ToString());

            GC.Collect();

        }
        while (true);

        return result.ToString();

    }

    static void ProcessNextGroup(string sequence, out char character, out int occurences)
    {
        int index = 0;
        character = sequence\[0\];
        occurences = 1;

        do
        {
            if (++index > (sequence.Length -1))
                break;

            if (sequence\[index\] == character)
                occurences += 1;
            else
                break;
        }
        while (true);
    }
}

98.4% of statistics are made up on the spot.

[Sascha Lefevre]

PeejayAdams wrote:

written on a whim in "do ... while" loops because someone recently said "for" loops are for dinosaurs

:laugh:

If the brain were so simple we could understand it, we would be so simple we couldn't. — Lyall Watson

[Kenneth Haugland]

No, the problem isnt that the strings are limited. You can load into memory the items you want, and perform the LookNSay from the partial string. When that is read to its end you simply load the next part into memory. Its just a little more complicated that's all:

string LookAndSayFile(int S, int n)
{
//If the string is above this length split it into multiple strings
int SplitStringSize = 2;

        //Setting the directory where files are stored relativce to the exe file
        string projectPath = System.IO.Path.GetFullPath(@"..\\..\\..\\..\\");

        //Delete files
        File.Delete(projectPath + "output.txt");
        File.Delete(projectPath + "input.txt");

        //Add intput and and empty file 
        System.IO.File.WriteAllLines(projectPath + "output.txt", new string\[\] { string.Empty });
        System.IO.File.WriteAllLines(projectPath + "input.txt", new string\[\] { S.ToString() });


        //Number of iterations you want to perform
        for (int i = 0; i < (n - 1); i++)
        {

            char currentItem = '#';
            int count = 1;
            StringBuilder NewString = new StringBuilder();

            using (StreamReader sr = new StreamReader(projectPath + "input.txt"))
            {
                while (sr.Peek() >= 0)
                {

                    string res = sr.ReadLine();

                    foreach (char j in res)
                    {
                        if (currentItem == '#')
                        {
                            currentItem = j;
                            continue;
                        }

                        if (currentItem == j)
                        {
                            count++;
                        }
                        else
                        {
                            NewString.Append(count.ToString() + currentItem.ToString());
                            if (NewString.Length > SplitStringSize)
                            {
                                System.IO.File.AppendAllLines(projectPath + "output.txt", new string\[\] { NewString.ToString() });
                                NewString.Length = 0;
                            }
                            currentItem = j;
                            count = 1;
                            continue

[Kenneth Haugland]

Oh, the integer sequence are difined for a bit larger range than Dave gave here: A006751 - OEIS[^]

[Dave Kreskowiak]

:laugh: Not so easy, is it?

System.ItDidntWorkException: Something didn't work as expected. C# - How to debug code[^]. Seriously, go read these articles.
Dave Kreskowiak

[Dave Kreskowiak]

That's cheating! :-D Oh, and what you linked to starts the sequence with a 2, not a 1.

System.ItDidntWorkException: Something didn't work as expected. C# - How to debug code[^]. Seriously, go read these articles.
Dave Kreskowiak

[PIEBALDconsult]

Strings? Why not use a List of Tuple ? :) After more thought and before reading Richard's response... Tuple Using a byte to implement such a Tuple. Having read Richard's response... Huh, yeah, use a byte to implement a Tuple,Tuple>... :D I would also try to use many threads.

[Kenneth Haugland]

I have been running the code (which I posted here somewhere) in the background as I'm reading for the exam. The text file that just holds the current number is 1.6 GB, and I guess that's one way of measuring the length. So I'll definitely post the end number on the Lounge forum :d Then it will at last contain something worthwhile :laugh:

[Richard Deeming]

This one's a real bugger for memory. String-based approaches are obviously out - 16 bits to store each character is overkill when the only symbols you need to store are 1, 2 and 3. List<byte> is obviously not going to work, because it would need to allocate an array big enough to hold the entire sequence. LinkedList<byte> has to create an object for every byte in the list, so the overhead far outweighs the payload. I settled on a custom singly-linked list of byte arrays, re-using two instances (previous and current) to reduce memory churn. But even that was eating huge amounts of memory. Finally, realising that the only numbers in the sequence are 1, 2 and 3, I decided to stuff four numbers into each byte, which brings the memory usage under control. However, it still takes a damn long time to run, and I haven't left it for long enough to get to the 100th iteration yet. Morris Sequence · GitHub[^] Having spent far too long thinking about this, now's the time for you to tell me there's some secret trick to calculate the sequence without having to store the whole thing. :-D

---

"These people looked deep within my soul and assigned me a number based on the order in which I joined." - Homer

[Kenneth Haugland]

PS. Do you want the text file? ... ... ... :laugh: :laugh: :laugh:

[GuyThiebaut]

Here we go - breaking the no code in the lounge rule here...

internal void Stream(int upTo)
{

    using (StreamWriter writer = new StreamWriter("E:\\\\temp\\\\MorrisSequence\\\\" + "line1.txt"))
    {
        writer.Write("1");
    }

    for (int i = 1; i <= upTo; i++)
    {
        Console.WriteLine(i);

        using (StreamReader reader = new StreamReader("E:\\\\temp\\\\MorrisSequence\\\\" + "line" + i + ".txt"))
        {
            int count = 1;
            char currChar = (char)reader.Read();
            char lastChar = currChar;
            char nextChar;
            string writeNum = (i + 1).ToString();

            using (StreamWriter writer = new StreamWriter("E:\\\\temp\\\\MorrisSequence\\\\" + "line" + writeNum + ".txt"))
            {
                while (reader.Peek() >= 0)
                {
                    nextChar = (char)reader.Peek();
                    if (nextChar != lastChar)
                    {
                        writer.Write(count.ToString() + currChar.ToString());
                        count = 0;
                    }
                    currChar= (char)reader.Read();
                    lastChar = currChar;
                    count++;
                }
                writer.Write(count.ToString() + currChar.ToString());
            }
        }
    }
}

[edit] small tidy up, giving filenames proper names that relate to what they contain.

“That which can be asserted without evidence, can be dismissed without evidence.”

― Christopher Hitchens

[Kirk 10389821]

I used a new AI program I trained on twitter... It responded: 1 Really large string of numbers NOBODY cares about, just like you. Delete yourself. I added "Plz" and it simply said "Go away Troll" I am having second thoughts about it having access to: - NEST devices (including garage door, locking all doors) - Internet access to unlock and start my car - IP Phone / Router... So I can't call for hel...

[Richard Deeming]

That's one way around it. But I hope you've got a large SSD! :-D Is there a reason you're writing strings instead of bytes?

---

"These people looked deep within my soul and assigned me a number based on the order in which I joined." - Homer

[Kenneth Haugland]

Doesn't look like there is a know shortcut: Look-and-say sequence - Rosetta Code[^] But they do post a formula here: A005150 - OEIS[^] But it looks complicated. I used chars, and the txt file is 2 GB now :S

[GuyThiebaut]

I was tempted to leave it running this evening but as you mention I think it will fill up the disk space - file 77 is just over 1 gig in size and it's only a text file. What I may do is compress then delete files prior to the one I am currently reading(the 1 gig file compresses to 80mb largely because it is composed of 1s,2s and 3s).

Richard Deeming wrote:

Is there a reason you're writing strings instead of bytes?

Um er yes, um err, um er because... that idea never occurred to me - thanks for the tip :thumbsup:

“That which can be asserted without evidence, can be dismissed without evidence.”

― Christopher Hitchens

[Kenneth Haugland]

I have already text files over 2.2 GB so I think you'll have to delete them as you gom at least that's what I do. And I think using bytes is cheating :laugh: also I didn't know that 3 would be the highest number. I don't think is enough not if you start at 3,4,5 or any other number, at least I got some 5 then. Or my code was wrong.

[GuyThiebaut]

My comment about 1,2,3 was based on looking at file 50 very briefly. I could run an analysis as I go through them. I have re-written it so that it zips and deletes files that are not being read - let's see how quickly my computer or hard drive goes up in a puff of smoke... My guess is that it may be one of those tasks where it is not possible to calculate up to 100 within the lifetime of the universe.

“That which can be asserted without evidence, can be dismissed without evidence.”

― Christopher Hitchens

[Kenneth Haugland]

GuyThiebaut wrote:

My guess is that it may be one of those tasks where it is not possible to calculate up to 100 within the lifetime of the universe.

Nah, I don't think so, I was able to run up to 77 before VS threw an out of memory exception. And as I told Richard, I think I found a formula, but It looked kind of complicated.

[Kenneth Haugland]

Well, he did only ask about the length of the 100 th number. So according to Look-and-say sequence - Wikipedia[^]. Dave told us that the 50th number had length:

L50 = 894810

And the wikipedia article said:

L_n+1/L_n= lambda = 1.303577269034

so....

L50*lambda^(50)= 511175198256

if my math is right enough. Very hard programming challange :D

[Dave Kreskowiak]

Oh, it's possible. My machine is sitting here listing the iteration, length, and time to calculate for each of the 100 numbers.

System.ItDidntWorkException: Something didn't work as expected. C# - How to debug code[^]. Seriously, go read these articles.
Dave Kreskowiak