I can turn my dictionary Dictionary<long,MyObj> to a list with .Values, where MyObj has the Vector3 field named position.
I would like to have a list of the positions (the value type): MyObj.position, like so: List<Vector3> positions.
Performance is critical.
Is there a way to achieve this in C# with LINQ or otherwise other than?
foreach e in myDict
mylist.Add(e.Value.position)
So, using the code below:
public static void Main(string[] args)
{
var stopwatch1 = new Stopwatch();
var dictionaryTest = GetDictionary(1000);
stopwatch1.Start();
var results = dictionaryTest.Values.Select(x => x.Position).ToList();
stopwatch1.Stop();
var stopwatch2 = new Stopwatch();
stopwatch2.Start();
var results2 = dictionaryTest.Select(obj => obj.Value.Position).ToList();
stopwatch2.Stop();
var stopwatch3 = new Stopwatch();
stopwatch3.Start();
var myList = new List<double>();
foreach (var pair in dictionaryTest)
{
myList.Add(pair.Value.Position);
}
stopwatch3.Stop();
Console.WriteLine("results1: " + stopwatch1.Elapsed);
Console.WriteLine("results2: " + stopwatch2.Elapsed);
Console.WriteLine("results3: " + stopwatch3.Elapsed);
Console.Read();
}
public static Dictionary<long, MyUser> GetDictionary(int numberOfRows)
{
var d = new Dictionary<long, MyUser>();
for (int i = 0; i < numberOfRows; i++)
{
d.Add(1000 + i, new MyUser { Age = 10 + i, Position = 100.01 + i });
}
return d;
}
This brought back:
So foreach looks like the quickest of the 3 proposed on my machine. Worth testing yourself though.
With 'optimize code' selected:
Related
I've been doing some reading on the generic Dictionary class and the general advice is to use Dictionary if you need really fast access to an item matching a specific key. This is because a dictionary is using a type-safe Hashtable under the hood. When accessing items the search complexity is O(1) in dictionaries whereas in a List we would need to loop through EVERY SINGLE item until we find a match making the complexity O(n).
I wrote a little console app to see just how significant the difference between the two would be. The app stores 10 million items in each collection and attempts to access the second last item. The time difference between the List and Dictionary<TKey,TValue> is only one second, making the dictionary a winner but only just.
Question - can you provide an example(verbal is fine) where using a Dictionary vs a List would yield significant performance improvements?
class Program
{
static void Main(string[] args)
{
var iterations = 10000000;//10 million
var sw = new Stopwatch();
sw.Start();
var value1 = GetSecondLastFromDictionary(iterations);
sw.Stop();
var t1 = sw.Elapsed.ToString();
sw.Restart();
var value2 = GetSecondLastFromList(iterations);
sw.Stop();
var t2 = sw.Elapsed.ToString();
Console.WriteLine($"Dictionary - {t1}\nList - {t2}");
Console.ReadKey();
}
private static string GetSecondLastFromList(int iterations)
{
var collection = new List<Test>();
for (var i = 0; i < iterations; i++)
collection.Add(new Test { Key = i, Value = $"#{i}" });
return collection.Where(e => e.Key == iterations - 1).First().Value;
}
private static string GetSecondLastFromDictionary(int iterations)
{
var collection = new Dictionary<int, string>();
for (var i = 0; i < iterations; i++)
collection.Add(i, $"#{i}");
return collection[iterations - 1];
}
}
class Test
{
public int Key { get; set; }
public string Value { get; set; }
}
Your own example is fine to show where using a Dictionary yields significant performance improvements. The problem is you're not looking at the right thing. Your code spends a lot of time creating the dictionary or list and then does just one access of it. You need to separate out the collection creation and time multiple accesses of the item.
The code below does this. I get multiple accesses of the dictionary take 0.001s, whereas of the list the same number of accesses takes 2 minutes 32 seconds. Assuming I've done that right I think it shows dictionaries are faster for access.
static void Main(string[] args)
{
var iterations = 100000;
var sw = new Stopwatch();
var dict = CreateDict(iterations);
var list = CreateList(iterations);
sw.Start();
GetSecondLastFromDictionary(iterations, dict);
sw.Stop();
var t1 = sw.Elapsed.ToString();
sw.Restart();
GetSecondLastFromList(iterations, list);
sw.Stop();
var t2 = sw.Elapsed.ToString();
Console.WriteLine($"Dictionary - {t1}\nList - {t2}");
Console.ReadKey();
}
private static Dictionary<int, string> CreateDict(int iterations)
{
var collection = new Dictionary<int, string>();
for (var i = 0; i < iterations; i++)
collection.Add(i, $"#{i}");
return collection;
}
private static List<Test> CreateList(int iterations)
{
var collection = new List<Test>();
for (var i = 0; i < iterations; i++)
collection.Add(new Test { Key = i, Value = $"#{i}" });
return collection;
}
private static void GetSecondLastFromList(int iterations, List<Test> collection)
{
string test;
for (var i = 0; i < iterations; i++)
test = collection.Where(e => e.Key == iterations - 1).First().Value;
}
private static void GetSecondLastFromDictionary(int iterations, Dictionary<int, string> collection)
{
string test;
for (var i = 0; i < iterations; i++)
test = collection[iterations - 1];
}
}
My function is supposed to split a string by "&" ";" and "," and return a triple jagged array and vice versa. (From data like: 1,2,3;4,5,6&1,2,3;4,5,6)
I've been struggling hard to make it work, now for some reason I'm getting a system.argumentnullexception on
Array.Copy(playerOneEnts, allEnts[0], playerOneEnts.Length);
Array.Copy(playerTwoEnts, allEnts[1], playerTwoEnts.Length);
Full code:
public string convertToString(string[][][] allEnts)
{
string Player = string.Empty;
string[][] playerOneEnts = new string[maxEnts][];
string[][] playerTwoEnts = new string[maxEnts][];
Array.Copy(allEnts[0], playerOneEnts, allEnts[0].Length);
Array.Copy(allEnts[1], playerTwoEnts, allEnts[1].Length);
for (int j = 0; j < playerOneEnts.Length; j++)
{
for (int i = 0; i < playerOneEnts[j].Length; i++)
{
Player += playerOneEnts[j][i] + ",";
}
Player = Player.TrimEnd(',');
Player += ";";
}
Player = Player.TrimEnd(';');
Player += "&";
for (int j = 0; j < playerTwoEnts.Length; j++)
{
for (int i = 0; i < playerTwoEnts[j].Length; i++)
{
Player += playerTwoEnts[j][i] + ",";
}
Player = Player.TrimEnd(',');
Player += ";";
}
Player = Player.TrimEnd(';');
return Player;
}
public string[][][] convertToArray(string ents)
{
string[] p = new string[2];
string[][] playerOneEnts = new string[maxEnts][];
string[][] playerTwoEnts = new string[maxEnts][];
string[][][] allEnts = new string[2][][];
p = ents.Split('&');
try
{
playerOneEnts = p[0].Split(new[] { ";" }, StringSplitOptions.RemoveEmptyEntries).Select(chunk => chunk.Split(',')).ToArray();
playerTwoEnts = p[1].Split(new[] { ";" }, StringSplitOptions.RemoveEmptyEntries).Select(chunk => chunk.Split(',')).ToArray();
MessageBox.Show(playerOneEnts.Length.ToString());
Array.Copy(playerOneEnts, allEnts[0], playerOneEnts.Length);
Array.Copy(playerTwoEnts, allEnts[1], playerTwoEnts.Length);
}
catch
{
MessageBox.Show("unable to convert string", "Fatal Error");
}
return allEnts;
}
The code looks like a disaster to me, if anyone knows a nicer way to convert this to a string I'd be happy for any ideas. Getting rid of my Error would help me enough already.
Thanks!
Here is example using Linq (requires .Net 4.0):
var source = "1,2,3;4,5,6&1,2,3;4,5,6";
string[][][] decoded = source.Split('&').Select(x => x.Split(';').Select(y => y.Split(',').ToArray()).ToArray()).ToArray();
string reencoded = String.Join("&", decoded.Select(x => String.Join(";", x.Select(y => String.Join(",", y)))));
Alternatively IEnumarable generics can be used to avoid conversion to arrays.
Warning: This code doesn't validates input e.g. array lengths.
Edit: Re-encoder for .NET 3.5:
var reencoded = String.Join("&", decoded.Select(x => String.Join(";", x.Select(y => String.Join(",", y)).ToArray())).ToArray());
You can do it using nested for-loops like this:
string p = "1,2,3;4,5,6&1,5,3;4,5,9";
List<List<List<string>>> result = new List<List<List<string>>>();
foreach (var a in p.Split('&'))
{
List<List<string>> level2 = new List<List<string>>();
foreach (var b in a.Split(';'))
{
level2.Add(new List<string>(b.Split(',')));
}
result.Add(level2);
}
var x = result[0][1][2]; // This will result in '6'.
The encoding is similar:
string encoded;
List<string> dec1 = new List<string>();
foreach (var a in result)
{
string e = "";
List<string> z = new List<string>();
foreach (var b in a)
{
z.Add(String.Join(",", b));
}
e = String.Join(";", z);
dec1.Add(e);
}
encoded = String.Join("&", dec1);
Heres a fun problem I have.
I have a function that returns a var of items;
var Items = new { sumList = SumList, ratesList = List, sum = List.Sum() };
return Items;
From a function that is dynamic:
public override dynamic GetRates()
and I return it to a function I else where and try to apply it to my code:
dynamic res = cl.mainC.GetRates();
List<double> MashkantaSumList = res.sumList;
Now when I try to apply it, it says the object doesnt exist. But if I look in the debugger the items are happily there as a generic list or what not.
How do I resolve this?
EDIT:
as per request I'll post the full code:
//virtual
public virtual dynamic TotalMashkanta(int i, double sum, double ribit, string[] discount)
{
return 0;
}
//override
public override dynamic TotalMashkanta(int i, double sum, double ribit, string[] discount)
{
double SumTemp = sum;
double monthlyRibit = ribit / 12;
Double permPayPerMont = Financial.Pmt(monthlyRibit, i, sum, 0, DueDate.EndOfPeriod);
List<double> MashkantaList = new List<double>();
List<double> MashkantaSumList = new List<double>();
for (int j = 1; j <= i; j++)
{
MashkantaList.Add(Mashkanta(j, sum, ribit, permPayPerMont) * (1 - CalcDiscount((j / 12) + 1, discount)));
SumTemp = getSum(j, sum, ribit, permPayPerMont * -1); ;
MashkantaSumList.Add(SumTemp);
}
var K_Mashkanta = new { sumList = MashkantaSumList, ratesList = MashkantaList, sum = MashkantaList.Sum() };
return K_Mashkanta;
}
//Function that calls the results
public void GetSilukinTable(string Path, string ClientID, DAL.Client client, string partner_checked, string insurance_Amount, string Premiya_Structure_Mashkanta, string Premiya_Life_Mashkanta, string Discount_Life_Mashkanta, string Loan_Period,string Loan_EndDate, string Bank, string Loan_Interest, string Loan_Amount, string Discount_Loan, string AgentNotes, string ManID)
{
BL.CalculateLogic.Companies t = BL.CalculateLogic.Companies.כלל;
if(ManID == "211") t = BL.CalculateLogic.Companies.הפניקס;
if(ManID == "207") t = BL.CalculateLogic.Companies.הראל;
if(ManID == "206") t = BL.CalculateLogic.Companies.מנורה;
if(ManID == "208") t = BL.CalculateLogic.Companies.הכשרה;
BL.CalculateLogic cl = new BL.CalculateLogic(client, t);
DateTime LoanEnd = DateTime.Now;
int months = 0;
if (DateTime.TryParse(Loan_EndDate, out LoanEnd))
months = BL.Calculating_Companies.Company.GetMonthsBetween(DateTime.Now, LoanEnd);
else
months = Int32.Parse(Loan_Period) * 12;
string[] Discount = Discount_Loan.Split('-');
dynamic res = cl.mainC.TotalMashkanta(months, Double.Parse(Loan_Amount), Double.Parse(Loan_Interest.Trim('%')), Discount);
var MashkantaSumList = res.sumList;
List<double> MashkantaList = res.ratesList;
List<double> MashkantaSumListPartner = new List<double>();
List<double> MashkantaListPartner = new List<double>();
List<double> MashkantaListSum = res.ratesList;
}
The compiler is happy about it because dynamic is compiled and checked at run time. Whatever the problem is, the types don't match. It evaluates this at run time, so you won't see issues at compile time. (Advice: use dynamic only when you really must! Else you will have this kind of problems all the time!)
I tried your code using this and it works fine:
static dynamic GetRates()
{
List<double> SumList = new List<double>();
List<double> List = new List<double>();
var Items = new { sumList = SumList, ratesList = List, sum = List.Sum() };
return Items;
}
static void Main(string[] args)
{
dynamic res = GetRates();
List<double> MashkantaSumList = res.sumList;
}
List A:
1, 2, 3, 4
List B:
2, 5
How to check if list A contains any value from list B?
e.g. something like A.contains(a=>a.id = B.id)?
If you didn't care about performance, you could try:
a.Any(item => b.Contains(item))
// or, as in the column using a method group
a.Any(b.Contains)
But I would try this first:
a.Intersect(b).Any()
I've profiled Justins two solutions. a.Any(a => b.Contains(a)) is fastest.
using System;
using System.Collections.Generic;
using System.Linq;
namespace AnswersOnSO
{
public class Class1
{
public static void Main(string []args)
{
// How to check if list A contains any value from list B?
// e.g. something like A.contains(a=>a.id = B.id)?
var a = new List<int> {1,2,3,4};
var b = new List<int> {2,5};
var times = 10000000;
DateTime dtAny = DateTime.Now;
for (var i = 0; i < times; i++)
{
var aContainsBElements = a.Any(b.Contains);
}
var timeAny = (DateTime.Now - dtAny).TotalSeconds;
DateTime dtIntersect = DateTime.Now;
for (var i = 0; i < times; i++)
{
var aContainsBElements = a.Intersect(b).Any();
}
var timeIntersect = (DateTime.Now - dtIntersect).TotalSeconds;
// timeAny: 1.1470656 secs
// timeIn.: 3.1431798 secs
}
}
}
You can Intersect the two lists:
if (A.Intersect(B).Any())
You can check if a list is inside of another list with this
var list1 = new List<int> { 1, 2, 3, 4, 6 };
var list2 = new List<int> { 2, 3 };
bool a = list1.Any(c => list2.Contains(c));
For faster and short solution you can use HashSet instead of List.
a.Overlaps(b);
Overlaps documentation
This method is an O(n) instead of O(n^2) with two lists.
I write a faster method for it can make the small one to set. But I test it in some data that some time it's faster that Intersect but some time Intersect fast that my code.
public static bool Contain<T>(List<T> a, List<T> b)
{
if (a.Count <= 10 && b.Count <= 10)
{
return a.Any(b.Contains);
}
if (a.Count > b.Count)
{
return Contain((IEnumerable<T>) b, (IEnumerable<T>) a);
}
return Contain((IEnumerable<T>) a, (IEnumerable<T>) b);
}
public static bool Contain<T>(IEnumerable<T> a, IEnumerable<T> b)
{
HashSet<T> j = new HashSet<T>(a);
return b.Any(j.Contains);
}
The Intersect calls Set that have not check the second size and this is the Intersect's code.
Set<TSource> set = new Set<TSource>(comparer);
foreach (TSource element in second) set.Add(element);
foreach (TSource element in first)
if (set.Remove(element)) yield return element;
The difference in two methods is my method use HashSet and check the count and Intersect use set that is faster than HashSet. We dont warry its performance.
The test :
static void Main(string[] args)
{
var a = Enumerable.Range(0, 100000);
var b = Enumerable.Range(10000000, 1000);
var t = new Stopwatch();
t.Start();
Repeat(()=> { Contain(a, b); });
t.Stop();
Console.WriteLine(t.ElapsedMilliseconds);//490ms
var a1 = Enumerable.Range(0, 100000).ToList();
var a2 = b.ToList();
t.Restart();
Repeat(()=> { Contain(a1, a2); });
t.Stop();
Console.WriteLine(t.ElapsedMilliseconds);//203ms
t.Restart();
Repeat(()=>{ a.Intersect(b).Any(); });
t.Stop();
Console.WriteLine(t.ElapsedMilliseconds);//190ms
t.Restart();
Repeat(()=>{ b.Intersect(a).Any(); });
t.Stop();
Console.WriteLine(t.ElapsedMilliseconds);//497ms
t.Restart();
a.Any(b.Contains);
t.Stop();
Console.WriteLine(t.ElapsedMilliseconds);//600ms
}
private static void Repeat(Action a)
{
for (int i = 0; i < 100; i++)
{
a();
}
}
Sorry, if this is irelevant, but will return list with matches using FindAll() in case you need this:
private bool IsContain(string cont)
{
List<string> ListToMatch= new List<string>() {"string1","string2"};
if (ListToMatch.ToArray().Any(cont.Contains))
{
return false;
}
else
return true;
}
And usage:
List<string> ListToCheck = new List<string>() {"string1","string2","string3","string4"};
List<string> FinalList = ListToCheck.FindAll(IsContain);
The final list contains only the matched elements string1 and string2 from list to check.
Can easy be switched to int List.
I use this to count:
int cnt = 0;
foreach (var lA in listA)
{
if (listB.Contains(lA))
{
cnt++;
}
}
I have run some tests for .Net CF. Basically, I wanted to compare for, foreach, extenstion method ForEach and LINQ query. Here is the whole code (you can skip it, to get to the point which bothers me)
namespace ForEachForLINQPerTest
{
class IntBox
{
public int fieldX;
public int PropertyX { get; set; }
}
public partial class MainPage : PhoneApplicationPage
{
/// <summary>
/// size of tested List
/// </summary>
public const int TEST_SIZE = 1000000;
//
private List<int> m_intList = new List<int>(TEST_SIZE);
//
private List<IntBox> m_intBoxList = new List<IntBox>(TEST_SIZE);
//
private Stopwatch m_stopwatch = null;
// Constructor
public MainPage()
{
InitializeComponent();
for (int i = 0; i < TEST_SIZE; ++i)
{
m_intBoxList.Add( new IntBox());
m_intList.Add(0);
}
}
private void startButton_Click(object sender, RoutedEventArgs e)
{
var forTest = ForTest(); // Jitter preheat
forTest = ForTest();
forResultTextBlock.Text = forTest;
var foreachTest = ForEachTest();
foreachTest = ForEachTest();
foreachResultTextBlock.Text = foreachTest;
var exTest = Extenstion();
exTest = Extenstion();
ExtensionResultTextBlock.Text = exTest;
var linqTest = LINQTest();
linqTest = LINQTest();
LINQResultTextBlock.Text = linqTest;
}
private string LINQTest()
{
m_stopwatch = new Stopwatch();
m_stopwatch.Start();
long temp = 0;
var result = from x in m_intList
select temp += x;
m_stopwatch.Stop();
var intListTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
result.ToList();
m_stopwatch.Start();
var result2 = from x in m_intBoxList
select temp += x.fieldX;
m_stopwatch.Stop();
var intBoxListFieldTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
result2.ToList();
m_stopwatch.Start();
var result3 = from x in m_intBoxList
select temp += x.PropertyX;
m_stopwatch.Stop();
var intBoxListPropertyTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
result3.ToList();
return String.Format("LINQ test List<int> = {0} \n List<IntBox> field = {1} \n List<IntBos> property = {2}", intListTime, intBoxListFieldTime, intBoxListPropertyTime);
}
private string Extenstion()
{
m_stopwatch = new Stopwatch();
m_stopwatch.Start();
long temp = 0;
m_intList.ForEach(i => temp += i);
m_stopwatch.Stop();
var intListTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
m_intBoxList.ForEach(i => temp += i.fieldX);
m_stopwatch.Stop();
var intBoxListFieldTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
m_intBoxList.ForEach(i => temp += i.PropertyX);
m_stopwatch.Stop();
var intBoxListPropertyTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
return String.Format("Extenstion test List<int> = {0} \n List<IntBox> field = {1} \n List<IntBos> property = {2}", intListTime, intBoxListFieldTime, intBoxListPropertyTime);
}
private string ForEachTest()
{
m_stopwatch = new Stopwatch();
long temp = 0;
m_stopwatch.Start();
foreach(int item in m_intList)
{
temp += item;
}
m_stopwatch.Stop();
var intListTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
foreach (IntBox item in m_intBoxList)
{
temp += item.fieldX;
}
m_stopwatch.Stop();
var intBoxListFieldTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
foreach (IntBox item in m_intBoxList)
{
temp += item.PropertyX;
}
m_stopwatch.Stop();
var intBoxListPropertyTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
return String.Format("ForEach test List<int> = {0} \n List<IntBox> field = {1} \n List<IntBos> property = {2}", intListTime, intBoxListFieldTime, intBoxListPropertyTime);
}
private string ForTest()
{
m_stopwatch = new Stopwatch();
m_stopwatch.Start();
long temp = 0;
for (int i = 0; i < TEST_SIZE; ++i)
{
temp += m_intList[i];
}
m_stopwatch.Stop();
var intListTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
for (int i = 0; i < m_intList.Count; ++i)
{
temp += m_intBoxList[i].fieldX;
}
m_stopwatch.Stop();
var intBoxListFieldTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
m_stopwatch.Start();
for (int i = 0; i < m_intList.Count; ++i)
{
temp += m_intBoxList[i].PropertyX;
}
m_stopwatch.Stop();
var intBoxListPropertyTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
return String.Format("For loop test List<int> = {0} \n List<IntBox> field = {1} \n List<IntBos> property = {2}", intListTime, intBoxListFieldTime, intBoxListPropertyTime);
}
}
}
And here I am confused
m_stopwatch = new Stopwatch();
m_stopwatch.Start();
long temp = 0;
var result = from x in m_intList
select temp += x;
m_stopwatch.Stop();
var intListTime = m_stopwatch.ElapsedMilliseconds;
m_stopwatch.Reset();
result.ToList();
The ouput is:
For loop test List = 93
List field = 119 // ref -> field
List property = 136 // ref -> property -> field properties are just functions for CF
ForEach test List = 88
List field = 140
List property = 152
Extenstions test List = 176 // another function is called.
List field = 220
List property = 239
LINQ test List = 0 Why?
List field = 163
List property = 165
Why intListTime == 0? What am I doing wrong? Also the last two values for field and property are almost the same (run it a few times). Does it mean that PropertyX in LINQ query is evaluated in-line?
The first time is zero because expression tree is built at compile time and it gets evaluated on ToList call that you have not included in timing.
For field and property access timing, I wouldn't worry too much - in reality, in release build, simple property accessor will be get inlined giving same performance as field access. For linq case, you might be seeing the same performance because linq internally might be converting property/field access into a method call and it would result in same timings (as I believe that method call overhead will be probably large compared to the field/prop access.
this is called "deferred execution". the linq statement isn't evaluated until it needs to be. move the ToList to be before you stop the clock and the time will go up