@ everyone saying this guy is amazing, and wishing their teacher was like this, this guys is insanely good, he started university when he was 12, got his bachelors degree when he was 14, and completed his PhD thesis when he was 20. He's the youngest every teacher at MIT
Example= MIA(2*2)------>4, M-77 I-105 A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
This guy is a genius. He founded his first software company when he was 8 with his dad. At 10 he graduated high school (not validictorian) and went on to university. At 13 he graduated university and sold his company for 1.2 million dollars. He went on to win a Politzer prize and patent for a computerized fabrics machine. MIT asked him to start teaching at the age of 16 and here we are.
Awesome lecture! I can't imagine the number of books he have read and the skills he have mastered to be this good. So much respect sir. You are worth emulating!
Example= MIA(2*2)------>4, M-77 I-105 A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
I can read here many people are complaining that if they had teacher like this lecturer,they would be happy. But the matter of fact is they all have him. Eric is teaching that's what teacher does , it's your job to make as much use of it and donate for such causes once you succeed in life.
my professor just gave us about 6 simple very dry PP slides, gave us starter code with data structure and gave us a pop quiz. I wish this guys was my professor.
So many cool things going on beyond the lecture content. First, the movable boards. That is so cool. Never saw anything like this before in my 53 years. Second, Erik's writing style on the board is very clear. Headings underlined, numbers circled, etc. It's all very on-purpose and makes me wonder if that comes from formal training or if Erik devised it on-the-fly. The content is great too. He explains the concepts with the ease of someone with deep understanding. Loving this series.
He's right! In Python, the id of an object is equal to the hash of the id. d = { 'a': 1, 'b': 2, 'c': 3 } for key,value in d.items(): print(key) print(id(key)) print(hash(id(key))) Output: a 5045920 5045920 b 5044240 5044240 c 4964624 4964624
@@patmaloyan620 yeah yeah. it's awesome to listen the courses at 1.5 or 2.0 speeds and it's increasing the focus. absolutely great. and we're finishing it in less time too. tons of time saved, thanks for the youtube videoplayer.
MIT is so freaking lucky. At UPenn and Temple, I had to write notes like a madman because no lectures were recorded. I wasted sooo much time writing notes and trying to write fast enough that the lecture was pretty much pointless, and resulted in me basically reviewing notes and asking questions at a later time.
Gary C I never took notes ever and graduated top of my class; it's best just to pay attention to the lecture in class and then use a textbook for anything concrete.
hash("\b0")=hash("\0\0C")=64 means that two different keys map to the same thing, 64. ideally you want if hash(x)=hash(y) then x=y that means the keys are the same. but sometimes we can have a collision where x!=y.
If teaching in 2022 was like this I would really be mf Genius in Data structure and IT, unfortunately they only stream videos and you MUST UNDERSTAND IT ALONE
@siddarth kamaria There is a Collection of hash functions you can use. 'a' and 'b' are random chosen in the beginning. Since then you must use the same hash function.
Example= MIA(2*2)------>4, M-77 I-105 A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
@MIT Opencourseware While explaining simple uniform hashing, Erik suggested that independence is necessary but is independence really necessary? The expected length of a chain comes out the same whether you assume independence or not. My solution without assuming independence: let R1 be indicator random variable for the event that a random key falls into the 1st bucket. R2 be the independent random variable for the event that a random key falls into the 2nd bucket and so on. Initially assume all slots or buckets are empty. A chain length becomes length one when a random key falls into one of the slots and remains the same when it falls into any other bucket. So, l1 be the length of chain in the first slot; l2 be the length of chain in the second slot and so on. So, total length of chain would be l= l1+l2+l3+...+lm. So, expected length of chain E(l)= E(l1+l2+l3+...+lm)= E(l1)+ E(l2)+ E(l3)+...+E(m).... (by linearity of expectation which does not need independence). E(l1)= 1.1/m+ 0.(1-1/m)= 1/m = E(l2)= E(l3)=...=E(lm) So, E(l) would be n/m. I did not assume independence. Where did I go wrong? Someone please explain.
They are actually seat cushions.... Srini explains that the seats in the hall are uncomfortable to sit on for long durations. So a cushion is a perk one earns for answering questions. I wouldn't be surprised if there a cult angle to the scenario though.
Pillows?! Wow that's peculiar. I almost thought that was some sort of secret invitation to a underground brainiac party held by him (he has that cult leader quality).
Yeah I've never heard of a prehash and hash the way he describes it. I know it as: the hash function turns an object into an integer. And the compression function gives you some integer within the bounds of the underlying array. A google search doesn't give that many results for pre-hash or prehash so... my guess is the are less popular terms (?)
From my understanding, prehash merely turns your non-int keys into integers but does not address the problem of space, so you hash the resultant integer using a more complex hash function which should theoretically convert the integer into another integer within a specific set of integers with m as the upper bound
The "prehash" is a function that converts the state of the key object to an integer; the "hash" function converts it to a hash table index. This is a logical division of labor; the key type "knows" how to convert the object to an integer, and the hash table "knows" how to convert that integer to a hash table index. But neither of them has to know anything about the other. What he's calling the prehash is done in Java, for example, by overriding the Object.hashCode() method; in C++, it is generally done by define a hash class and implementing operator(). The hash function then generally uses either the division method or the multiplication method, as described in the video. I personally prefer the multiplication method since it works with hash table sizes that are powers of 2, and that allows you to use the table resizing discussed in the next lecture in this series.
the issue with collision, if the developers had just taken the string of bytes, appended itself, for example /0B and /0/0C, turn that into /0B + B0/, and /0/0CC0/0/ so that you have palindromes, the collision should occur less if at all, instead of using the raw string of hash.
In the beginning it was said that the key is the index in the array, so if you have negative integers as keys, you have to have negative indexes. Later on, it is said, that you have some kind of prehashing, that gets rid of most assumptions.
For that useless piece of paper that says doing some shitty handwork by some professor is a better way to be recognized on the job market than learning by your self .
+Gherbi Hicham Employer doesn't have time to test every job applicant so he/she wants a proof of knowledge issued by institutions which he/she trusts same way you want to use medicine approved by official institution for testing medicine.
ionezgb I didn't say every job , sure i won't go to someone without a medicine degree if i felt pain , some jobs just need to be certified by an institution , some can only be proven by institution , again how on earth are you going to prove that you can prescribe drugs without a university degree ? that's not my point , my point is that a lot of jobs don't need a university degree at all and its good respectable well payed jobs as well, being a programmer for example can have multiple other proofs of knowledge other than a University degree , good projects on Github , a good Stackoverflow reputation count , even down right list of the projects you've built and several other things can land you a job as a developer in a lot of places, infact some jobs will even require things like your programming profile. This misconception that you need college in order to work a proper or a well paying job is just getting ridiculous, a lot of people don't have the time nor the financial resources to engage in a university program, and a lot of people who enrole in college for a useless liberal arts degree only find themselves several years later with no job and a massive loan , University shouldn't be for every one , and not everything needs a degree . This is coming from someone who spent 5 years at college for a masters in Computer Science .
Does anyone feel why the lecture is bounded by time, I just he feel he explains whatever knowledge he has in how much ever time he needs for it. Awesome lecture @ErikDemaine
If a and b are random, does that make the hash function non-deterministic? Do we have to cache the calculated hashes in order to look items up in a hash table after they've been inserted?
At 33:15 Erik draws a linked list to store multiple items where their key collides. How does one retrieve the correct item on a lookup? The keys themselves seem to be absent from the linked list, so how do you know which item to return?
I think the assumption is that the item is a tuple with the key and corresponding value. As in the Python implementation explained earlier in the video.
The key is immutable for each object. Otherwise you couldn't reconstruct it later on. So either you calculate the key for each object in the list anew or you save tuples.
I understand how to put stuff in the hashtable, what I don't understand is how to look it back up. If your key is random, how do you know which key to look up?
Outdated information... Just use a large hash table so there is a high probability of 0 collisions. Space is cheap now. Having a hash table that is the same size as the number of keys inserted into it is not necessary. Who cares if the hash table is 10x the size of the # of keys? Memory is so cheap. Wasted space many times with computers = more speed. When i was in college, one good application I found for a 0 collision hash table was to determine in a programming language, if a word in the input file was a keyword (such as if, then, while...). By having a separate lookup table for those, the compiler could quickly determine if it is a keyword or not simply by hashing and then comparing letter for letter and stopping at the first mismatch. Since keywords are very frequent and the # of them is rather small (compared to the possible # of different non keywords), it makes some sense to put all the keywords (maybe 50 or so of them), in a separate table. Some people may argue that a single symbol table is sufficient or even better and they may be right. Computers are so fast these days I doubt it would make much difference anyway.
Memory is cheap depending on your system. If you’re designing a solution to run on embedded hardware, you may not have access to the same amount of memory. You might be sharing with other key processes.
How can 'a' and 'b' be random? If we call the hash function another time it might generate another values for 'a' and 'b'. I'm unclear what is he referring to by describing 'a' and 'b' as random...Any help would be appreciated.
Example= MIA(2*2)------>4, M-77 I-105 A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
Thank you for the video. I have question. Why don't we instead of using hash table and liked list together, use two hash tables and link them with a unique queue.
Can someone explain what the point of this is? You mapped your universe to a small array but each index of the array contains a linked list and the entire universe fits into the array so the array has to use a lot of memory so I don't see how this solves problem (2) gigantic memory
00:55 dictionary
05:10 dict in Python
06:30 motivation
13:40 simple approach
17:50 prehash
24:00 hashing
30:15 chaining
42:05 hash functions
@ everyone saying this guy is amazing, and wishing their teacher was like this, this guys is insanely good, he started university when he was 12, got his bachelors degree when he was 14, and completed his PhD thesis when he was 20. He's the youngest every teacher at MIT
The real goat 🐐
Example= MIA(2*2)------>4,
M-77
I-105
A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
@@Utuber5000 okaaaayyyyy
and he looks like he's 10! =)
what his name ??
The core strength this man displays with that chalk-handing is god-like.
Especially at 1.5 speed lol
@@Xaminn 😜
My Algorithms teacher is shit. She reads off her power point VERBATIM!! I want my money back because this guy has taught me more than she ever could.
Dude same...
If anyone is curious, hash('\0B') does NOT equal hash('\0\0C') (or '\0\0c') anymore in python (ipython 7.13)
This guy is a genius. He founded his first software company when he was 8 with his dad. At 10 he graduated high school (not validictorian) and went on to university. At 13 he graduated university and sold his company for 1.2 million dollars. He went on to win a Politzer prize and patent for a computerized fabrics machine. MIT asked him to start teaching at the age of 16 and here we are.
😵🤯
is this satire
Awesome lecture! I can't imagine the number of books he have read and the skills he have mastered to be this good. So much respect sir. You are worth emulating!
This teacher is awesome
+Snake agree
He had bachelors degree at age 14 and phd at 20
@NameHe was 30 or 31, born in 1981. But who cares? The lecture is awesome all the same.
Example= MIA(2*2)------>4,
M-77
I-105
A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
I'm jealous of those MIT students, so lucky to have such a great prof.
Yet, we are all here 😎. Good for you for learning on your own
I can read here many people are complaining that if they had teacher like this lecturer,they would be happy.
But the matter of fact is they all have him. Eric is teaching that's what teacher does , it's your job to make as much use of it and donate for such causes once you succeed in life.
God I wish I had teachers like that.
Now you have it. How is your algorithms?
@@weizhixie9678 XD
my professor just gave us about 6 simple very dry PP slides, gave us starter code with data structure and gave us a pop quiz. I wish this guys was my professor.
So many cool things going on beyond the lecture content. First, the movable boards. That is so cool. Never saw anything like this before in my 53 years. Second, Erik's writing style on the board is very clear. Headings underlined, numbers circled, etc. It's all very on-purpose and makes me wonder if that comes from formal training or if Erik devised it on-the-fly. The content is great too. He explains the concepts with the ease of someone with deep understanding. Loving this series.
Thank you, Mr. Demaine, for rocking the chalk like you rock the ponytail: with gusto and aplomb.
He starts talking about at chaining at 30:15.
I thought he just skipped the solution to the problem 1 of map to non-negative integer. which relate to hash functions.
He's right! In Python, the id of an object is equal to the hash of the id.
d = { 'a': 1, 'b': 2, 'c': 3 }
for key,value in d.items():
print(key)
print(id(key))
print(hash(id(key)))
Output:
a
5045920
5045920
b
5044240
5044240
c
4964624
4964624
Actual hashing is only addressed at 24:00
Coolashacka thanks a lot for that....
Thanks mate
@JustOne this is literally their actual course
Thanka
This guy is even more awesome at 1.5 speed
Also at 2.0
@@icosmini Agree
hahahah. I keep forgetting this and these comments keep saving me. 1000% focus increase.
@@patmaloyan620 yeah yeah. it's awesome to listen the courses at 1.5 or 2.0 speeds and it's increasing the focus. absolutely great. and we're finishing it in less time too. tons of time saved, thanks for the youtube videoplayer.
40:30 how could they not laugh at that???? It's gold.
+Luis Daniel Mesa Velasquez can u explain the star wars reference?
+and1fer Star Wars, A New Hope, Ben kenobi to the stormtrooper: "These aren't the Droids your looking for"
+Luis Daniel Mesa Velasquez probably because they never watched it :(
spider :´(
MIT is so freaking lucky. At UPenn and Temple, I had to write notes like a madman because no lectures were recorded. I wasted sooo much time writing notes and trying to write fast enough that the lecture was pretty much pointless, and resulted in me basically reviewing notes and asking questions at a later time.
An aside I notice when I was like WTF why is no one writing notes
Gary C I never took notes ever and graduated top of my class; it's best just to pay attention to the lecture in class and then use a textbook for anything concrete.
i would love to have a prof. like this :D i sleept in so many time in our class about hashing x.x but he get me the motivation!
greets from germany!
hash("\b0")=hash("\0\0C")=64 means that two different keys map to the same thing, 64. ideally you want if hash(x)=hash(y) then x=y that means the keys are the same. but sometimes we can have a collision where x!=y.
If teaching in 2022 was like this I would really be mf Genius in Data structure and IT, unfortunately they only stream videos and you MUST UNDERSTAND IT ALONE
This guy is awesome, even 9 years later, still better than my professors. i just wish I had a CS prof that could speak english :/
@siddarth kamaria There is a Collection of hash functions you can use. 'a' and 'b' are random chosen in the beginning. Since then you must use the same hash function.
You are an amazing teacher, Erick :-)
Btw "Hasheesh" in Arabic means "marijuana"!!
It's so cool that these top schools release courses like this one online free of charge. I may not get a chance to go to MIT
15:46 Why something is hard to associate with integers
24:08 hashish vs hashing
Example= MIA(2*2)------>4,
M-77
I-105
A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
@@Utuber5000 how's this related to my comment?
@@Utuber5000 yeah, it can be done. I did something similar to this some time ago
@MIT Opencourseware While explaining simple uniform hashing, Erik suggested that independence is necessary but is independence really necessary?
The expected length of a chain comes out the same whether you assume independence or not.
My solution without assuming independence:
let R1 be indicator random variable for the event that a random key falls into the 1st bucket. R2 be the independent random variable for the event that a random key falls into the 2nd bucket and so on. Initially assume all slots or buckets are empty. A chain length becomes length one when a random key falls into one of the slots and remains the same when it falls into any other bucket.
So, l1 be the length of chain in the first slot; l2 be the length of chain in the second slot and so on.
So, total length of chain would be l= l1+l2+l3+...+lm.
So, expected length of chain E(l)= E(l1+l2+l3+...+lm)= E(l1)+ E(l2)+ E(l3)+...+E(m).... (by linearity of expectation which does not need independence).
E(l1)= 1.1/m+ 0.(1-1/m)= 1/m = E(l2)= E(l3)=...=E(lm)
So, E(l) would be n/m. I did not assume independence.
Where did I go wrong? Someone please explain.
damn this videos are so damn cool, i wish i had an actually good teacher like him
Thank you teacher for just being there
great stuff !
now that's the typical nerd look you would get after becoming an algorithms teacher. love it.
They are actually seat cushions.... Srini explains that the seats in the hall are uncomfortable to sit on for long durations. So a cushion is a perk one earns for answering questions.
I wouldn't be surprised if there a cult angle to the scenario though.
Lolwut
I love how he explain the algorithem problems also he look young.
Thanks for your comments! A little trivia: Erik Demaine became MIT's youngest professor in Fall 2001: news.mit.edu/2003/demaine-0226.
Pillows?! Wow that's peculiar. I almost thought that was some sort of secret invitation to a underground brainiac party held by him (he has that cult leader quality).
I didn't even feel when those 50+ minutes passed!!
The lecture was awesome, but the starwars reference was even better!
Who else is Still watching this video in 2021??? Great lecture!
Dopest class, much better than my dear teachers, or almost dear... whetheaver, thanks!!!!
Wow MIT has masterpiece teachers!
42:14 hash function fundamentals.
You are awesome! Big Thanks
What is the difference between a prehash and a hash function? I didn't feel totally clear on that point after this lecture.
It's also known as hashing (prehash) and compression (hash), maybe that helps
if the key are not an intergers, you use the prehash to make then in a intergers and then use the hash fuction. I think.
Yeah I've never heard of a prehash and hash the way he describes it. I know it as: the hash function turns an object into an integer. And the compression function gives you some integer within the bounds of the underlying array.
A google search doesn't give that many results for pre-hash or prehash so... my guess is the are less popular terms (?)
From my understanding, prehash merely turns your non-int keys into integers but does not address the problem of space, so you hash the resultant integer using a more complex hash function which should theoretically convert the integer into another integer within a specific set of integers with m as the upper bound
The "prehash" is a function that converts the state of the key object to an integer; the "hash" function converts it to a hash table index.
This is a logical division of labor; the key type "knows" how to convert the object to an integer, and the hash table "knows" how to convert that integer to a hash table index. But neither of them has to know anything about the other.
What he's calling the prehash is done in Java, for example, by overriding the Object.hashCode() method; in C++, it is generally done by define a hash class and implementing operator().
The hash function then generally uses either the division method or the multiplication method, as described in the video. I personally prefer the multiplication method since it works with hash table sizes that are powers of 2, and that allows you to use the table resizing discussed in the next lecture in this series.
the issue with collision, if the developers had just taken the string of bytes, appended itself, for example /0B and /0/0C, turn that into /0B + B0/, and /0/0CC0/0/ so that you have palindromes, the collision should occur less if at all, instead of using the raw string of hash.
Why non-negative integers restriction is a problem? Why can't we just throw out the minus sign?
Collisions. Not that that is the end of the world, but at that point in the lecture we are assuming that there's a unique integer-key for every value.
Kudos, Damien
No problem!
In the beginning it was said that the key is the index in the array, so if you have negative integers as keys, you have to have negative indexes. Later on, it is said, that you have some kind of prehashing, that gets rid of most assumptions.
I love these videos!
Great teacher !
i dont even know why i go to class anymore..cept for midterms
For that useless piece of paper that says doing some shitty handwork by some professor is a better way to be recognized on the job market than learning by your self .
Gherbi Hicham You're very misguided.
stampsr92 Elaborate please ...
+Gherbi Hicham Employer doesn't have time to test every job applicant so he/she wants a proof of knowledge issued by institutions which he/she trusts same way you want to use medicine approved by official institution for testing medicine.
ionezgb I didn't say every job , sure i won't go to someone without a medicine degree if i felt pain , some jobs just need to be certified by an institution , some can only be proven by institution , again how on earth are you going to prove that you can prescribe drugs without a university degree ? that's not my point , my point is that a lot of jobs don't need a university degree at all and its good respectable well payed jobs as well, being a programmer for example can have multiple other proofs of knowledge other than a University degree , good projects on Github , a good Stackoverflow reputation count , even down right list of the projects you've built and several other things can land you a job as a developer in a lot of places, infact some jobs will even require things like your programming profile.
This misconception that you need college in order to work a proper or a well paying job is just getting ridiculous, a lot of people don't have the time nor the financial resources to engage in a university program, and a lot of people who enrole in college for a useless liberal arts degree only find themselves several years later with no job and a massive loan , University shouldn't be for every one , and not everything needs a degree .
This is coming from someone who spent 5 years at college for a masters in Computer Science .
That was an awesome lecture. Thanks MIT :)
amazing class. Thanks for making this public
Does anyone feel why the lecture is bounded by time, I just he feel he explains whatever knowledge he has in how much ever time he needs for it. Awesome lecture @ErikDemaine
a Very much satisfying place to learn cs
What is he throwing at 17:07 ?
Thanks for uploading this.
Did anyone else notice the blunt behind the guy’s ear at 1:48 😂😂😂
I died laughing at his references, Erik Demaine is such a cool geek
What was the vampire reference?
I like the way when he explains hash meaning hhh.. Thank you
Can someone tell what's difference between hash and prehash.
ak+b mod p and p is a huge number? ain't that just equal to ak+b?
If a and b are random, does that make the hash function non-deterministic? Do we have to cache the calculated hashes in order to look items up in a hash table after they've been inserted?
maybe a and b are chosen from very first and stored as constants
@Bede Kelly I was wondering the same thing. I think maybe yes.
Literally a good explanation I got❤
Big Thanks from Austria!! :D
professor is really an heavy job
Thanks teachers, lession is awesome.
Ok, I'm inspired. Time to go code.
why they don't talk about expression trees?!
These courses are awesome N helpfull
excellent course
why are they talking about pointers if there is no function of pointer in python can anyone answer that please
3:05 what's the point to override a key? nothing changes
I guess he meant to override the item not the key
writing with chalks and erasing with hands?
Excellent explanation. A++
Excellent lecturer. Bravo sir
Why does it suck to have the division method for hashing? I can't see why he metioned half of the table would be used if m is the power of 2.
I see it now.
Hashing is cool ~ Erik Demaine
why don't we use a pointer to another dictionary instead of a linked list?
awesome instructor
he is handing pillows - see the first lecture of the series, there is some explanation
can someone explain key space? why is memory allocated to key space a problem in direct access tables?
Wonderful lecture.
Thankful for this
Great lecture
great teaching
At 33:15 Erik draws a linked list to store multiple items where their key collides. How does one retrieve the correct item on a lookup? The keys themselves seem to be absent from the linked list, so how do you know which item to return?
I think the assumption is that the item is a tuple with the key and corresponding value. As in the Python implementation explained earlier in the video.
The key is immutable for each object. Otherwise you couldn't reconstruct it later on. So either you calculate the key for each object in the list anew or you save tuples.
I understand how to put stuff in the hashtable, what I don't understand is how to look it back up. If your key is random, how do you know which key to look up?
i love MIT! i love EriK Demaine!
Blackboards are the key
Outdated information... Just use a large hash table so there is a high probability of 0 collisions. Space is cheap now. Having a hash table that is the same size as the number of keys inserted into it is not necessary. Who cares if the hash table is 10x the size of the # of keys? Memory is so cheap. Wasted space many times with computers = more speed.
When i was in college, one good application I found for a 0 collision hash table was to determine in a programming language, if a word in the input file was a keyword (such as if, then, while...). By having a separate lookup table for those, the compiler could quickly determine if it is a keyword or not simply by hashing and then comparing letter for letter and stopping at the first mismatch. Since keywords are very frequent and the # of them is rather small (compared to the possible # of different non keywords), it makes some sense to put all the keywords (maybe 50 or so of them), in a separate table. Some people may argue that a single symbol table is sufficient or even better and they may be right. Computers are so fast these days I doubt it would make much difference anyway.
Memory is cheap depending on your system. If you’re designing a solution to run on embedded hardware, you may not have access to the same amount of memory. You might be sharing with other key processes.
How can 'a' and 'b' be random? If we call the hash function another time it might generate another values for 'a' and 'b'. I'm unclear what is he referring to by describing 'a' and 'b' as random...Any help would be appreciated.
Example= MIA(2*2)------>4,
M-77
I-105
A-97, Equation= sum of axes divided by array boxes and then 77+105+97 / divided by 11, which is the number of array boxes. 279/11=8 , I know afterwards I would find the next index number for the new key/name given , its just that I wrote this down about a year ago , but the part about how I found the axes 77,105,& 97 is missing , so can u renew my memory on how I go about finding the axes again?
Thank you for the video. I have question. Why don't we instead of using hash table and liked list together, use two hash tables and link them with a unique queue.
Its called double hashing and it will leave us with more unallocated space.
38:50 That's what she said
Why load factors means constant time? can someone explain?
load factor=O(1), if m=O(n), is this mean, if m is big enough to cover n, it's a constant time?
why you not use a projector?
Recap for this video in Chinese
medium.com/@ipalm0423/mit-6-006-introduction-to-algorithms-8-90ca9cbce927#.neihzsm23
Can someone explain what the point of this is? You mapped your universe to a small array but each index of the array contains a linked list and the entire universe fits into the array so the array has to use a lot of memory so I don't see how this solves problem (2) gigantic memory
Earlier if you had to look for a specific key, you had to travel whole universe. Now you just need to go to specific location.
Very helpful. Thanks.
very very awesome ...
hash("\b0")=hash("\0\0C"=64 also = hash(64) ?