Sometimes it is also because at this point you already have some idea of algorithm and understand some of its key concepts, so when William teaches you this you have an easier time grasping it.
Good video, but there's a few things that could be improved. Residual edge is incorrectly defined at 6:00. At 4:20 he says you're probably wondering what a residual graph is, and then he proceeds to use the word residual before defining it! In Ford-Fulkerson, we assume each edge has an edge in the opposite direction. Where the capacity is only given in 1 direction, set the other direction's capacity to 0. Assign a flow of 0 to each edge, and then as the algorithm proceeds, the flow on each edge can be changed. But when you change the flow of an edge to x, you MUST assign -x to the flow in the opposite direction. This is NOT called a "residual". The residual exists in both directions. The residual (which means "quantity remaining") of an edge is what at 7:29 he calls "remaining capacity". It is simply the capacity of the edge minus the current flow. i.e. If the residual is > 0, there is still room for more flow. The residual graph is just the whole set of residual edges. The word saturate in general can mean "fill until no more can be held". In this case, a saturated edge has residual == 0.
Best explanation!! I watched some videos but this one was really in detail, giving examples and again repeating what a risidual network acually is, how capacities work and the principle of augmenting paths. Thanks a bunch my assignment was even fun when I understood it thanks to you~
Amazing explanation and I love the animation as well! I think it is worth pointing out that the key observation that the sum of the bottlenecks (found in each augmenting path) equals the max flow is only true because we assumed the initial arc flow is 0 for each arc. If we do not assume this, then you need to add the initial value of the flow (of the graph) to the sum of bottlenecks as well.
Thanks William, that was the best video i have seen for Residual Graphs and Augmenting paths, and trust me i saw many. You managed to make me undestand what my professor's couldn't in 100 silides.
If Timotheus' pole has a network flow capacity of 8 units spunk along its edge, delta net, how much bottleneck maximum flow can Timotheus' hole take before it fissures?
I got totally lost around 4:47 when he starts talking about edges being saturated...? And what is a residual graph? And I don't remember depth first search... What if there are two bottlenecks in the graph?
Good video, but there's a few things that could be improved. Residual edge is incorrectly defined at 6:00. At 4:20 he says you're probably wondering what a residual graph is, and then he proceeds to use the word residual before defining it! In Ford-Fulkerson, we assume each edge has an edge in the opposite direction. Where the capacity is only given in 1 direction, set the other direction's capacity to 0. Assign a flow of 0 to each edge, and then as the algorithm proceeds, the flow on each edge can be changed. But when you change the flow of an edge to x, you MUST assign -x to the flow in the opposite direction. This is NOT called a "residual". The residual exists in both directions. The residual (which means "quantity remaining") of an edge is what at 7:29 he calls "remaining capacity". It is simply the capacity of the edge minus the current flow. i.e. If the residual is > 0, there is still room for more flow. The residual graph is just the whole set of residual edges. The word saturate in general can mean "fill until no more can be held". In this case, a saturated edge has residual == 0.
Is there a particular reason why the "maximum flow" calculated using the "sum of bottleneck values"? Instead of using the "sum of outflow from the source node"? I found the "sum of outflow from the source node" easier to interpret, that's all.
The bottleneck approach is more useful to understand min-cut. Looking at the capacities that are actually filled gives you more information on how to reconstruct the optimal answer than looking at the outflow from the source. Sometimes, designing network flows is also easier to come about if you think about it in terms of min-cut.
I have a doubt: why we have to compute the augmenting path (s -> 0 -> 3 -> t), if the arrow goes from 3 to 0 (3 -> 0), and not from 0 to 3 (0 -> 3)? Thanks.
Can you please teach the same thing to our professor who is teaching us the network flow. He is trying to explain the same thing to us from last 3 lectures and out of 250 not even 25 of us is able to understand the same.
There is a lot good work in this video. At least give a hint on what is not clear to you. I think if you have your display and sound on, it is easy to follow and well explained
I wonder why this video is not highlighted in the search results. This is a wonderful video with effective teaching. You have done a good job William. Please Keep it going for lot more people to benefit!
Maybe when people learn theses concepts by themselves rather than being forced by the college it will be easier for them to understand. Anyway this video is awesome. Brief and clear. Thank you so much.
You spent less time than my Prof to explain this and you nailed it. Awesome!
My prof is so old and senile I swear to God he was out there helping Alan Turing crack the enigma back in the day
@@yoavmor9002 lol
Sometimes it is also because at this point you already have some idea of algorithm and understand some of its key concepts, so when William teaches you this you have an easier time grasping it.
I would thank my professor explaining the algorithm so terriblely otherwise I won't find this awesome channel!
Watching this in 2024 for my exam in Algorithms, love the explanation
My thoughts when I first saw this algorithm 4:17 - 4:23
Literally said it out-loud in zoom class last week ... accidentally.
@Archer Grayson lmao u really made two accounts for this
Good video, but there's a few things that could be improved.
Residual edge is incorrectly defined at 6:00. At 4:20 he says you're probably wondering what a residual graph is, and then he proceeds to use the word residual before defining it!
In Ford-Fulkerson, we assume each edge has an edge in the opposite direction. Where the capacity is only given in 1 direction, set the other direction's capacity to 0.
Assign a flow of 0 to each edge, and then as the algorithm proceeds, the flow on each edge can be changed. But when you change the flow of an edge to x, you MUST assign -x to the flow in the opposite direction. This is NOT called a "residual". The residual exists in both directions.
The residual (which means "quantity remaining") of an edge is what at 7:29 he calls "remaining capacity".
It is simply the capacity of the edge minus the current flow. i.e. If the residual is > 0, there is still room for more flow. The residual graph is just the whole set of residual edges.
The word saturate in general can mean "fill until no more can be held". In this case, a saturated edge has residual == 0.
thank you!
This is the best algorithm channel on youtube, thank you for this video.
Best explanation!! I watched some videos but this one was really in detail, giving examples and again repeating what a risidual network acually is, how capacities work and the principle of augmenting paths. Thanks a bunch my assignment was even fun when I understood it thanks to you~
Amazing explanation and I love the animation as well! I think it is worth pointing out that the key observation that the sum of the bottlenecks (found in each augmenting path) equals the max flow is only true because we assumed the initial arc flow is 0 for each arc. If we do not assume this, then you need to add the initial value of the flow (of the graph) to the sum of bottlenecks as well.
Love your network flow playlist, getting ready to watch it all again for the 2nd time. Great videos :D
Amazing as always !! The impact these videos will have over time will be phenomenal ... Let my comment bear testimony to that prediction ...
Thank you!
Agreed
this is so clear. my professors explanation was over an hour & made no sense. and you didn't oversimplify either! cheers
Only video that explains the concepts clearly without confusing the mess out of you🤣
This is so awesome!! Thank you! How come people dislike it?? 18 university prof?
Thanks William, your videos have already helped to get Tarjan's algorithm and this video is super clear, as well! ;-)
Your explanation is so clear! Also, I love the illustrations. Keep up the good work!
Best series on youtube yet
someone give this man a GOLD......wait this isn't reddit
Really well done! Best video for Max flow i have seen so far!
But what is a residual graph? You went straight to augmenting path.
Very nice explanation
Very clear
I love this so much
Great work on this! One of the clearest explanations I have seen
Can you make one video on Push-Relabel algorithm also?
Thanks William, that was the best video i have seen for Residual Graphs and Augmenting paths, and trust me i saw many. You managed to make me undestand what my professor's couldn't in 100 silides.
very good and clear video! I can now understand what is going on in the reverse operation!
i have an exam in an hour , i just love u
wached 3 videos before that didn't help me, I'm glad you did
Appreciate the videos! I'm in an operations research course and it's hard finding clips on some of these algorithms!
If Timotheus' pole has a network flow capacity of 8 units spunk along its edge, delta net, how much bottleneck maximum flow can Timotheus' hole take before it fissures?
I got totally lost around 4:47 when he starts talking about edges being saturated...? And what is a residual graph? And I don't remember depth first search... What if there are two bottlenecks in the graph?
Good video, but there's a few things that could be improved.
Residual edge is incorrectly defined at 6:00. At 4:20 he says you're probably wondering what a residual graph is, and then he proceeds to use the word residual before defining it!
In Ford-Fulkerson, we assume each edge has an edge in the opposite direction. Where the capacity is only given in 1 direction, set the other direction's capacity to 0.
Assign a flow of 0 to each edge, and then as the algorithm proceeds, the flow on each edge can be changed. But when you change the flow of an edge to x, you MUST assign -x to the flow in the opposite direction. This is NOT called a "residual". The residual exists in both directions.
The residual (which means "quantity remaining") of an edge is what at 7:29 he calls "remaining capacity".
It is simply the capacity of the edge minus the current flow. i.e. If the residual is > 0, there is still room for more flow. The residual graph is just the whole set of residual edges.
The word saturate in general can mean "fill until no more can be held". In this case, a saturated edge has residual == 0.
Watching this in 2023 for my exam in Algorithm and Datastructures, love the explanation
where do you study bro ?
In so sure hes also German haha @@Iam_number_one
teacher thanks. Muchas gracias con lo poco que entiendo ingles me ayudaste gracias profe.
Residual edges is more like an undo
In augmenting paths, we can only select the path whose remaining capacity is greater than zero
Great Explanation
This is beautiful, thank you so much!
Thank you for your helpful explanation!!!
Is there a particular reason why the "maximum flow" calculated using the "sum of bottleneck values"? Instead of using the "sum of outflow from the source node"?
I found the "sum of outflow from the source node" easier to interpret, that's all.
The bottleneck approach is more useful to understand min-cut. Looking at the capacities that are actually filled gives you more information on how to reconstruct the optimal answer than looking at the outflow from the source. Sometimes, designing network flows is also easier to come about if you think about it in terms of min-cut.
Great explanation!
love your stuff, now I need to rewatch it to sink it xd
Algorithme Ford-Fulkerson (Flot maximal) Méthode marquage
ua-cam.com/video/YnKXJAxUAu4/v-deo.html
Awesome introduction, thanks!
Best explanation. Thanks!
best youtube university
Awesome video man, life-saving. Very well explained
I have a doubt: why we have to compute the augmenting path (s -> 0 -> 3 -> t), if the arrow goes from 3 to 0 (3 -> 0), and not from 0 to 3 (0 -> 3)? Thanks.
He said at ( 9:09 )that we can use the residual edge, we have created. So basically we used the residual edge 0->3
@@ahlahous8128 Thanks!
thank you
Very useful. Makes good sense! :D
Thank you.
great explanation. thanks for the video and quality explanation with the visualizations!
At 2:42, the max flow is 7. But I am not able to get any min-cut of value 7, the least I cab generate is 8. Can anyone find a mincut?
What about push relabel. Its not in the playlist!!
Thanks mate appreciate it
From what time the F-F algo actually starts on example?
Not for someone who's trying to understand this for the first time. If one already knows, animations are nice.
Can you please make videon busacker and gowen method for network flow
Thanks!
Could you explain how you came at 2/100 at 11:15 ?
great video! keep it up :D
You sound a bit like Ross Geller from time to time lol! Great material tho, much appreciated hah!
You are the guy from data structures crash course by freecodecamp right??
which tool do u use for animation ?
thanks sir 😊
was this video supposed to be revision or first time learning? bc if it was for learning it was terrible cant lie
What is missing? Anything I can clarify?
@@WilliamFiset-videos you skimmed past probably the most confusing part of all of this: residual flow
if it weren't for youtube, i wouldn't be graduating college
Can you please teach the same thing to our professor who is teaching us the network flow. He is trying to explain the same thing to us from last 3 lectures and out of 250 not even 25 of us is able to understand the same.
kaunsa clg me padhte ho bhai
#Excelent!
we miss one or 2. checkout hieros gamos..
11:34
He took an augmenting path and augmented all over the place
Great video, shitty algorithm
εσπασες μας
8:08 😑😑
I WANNA KISS THIS MAN HOLY FUCK THIS SAVED ME A LOT OF TIME
Why not just do the opposite of a shortest path algorithm? wouldn't that give the maximum flow?
Hold on....isn’t this the guy from 3b1b?
poor explanation ! make an example and solve it instead of bla bla
i dont understand shit
Useless explanation
i now understand it less after watching, f***....
makes no sense
voice is copy of bill gates
i dont get it dude you explain too fast cant keep up disliked
terrible explanation, not clear at all
There is a lot good work in this video. At least give a hint on what is not clear to you. I think if you have your display and sound on, it is easy to follow and well explained
Good try but wrong information in some parts.
I wonder why this video is not highlighted in the search results. This is a wonderful video with effective teaching. You have done a good job William. Please Keep it going for lot more people to benefit!
thank you
13:25
how did the 4th/last augmented path had bottleneck value - 6?
Maybe when people learn theses concepts by themselves rather than being forced by the college it will be easier for them to understand. Anyway this video is awesome. Brief and clear. Thank you so much.