000000000000000000000

0:00 the following content is provided under a creative commons license 0:03 your support will help m_i_t_ opencourseware continue to offer high 0:07 quality educational resources for free 0:10 to make a donation or view additional materials from hundreds of m_i_t_ 0:14 courses 0:15 visit m_i_t_ opencourseware ato cdw dot m_i_t_ dot e_d_u_ 0:25 there are welcome to six oh four one six four thirty one the class and 0:29 probability models and the like 0:31 i'm john city place that would be teaching these classes and 0:35 ivan looking forward to these 0:37 an enjoyable and also useful experience 0:41 we have a fair amount of staff involved in this course of your his station use 0:45 tracked or send us a bunch of d_a_'s but they want to single out 0:49 our heads the eighth was llama 0:52 who is the key person in this class everything has to go through him 0:56 if she doesn't know inc which is a patient section you are 0:59 then simply do not exist 1:01 so keep that in mind 1:03 all right so that we want to jump 1:05 right-wing into this object but i'm going to take just a few minutes to talk 1:09 about the few 1:11 administrative the face and how the courses ron 1:14 so we're going to have lectures twice a week 1:17 and i'm going to use old-fashioned transpired since now you get copies of 1:21 these slides with plenty of space for you to keep notes on them 1:25 uh... is useful way of uh... 1:27 making good use of the slides 1:29 is to 1:30 use them as this sort of mnemonic summary of quite happens in lecture 1:35 not everything that's been going to say it is of course on the slides 1:39 but they looking them to get the sense of cuts happening right now and it may 1:43 be a good idea to review them 1:45 before you go to the station 1:47 so quick happens here is stationary station heuristic datians track there is

1:51 going to maybe review some of the fury 1:54 and then solve some problems for you 1:57 and then you have tutorials where you meet in very small groups together with 2:01 your ta 2:02 and quick advancing tutorials is that you actually do the problem solving with 2:06 a couple of your ta and the country if your classmates injured injured tutorial 2:11 section 2:12 now probabilities that tricky subject you may be reading the text missing two 2:16 lectures everything makes perfect sense 2:18 and so on but until you actually sit down and try to solve problems you don't 2:23 quite appreciates the subtleties and the difficulties that are involved 2:27 so problem solving is a key part of this class 2:30 and tutorials are extremely useful just for this reason because that's where you 2:34 actually get the practice or solving problems in your own as opposed to 2:38 seeing someone else 2:40 whose solving them for you 2:43 okay about mechanics a key part of what's going to happen two days that you 2:47 would alternate juror 2:50 schedule forms that their at the end of the handouts that you have in your hands 2:55 vendor to his will be at working frantically through the night 2:59 and they're going to be producing 3:01 barely has to offer who goes into quite section 3:05 and find that happens 3:07 any person in this class with grow beating ninety percent is going to be 3:11 happy with their assignment 3:13 and with probity ten percent they're going to be unhappy 3:17 now unhappy people haven't options though 3:20 you can't resubmit your form together with your for scheduling constraints and 3:25 give it back to the fifty eight 3:27 who will then do some further juggling 3:30 and reassigned people 3:32 and after that happens ninety percent of those unhappy people will become happy 3:37 and ten percent of them 3:39 will be 3:40

left unhappy 3:42 well pays so finds the probative that out that random person is going to be 3:46 unhappy at the end of this process 3:49 it's one percent excellent but maybe you don't need this class 3:53 okay so one percent would have about one hundred people in this class 3:56 so there's going to be about one unhappy person 3:59 i mean anywhere you look and life there's hope in many group you look at 4:03 there's always want unhappy person writes okay 4:06 what can we do about that 4:09 for rights and other important part about the mechanics is to read carefully 4:12 the statement that we have about collaboration with identical amnesty and 4:16 all that 4:17 you're encouraged its a very good idea to work with other students 4:21 you can consult sources that are out to their but when you sit down and write 4:25 your solutions 4:26 you have to do that's by setting things aside and just strike them 4:31 when your own you cannot 4:32 copy something that somebody else has given to you 4:37 one reason is that we're not going to like it frequently happens 4:42 and another reason is that you're not going to do yourself any favor at really 4:46 the only way to do well in this class is to get them out of practice by solving 4:49 problems yourselves so if you don't do that on your own then 4:54 quinn quiz and exam time comes 4:56 if things are going to be difficult 4:59 so as i mentioned here so we're going to have section three special sections 5:03 but some of them are 5:04 for six oh four one student summer for six thirty one students the gadget 5:08 section of the class 5:10 now undergraduates and see it in good budgetary situation sections what's 5:14 going to happen there is that 5:16 things maybe just a little faster and you may be covering 5:20 a problem but certainly tomorrow constant is not covered in the undergrad 5:23 sections 5:24 but if you sit in the dreaded section just they just respond to your an 5:28 undergraduate you're still just responsible

5:30 for the undergraduates 5:32 materiel 5:33 but these you can just do the undergraduate work in the class but 5:36 maybe be exposed at the different section 5:43 few words about the style of this class 5:46 we want to focus on basic ideas 5:48 and concepts 5:50 there's going to be lots of formal s but quite we try to do in this class is to 5:54 actually have you understand clinton's former lesbian 5:58 auntie ny here from now when almost all of the form this could be wiped out from 6:02 your memory 6:03 you could still have the basic concepts you can understand them so quinn you 6:07 look things up again 6:08 the loose tooth 6:09 make sense 6:11 uh... it's not the 6:12 it's not the plugin sog 6:14 kind of classwork 6:16 mhm you're given list or formulas you given numbers and you plug in and you 6:20 get the answers 6:21 through the feely hearts parties usually too 6:24 shuru squeaks form unless you're going to use unit judgment you'd need 6:27 inflation 6:28 lots of probability problems to please the interesting ones 6:32 orphan cover lots of different solutions some are extremely long summer extremely 6:36 short 6:37 the extremely short once usually involves some kind of deeper into 6:41 spending boards 6:42 of what's going on 6:43 so that you can't because shortcut 6:45 and use it and hopefully you if you're going to develop these skills 6:49 during this class 6:52 dash 6:53 i could spend a lot of time these lecture talking about quiet this subject 6:57

is important for help keep it short because i think it's 7:00 almost all of these 7:02 anything that happens in life is uncertain 7:05 there's uncertainty any question of whatever you try to do you need to cut 7:09 some way of dealing or thinking about the sand certainty 7:13 and the way to do that in the system but the kuwaitis by using the models that 7:17 are given to us by probability theory 7:20 so if you're an engineer and and you're dealing with the communication system or 7:23 signal processing basically you're facing a 7:26 against noise 7:28 noises random is uncertain how do you model it come to you deal with it 7:33 if you're in my mentor 7:34 yes you're dealing with customer the month which is of course rundle or 7:38 you're dealing with the stock market 7:40 who each is definitely a problem or 7:43 you play at the casino 7:45 which is a game condom and so on 7:48 and the same goes for pretty much any other field that you can finkel 7:52 but uh... independent book week 7:55 field you're coming from 7:57 the basic concepts and tools such a deal to sing 8:00 so you may see in bookstores that there is 8:03 their books 8:03 uh... probability for scientist from a breeder for engineers probability for 8:07 social scientists 8:09 probably deforest religious ts 8:11 well quite all those books have 8:13 inside then these exactly the same are those the same equations the same 8:17 problems 8:18 they just make them some put the difference words problems 8:21 the basic concepts are just one of the same 8:24 and 8:25 who will take this is an excuse for not going to much into the specific domain 8:30 applications 8:31 we will cut problems and examples that their

8:34 motivated in some loose ends from real-world situations 8:38 but we're not really tried in this class too 8:41 develop the skills for specific 8:44 for domain specific problems 8:46 rather worked from going to try to stick to general understanding of the subject 8:52 okay so the next slide oak which you do 8:55 which you do have been here 8:56 sent out to gives you a few more debates about the class 9:00 uh... made you wanting to comment here she is the truth do we need to privy to 9:04 the facts 9:06 and with kyle cruise books perhaps you can live with adjusted to page summary 9:10 of for the 9:11 interesting formulas and cut loose ends you can get by 9:14 uh... 9:16 just with those formulas but here because we want to develop concepts and 9:20 intuition actually reading words 9:22 as opposed to 9:23 just browsing through questions 9:25 does make a difference 9:27 the in the beginning the class is kind of easy 9:30 when we deal with discrete probability that's too much young killer first weeds 9:35 and some of you may get by without being through systematic about forming the 9:39 material 9:40 but it does get substantially car there afterwards 9:44 and they will keep 9:45 he stating that if you do have to read the text to really understand the 9:49 material 9:53 so now we can 9:54 uh... starts 9:55 who theory a part of the lecture 9:57 let us sets the golds for today 10:01 so probability of probability theory 10:04 is a framework 10:05 for dealing with uncertainty 10:08

for dealing with situations and quit without some kind of front on this 10:12 so quickly want to do is by the end of today's lecture 10:15 to gave you 10:17 anything that you need 10:18 to know how to set back what does it take to set up a probabilistic model 10:24 and 10:25 quarter of the basic 10:26 could rules of the game for dealing with probably stick models 10:30 so by the end of this lecture you will have essentially recovered couple for 10:34 this semester's tuition 10:36 so we're going to talk about probabilistic models in more detail if 10:40 the sample space which is basically descriptions off all the things that may 10:44 happen 10:45 during a random experiment 10:47 and the probability lock which describes our beliefs about which outcomes are 10:51 more likely to hope for 10:53 compared to other outcomes 10:56 uh... probability laws have to obey certain properties that we call the 10:59 actions of probabilities so 11:01 bait main part of today's lectures to describe those actions which are the 11:05 rules of the game 11:07 and concede their if u delete reveal examples 11:12 poke it so let's start with our agenda 11:15 the first piece in the probabilistic models is a description of the sample 11:19 space with an experiment 11:21 so if we do an experiment 11:24 and by experiment we just to mean 11:27 that's just something happens out there 11:30 and that's something that happens it would be flipping a coin 11:34 or it would be her calling uh... guy 11:37 it could be 11:38 doing something in a card game 11:41 so we fix a particular experiment 11:44 and sweet come up with the latest or for them 11:47 possible things that may happen through this experiment

11:51 so we have rights down the list of all the possible outcomes 11:54 so here's the latest of all the possible outcomes of the experiment 11:59 uh... used the word lists 12:00 but see if you want to be a little more formal 12:03 it's better to think of at least as a set 12:06 so we have a set 12:08 that said these are some things space 12:11 and feeds us at www spell amends are the possible outcome so 12:14 the experiment 12:15 so for example if you're dealing with flipping a coin your sample space would 12:19 be 12:20 pads that this is one outcome oakdale sees one outcome 12:24 and to be set which has two elements is the sample space of the experiment 12:29 opaque co at do we need to think about when we're setting up the sample space 12:34 first of the list should be mutually exclusive collectively exhaustive web 12:38 does that mean 12:39 collecting the exhaustive means that's not my clerk what happens in the 12:42 experiment you're going to get 12:45 one of the outcomes inside here 12:47 so you have not forgotten any of the possibilities of but make happen in the 12:51 expected 12:53 mutually exclusive means that if this happens 12:56 then thoughts 12:57 cannot happen 12:58 so at the end of the experiment you should be able to point out to me 13:03 just one exactly one of these outcomes and say this is the outcome dot 13:10 well so these are sort of basic requirements 13:13 there's another requirement which is a little more loose when you set up your 13:17 sample space sometimes you do have some freedom about the dictates of them 13:21 well quite your 13:22 or how you're going to describe it 13:24 and the question is how much detail are you going to include 13:28 so let's take this coin flipping experiments and think of the following 13:32 something space 13:34

one possible outcome is heads second possible outcome is fails and it's 13:39 raining up 13:41 and the third possible outcome-based pails and it's not rain 13:49 so this is another poor symbol of sample space for the experiment 13:53 square a flip a coin just once 13:56 it's illegitimate one these three possibilities are 14:00 mutually exclusive and collectively exhaustive which one is the rights of 14:04 the samples bases it to this one or dot blot 14:08 well you do you think that's my point flipping inside this room is completely 14:12 and related to the weather outside 14:15 then you're going to speak with the sample space 14:18 he phoned the other hand you have some superstitious beliefs that's maybe rain 14:23 has an effect on my points 14:25 you might 14:26 work with the sample space of the sky 14:30 probably wouldn't do that but that's a legitimate sorption strictly speaking 14:35 now this example is a little bit date on the freemen decides about the issue that 14:39 comes up here she is a basic one that shows up 14:42 inquiry in science and engineering whenever you're dealing with the model 14:46 or with the situation 14:48 the zealand's of debates in that situation and quinn who come up with the 14:51 model you choose 14:53 some of those details that you keeping your model 14:57 and some of that you say well these are irrelevant or maybe they're small effect 15:01 secondly that neglect them 15:03 and you keep them outside your model so there's definitely cling to go to the 15:07 real world is definitely an element of art 15:10 and some judgment that you need to do in order to set up 15:14 an appropriate sample space 15:20 sewing easy example amount 15:24 course the elementary examples are callings cards dyess 15:29 so let's deal with diced best to keep the diagram small instead of her six 15:33 idea died where going to think about the died at the only has four faces 15:38 so you can do that with it the procedure and doesn't really matter basically it's 15:41 a dike

15:42 but when you are all if you get the result which is one two three or four 15:47 however the experiment that i'm going to think about 15:50 will consist of 15:51 to roles 15:53 orf at di 15:56 crucial point here 15:57 i'm rolling the dice twice but then thinking of this there's just one 16:02 experiment 16:03 to different experiments not a reputation for 16:08 twice of the same expect 16:10 so it's one big experiment 16:12 during that big experiment various things will happen such as and running 16:16 the died once and then i'm rolling 16:18 the dike place 16:22 so quite a sample space for that experiments one the sample space 16:26 consists of the possible outcomes 16:28 one possible outcome is that your first of all resulted into 16:34 and the second floor resulted in three ink which days the outcome that you get 16:39 is this one 16:40 at to followed by three 16:42 this is one possible outcome 16:45 the way a m said describing things these outgunned is to be distinguished from 16:50 these outcome here 16:52 quirky three 16:53 is followed by two 16:57 you're playing but come on 16:58 it doesn't matter 16:59 which one of the two 17:02 but if you had 17:03 do we deal with the probabilistic model that wants to keep track of everything 17:07 that happens nbsp composite's experiment 17:11 there are good reasons for distinguishing 17:14 between these two outcomes 17:16 quinn this happens it's definitely something different from that's 17:19

happening at the photo basically is different from the three followed by yet 17:23 to 17:24 so this is the correct some pool space for this experiment quit we've rolled 17:28 the dice twice 17:29 it has a total sixteen elements and it's of course a finite set 17:35 sometimes 17:37 instead of describing sample spaces in terms of lists or sets or diagrams of 17:42 this kind 17:44 it's useful to describe the experiment in some sequential witty 17:48 whenever you have an experiment that consists of multiple stages 17:52 it's might be useful at least visually 17:55 to give a diagram from that shows you how those pages involve 17:59 and that's what we do by using in sequential description caller id 18:04 three based description 18:06 by drawing it three all for the possible if a lucien ps 18:09 during our experiment 18:11 so industry i don't think you off the first stage in quick i_d_'s role of the 18:15 first died 18:16 and therefore forcible results one two three and four 18:20 and given what happened in the first in blitzing the first of all suppose i got 18:25 that one then i'm rolling the second die 18:28 and therefore possibilities for quite a make up into the second died and a 18:32 possible results are one two three and for again 18:36 so quick to relation between the two diagrams 18:39 for example the outcome to followed by a three 18:42 corresponds to this 18:45 owns the tree 18:46 so this possible responsive to folded by you three 18:50 annie potts these associates to a particular outcome any outcome is 18:54 associate that particular box 18:57 and things of the five th you may want to think in terms of believes of these 19:01 diagram sing-sing finkel each one of the leafs is being one possible outcome 19:08 and of course we have sixteen outcomes here 19:10 with f_-sixteen outcomes here 19:12 uh... maybe you're not just the subtlety that they used in might land which they

19:16 said they are all the first died 19:17 and the result that they get for the is it too 19:20 i didn't use the word outcome 19:22 i want to reserve the word outcome to mean 19:26 the overall outcome 19:28 at the and to feel their own experiment 19:33 is the 19:34 outcome or for the experiment the experiment consisted of stages 19:38 to was the result in the first stage three was the result in the second stage 19:43 you put all those results together and you get your outcome 19:47 okay perhaps you're splitting hairs here but it's useful to move 19:51 keep the concepts to keep the concerts right 19:56 quite special about this example of his that's besides being trivial because of 20:00 samples based which is finite 20:03 there's sixteen possible total outcomes 20:06 not every experiment has a finite sometimes face 20:09 she is an experiment think which the sample space is infinite 20:12 so you are 20:13 you're playing darts 20:15 and that dark it is the square 20:17 and you're a perfect for your perfect at that game so you're sure that your darts 20:22 we've always fall inside the square 20:26 so but query exactly if your doctor would fall inside that squared up itself 20:30 is behind them we don't know what it's going to be it's uncertain 20:34 so all the possible point inside the square are possible outcomes of the 20:38 experiment 20:39 so it typical outcome of the experiment is going to be at the airport numbers 20:43 ex-wife critics of the whitewater real numbers between zero and one 20:48 knob is infinitely many of the other numbers 20:51 the scene finicky many points 20:52 into square 20:54 so this is an example ink which s 20:56 sample space 20:57 is an infinite set 21:01

smoking so we're going to the revisits this example immediately 21:06 okay so these are for example also quiet the samples base 21:10 might to be 21:11 in simple experiment 21:13 now the 21:14 more importance order of business 21:17 is not to look at those possible outcomes 21:20 and make some statements about their relative likely coats 21:23 quakes outcome is more likely to a fair will clear compared to 21:28 the actors 21:29 and the way we do this 21:30 is biased signing 21:32 probabilities 21:33 to the outcomes 21:36 well not exactly 21:38 suppose that you or you work to do was to assigned probably just to individual 21:43 outcomes 21:44 if you go back 21:45 to this example 21:48 and you concede that one particular outcome let's say this point 21:52 quite could be the probability that you kids exactly this points to infinite 21:56 precision 21:58 intuitively that there be would be zero 22:01 so any individual points in this diagram 22:04 in any reasonable model should have zero probability 22:08 so if you just tell me 22:10 that's any individual outcome zero probability you're not really telling me 22:14 much 22:15 to work with 22:16 for that reason quest instead we're going to do 22:19 is to assign probabilities to subsets of the sample space 22:24 as opposed to assigning probabilities 22:26 individual outcomes 22:29 so here's the picture 22:32 we have our sample space

22:34 shes ul magda 22:36 and sweet pussy there's some stops that for the sample space colleagues any 22:41 and they want to assign 22:43 in yuma and number in numerical probability prove this particular subset 22:49 quenched them 22:50 uh... represents my belief about how likely 22:54 this set 22:55 is twill clear 22:57 will kick whether we mean cool color 22:59 and and introduce each year language that's being used in proably dethier 23:03 teak when we talk about subsets of the sample space 23:06 week usually called them events as opposed to subsets 23:10 and there is a missed because it works nicely with the language so that's the 23:15 describes what's going on 23:16 so the outcome is a point 23:19 the outcome is random the outcome maybe 23:22 insides 23:23 these set in quick stays we say that the vents a poker 23:29 if we get them out come inside here 23:31 or the outcome make for outside the set 23:34 ink which days we say that the vent eight 23:37 did not occur 23:38 so we're going to assign probabilities to events 23:42 and 23:43 dahil how should we do the society swell 23:46 probing sermons to describe your beliefs about quick sets are more likely to 23:50 occur 23:51 versus others said so there's many ways that you can assign those probabilities 23:56 but there are some ground rules for this game 23:59 first week one probably just to be numbers between zero and one because 24:03 that's unusual convention 24:06 uh... so probably qf zero means we're certain that something is not going to 24:10 happen probably do if one means that we're essentially served in that 24:13 something's not going to happen so we want to numbers between zero and one 24:17

we also want to few other things and those few other things are going to be 24:21 encapsulated in this set of 24:23 axioms 24:25 quests axioms means in this context it's no ground rules that's 24:29 an illegitimate probabilistic model shoots will be 24:33 you have a choice of how 24:35 what kind of probabilities you use but not much of a quite you use 24:39 they should still be a certain inconsistency properties 24:42 because if they obey those properties then you can go ahead and do you swear 24:46 calculations and do some useful reasoning 24:49 soak whether these properties 24:51 first probabilities should be 24:53 non-negative 24:56 and that's our convention we want pro-business to be numbers between zero 24:59 and one 25:00 so they should certainly be non-negative the probative at the event angle person 25:04 should be in on negative number 25:06 has a second back to you 25:08 the proably be over the entire sample space 25:11 is equal to one 25:13 why does this make sense 25:16 the outcome is certain to be and elements of the sample space 25:21 because we set up a sample space which is collecting the exhausted 25:24 no matter who aren't 25:25 no matter what the outcome is is going to be an element of the samples phase 25:29 where certain 25:30 that event omega is going to occur 25:33 therefore we require that we represent lisa seconded by saying that the probe 25:37 b_d_o_ almeida is equal to one 25:41 pre fee 25:42 straight 25:43 forward so far 25:47 the more interesting axiom is the third 25:52 uh... before getting into it for just a quick reminder if you'll have to assess 25:57 a and bonnie off

25:59 the intersection of a ends up being consists of those elements that the long 26:04 both to a 26:05 and to be 26:07 and would you not to talk this way 26:09 when you think probabilistic live the way to think of intersection is by using 26:12 the word and 26:15 these events this intersection 26:18 is the event that able colored 26:20 and beat or colored if i get them out come inside here eight has acquired and 26:25 to be has occurred at the same time 26:27 so if you may find the word tends to be a little more convenient than the word 26:31 intersection 26:33 and similarly without some updation for the union bluff to leave ends 26:39 which group are arrived by two eight 26:42 the union of two sets or to leave ends is the collection of 26:46 all elements that belong i'd there too the first set sort of the second or two 26:50 blocks 26:51 quick to talk about the events you can't use the word or 26:55 so this is the event that 26:57 in your current 26:58 or beetle current 27:00 and the beast or it means that it would also be that's both of them 27:03 orkut 27:08 okay so now that we have this notation what does the first the third axiom sick 27:14 third acts hill says that 27:16 if we have to leave and stay in beat 27:19 that have no coleman elements 27:23 so shears eight 27:26 she is being 27:28 subspecies are big something space 27:31 the two events have no common elements of the intersection of the two events 27:35 he's the empty set there's nothing in there intersection 27:39 then that bought the probability of aid to gather with b 27:42 has to be able to the sample for the individual probabilities 27:46

so the probative that they'll press or be evil person is equal to the probe 27:50 into that field first plus the probability that's being smokers 27:55 so think of probability as being cream cheese 27:58 you'll have one pound of 28:00 cream cheese 28:01 that the other probability assigned to the 28:04 in fact something space and for that cream cheese is spread out 28:07 or where 28:08 this 28:10 uh... over it 28:11 this set 28:12 the probability of 28:13 date is how much pinches sits on top of any prbly kilby's how much sits on top 28:18 of b 28:20 the probability corp 28:21 any when you'd be is that part of the month of cream cheese sitting on top of 28:26 the issue 28:27 and dot 28:28 which is all bruce lee the sound of how much is sitting here how much is sitting 28:32 there 28:33 so probabilities behave like cream cheese or if they became like 28:37 months 28:38 for example of the total miles 28:40 uh... older 28:42 the miles of set 28:45 you think of some material objects to the moscow 28:48 the set consisting of two pieces is obviously the sample of the two mosses 28:53 so these properties of the included one itself 28:56 treating natural onto ca 29:00 uh... how are these actuals enough for what we want to do 29:03 i mentioned acquire the gold 29:05 that's we want probabilities to be numbers between zero and one 29:10 she has a much so that those who have probably bizarre non-negative 29:13 should we have another axial that tells us 29:16 that's probabilities are

29:18 less 29:19 than or equal to one 29:21 it's a desirable property we would like to have it in our patents 29:26 opaque wise it's not feel that list 29:29 the people who are in the actual nb 29:31 making business or mathematicians and mathematicians tend to be pretty lock on 29:35 it 29:36 you don't say something if you don't have to say 29:40 and this is a cake this is the case here 29:42 we don't need that extraction because we can do you drive it 29:46 from the existing axioms 29:48 here's how it goes 29:50 one used the probability or the entire sample space 29:55 here was using the second axiom 30:00 now the sample space 30:02 is consists twelve 30:04 update two gether with a compliment of 30:07 okay so this is 30:10 uh... when we try to read the complement of n_e_a_ mean the complement of h 30:14 inside the set 30:15 omid 30:16 so we have almeida 30:18 sheer state here's the complement of create and here we're all set 30:23 is ul magda 30:25 now 30:26 quest the next step question they do next which action should they use 30:31 we use axiom three because as seth and the complement of that said 30:35 are disjoint they don't have any common elements 30:38 so i had two or three applies 30:41 and tells me left vcs different okay 30:45 plus the proverbial fewer a compliment 30:48 in particular 30:49 the probability of date 30:52 is equal to one minute of the probability of copy compliments 30:56

and this is less than or equal to one 31:00 quietly 31:03 because probabilities are non-negative 31:06 by the first axiom 31:10 okay so we got to the conclusion that we wanted probabilities are always less 31:14 than or equal to one 31:15 and this is a simple consequence of the three actions that we cut 31:20 this is this 31:21 truly nice argument because it actually uses 31:24 each one of those axioms the argument is simple but you have to use all of these 31:28 people properties 31:30 to get the conclusion that you want 31:33 so we can get interesting things out of her actions 31:37 can we get some 31:38 work interesting ones 31:40 how about 31:40 the union 31:42 three sets 31:44 quickened of probability should be cut 31:47 so here's an event consisting off 31:50 of three pieces 31:52 and they want to say something about the probability of aid union v 31:57 union c 31:59 quite i would like to say if he is that these probability 32:02 is equal to the sample of the three individual probabilities 32:07 how can they do it 32:08 i have an action that those me that they can do it for two weekends i don't have 32:13 a maximum for three events 32:15 well may become my size things 32:17 and still be able to use that back to you 32:20 and here's the trick 32:22 uh... the union of three sets you can think over it as 32:27 forming the union of the first two sets 32:30 and then taking the union with the cleared 32:36 so taking unions you can take the unions in any order that you walked

32:40 so here we have the union off 32:43 sets 32:45 a_b_c_ sardi's joint 32:48 bias on two nd or 32:49 that's how it through it 32:51 so if they'd be n_c_i_ disjoint then aid 32:54 union be these these joint from c 32:58 so here we have the union of foodies joint sets so by the edit delete the 33:02 axiom 33:03 the probability or that the union is going to be the probably chiapas first 33:07 step 33:09 plus the probability of the second step 33:12 and now i can use the edit delete the axiom once more 33:15 to writes that places probably of a task will be if you are being 33:20 chloroplast probably you'll see 33:23 so by using this action quicklist dated four two sets we can actually do dr 33:28 missy miller property for the union of 33:30 three disjoint sets 33:32 and then you can repeat this argument as many times as you want it's valid for 33:36 the union offended joint sets for that you know for a hundred disjoint sets 33:41 for the union of any finite number of sets 33:44 a soviet anyone up quite and at parties join 33:51 then appropriately 33:53 of anyone union 33:55 union 33:56 these people to do something for the probabilities of the thing to do is it 34:05 special case of this 34:08 east when we're dealing with finite sets 34:10 suppose they have 34:11 just the finite set of outcomes 34:14 i put them together unicef 34:16 and them interesting the probability of that stuff 34:19 so here's our sample space 34:22 there's lots of outcomes but then taking 34:24

if you will be just 34:26 and they form cassette 34:28 out of them 34:30 this is a set consisting of 34:32 in the speaker of three elements 34:34 in general it consists of j elements 34:39 the finite sets i can drive if as a union 34:42 of single element sets so they said he is is the union obviously want elements 34:48 epf together with this one element sets together without 34:52 one elements f 34:53 so the bottom probity of the sad thing is going to be this some of the 34:57 probabilities 34:59 of the one element set 35:03 probability of one elements that 35:06 you'd need to use the brackets here 35:09 because probabilities are assigned two sets 35:12 but this gets kind of deduce so here one abuses notation in a little bit 35:17 and we get rid of those brackets and just write probability obvious single 35:22 individual outcome 35:24 mandates conclusion from this exercise is that 35:28 the thought the probability cough ah... 35:30 finite collection of possible outcomes 35:34 that bought the probabilities 35:36 equal to the some of the probe in piece of 35:38 individual elements 35:42 so these are basically the axiom suf 35:44 probability theory 35:46 or well 35:47 their almost to the actions 35:50 there's some 35:50 subtleties that are involved here 35:53 one subtlety 35:54 is that these facts and here it is 35:57 doesn't quiet to do the job for everything we would like to do 36:01 and we're going to come back to this at the end of the lecture

36:05 as second subtlety 36:06 is has to do with weird sets 36:10 we said that in any event is a subset of the sample space and we assign 36:14 probabilities to events 36:16 does this mean that we are going to assign probability to every possible 36:20 stops that for the sample space 36:23 ideally we would wish to do that 36:26 unfortunately this is not always possible 36:29 if you take a sample space such as the square 36:34 the square has nice subsets of those that you can describe by cutting it with 36:38 lines and so on 36:40 but that does have some very obvi 36:42 subsets as well that the 36:45 that they're enforceable to visualize impossible to imagine but they do exist 36:50 and those very weird send sets are such that there's no way to assign 36:54 probabilities to them 36:55 in the way that's consistent with the actions probit 36:58 okay so this is a very very fine points 37:01 that you can't immediately 37:03 forget for the rest of this class 37:05 uh... you will only encounter the sets if you end up doing doctoral work 37:10 on the theoretical aspects of for would be good fury 37:14 so 37:15 so it's just the mcmichael subtlety 37:18 uh... that's something we had sets do not have probabilities assigns to them 37:22 but we're not going to encounter these sets and they do not show up 37:26 you mean you're petitions 37:29 okay so now let's revisit are examples let's go back to the dying exotic 37:34 we have our sample space 37:36 now we need to assigned at probably did not 37:40 there's a lot of possible probability laws that you can assign 37:44 i'm picking one here 37:46 arbitrarily 37:48 inc which they say that every possible outcome has the same probability orf 37:52

one over sixteen 37:56 quiet do i make this model who work in theory cli if you could overwhelm 38:00 manufactures dies 38:02 date bands to behave that way 38:04 were will be coming back to this kind of story later in this class 38:08 bucks 38:09 i'm not saying that this is the only probability law that there can be you 38:14 might have we had to dies 38:15 in quick certain outcomes are more likely 38:18 that matters 38:19 but to keep things simple let's take every outcome to have the same probe 38:22 into flying where sixteen 38:24 okay so let's now 38:26 now that we're having our hands this sample space and the probate law 38:30 wiccan actually solve any problem there is 38:33 we can answer any question that would be post looks for example 38:37 clubs the probative that the outcome which is the spare 38:40 these either one one or want to 38:43 we're talking here 38:45 about these particular events one one or 38:48 want to 38:50 so it's an event consisting of these two items 38:53 according to but who were just discussing 38:55 the probe ngo for finite collection of outcomes is this some of the individual 39:00 probabilities 39:01 each one of them chris probe into one over sixteen th so the probe in q_v_c_'s 39:05 to over it 39:07 how about the probability of vive en 39:10 that x is equal to one x is the first of all so that's the probative that the 39:15 first of all 39:16 uh... is equal to one 39:18 not is this impex that's being used here 39:22 probys are assigned to stop sets two sets 39:26 so we think of this as meaning the set of all outcomes 39:30 soch sadat's excuse equal to one

39:33 how do you want sir this question for you go back to the picture and you try 39:36 to visualize or identified these event of interest 39:40 x is equal to one corresponds to 39:44 these event here these are all the outcome so quick specs is equal to one 39:48 there's four outcomes each one has probably the one over sixteen so the 39:53 answers for over sixteen 39:57 how bollocks the 40:01 probally that's explains why 40:03 is ald 40:06 hoping that will take a little bit smaller worth 40:09 but you go to the sample space injured mt five all of the outcomes at which the 40:14 song isn't load the number 40:16 that's a place where it be dishonest olds 40:20 these are other places 40:24 and ideas that that soloists all the possible outcomes 40:28 etc which we have 40:29 and old song 40:31 uh... we count them how many of their there's a total of eight of them each 40:35 one has per week he won over sixteen total probity spades over sixty 40:40 and harder question of course the probe into the demeaning war for the two 40:43 worlds is he quoted to 40:45 this is something that you probably couldn't do in your head 40:49 without the couple for diagram 40:51 but once you have a diagram things are simple 40:54 you ask the question okay 40:57 this is an event that demeaning or more that we were all sitting close tutu this 41:01 can happen in several ways what are the several ways that it can happen to go to 41:05 the diagram and try to identify them 41:08 so the meeting will musical to to if both of them part to lose 41:13 antha or it could be that excuse to and why is bigger worldwide is to and x is 41:19 big 41:23 yes you will rediscover that uh... 41:26 yellow and blue make green 41:29 who we see here 41:30

that there is a total of 41:33 forcibly outcomes 41:34 the probability of bc bent 41:37 usa five over sixteen 41:41 simple example 41:45 but the procedure that we followed in this example actually applies to any 41:50 probabilities 41:51 model you might ever encounter 41:54 you set up your sample space 41:56 you make a statement that describes the probability law over that sample space 42:00 than somebody asks you questions about various events you go to your pictures 42:05 identified those events 42:07 seeing them down 42:08 and then starts kind of counting and calculating the total probability 42:13 for those outcomes that you're conceding 42:16 uh... this example is a special case so quiet is called the discrete uniformed 42:20 law 42:21 these 42:22 and model obeys the discrete uniform law if all outcomes 42:26 art equally likely 42:28 it doesn't have to be that way that's just one example off a probability law 42:33 but when things are about to eighty four lookout comes alert equally likely 42:37 and we have and love them 42:42 and and you have 42:43 and set a that has little ten elements 42:48 then each one of those elements has probably the one where capital and since 42:54 all outcomes are equally likely 42:56 and for approval be destroyed up to one teach one must have this much 42:59 probability 43:01 and there's little and elements that gives you 43:03 the probably just be vento interest 43:06 so problems like the one in the previous slide the more generally of the tide 43:09 described here under discrete uniform law 43:12 these problems reduced to just counting 43:15 how many elements are there nice ample space how many elements are there

43:19 inside the vent off interest 43:21 counting is generally simple but for some problems it gets pretty complicated 43:25 hansa 43:26 in a couple of weeks we're going to have to spend the whole lecture 43:30 just on the subject of how to come systematically 43:33 now the procedure we followed in the previous example 43:36 is the same as the procedure you would follow in continues probe into problems 43:41 so going back to our docs problem we get there around them point inside the 43:45 square 43:46 that's our sample space we need to assign the probability lol 43:50 for lack of committee nation on taking the probing feel ought to be the area 43:54 well for subset 43:56 so if we have 43:57 to subsets of the sample space 44:00 that have equal areas 44:02 then compost relating that they are equally likely to occur 44:06 the probe either they for here is the same as the probe into that they fall 44:09 out there 44:11 the model doesn't have to be that way 44:13 but if i have sort of complete even around so quick sponsor more likely than 44:17 others 44:18 but mine to be the reasonable model two years 44:21 so equal areas mean equal probabilities 44:24 uh... if the n_e_a_'s twice as large the probate is going to be twice as big 44:28 so this is our motto 44:32 and i would come now 44:33 answer questions let that should be easy one quest the probability that the 44:37 outcome is exactly this point 44:40 that of course is 44:42 zero because a single point 44:45 president zero area 44:47 spence's probate is equal to area that's zero four will be 44:51 how about the probability that through some 44:54 over the coordinates over the points that we've got 44:57

is less than or equal to one cup 45:00 how do you deal with it 45:01 well you look at the picture again invade your sample space 45:04 and try to describe the event that you're talking about 45:08 there's some being less than one-half 45:10 corresponds to getting in outcome that's below this line 45:14 square this line is the line court explicit way close to one-half 45:19 loading intercepts over that line with the axes are a one half and one half 45:25 so you describe the events visually 45:28 and then you use your probability law the probe into law that would have is 45:32 that the probability or first set 45:34 is equal to the area of that sets old we need to find is the area of this trial 45:38 go 45:39 quick cheese one half times one tough times will huh 45:45 equals to one p 45:48 okay morrow from these two examples is that it's always useful to have the 45:52 picture and worker with a picture to visualize 45:56 defense that you're talking about 45:58 and once you have a probability law in your hands then it's a matter of 46:02 calculation to find the probe into something event off interest 46:06 the decorations we did in these two examples of course we're very simple 46:10 sometimes calculations may be unlawful harder 46:14 beats a different business it's a business of congress for example or 46:18 being bored in algebra and so on this fire is probably he's concerned 46:22 uh... it's clear quite you will be doing and then maybe you're faced with a 46:26 credit card to brake parts to actually carry out the calculations 46:30 the area for a triangle is easy to compute the fact that looks down a very 46:34 complicated shape then you might 46:36 needs to solve the card integration problem to find the area of that shape 46:40 but that stuff that's belongs to another class that you have 46:44 presumably mastered by now 46:46 food ok so now who spend just a couple of minutes to return to a point that 46:50 they days before 46:52 i was saying that the action but we had about spybot dvd 46:56 might not quite be enough

46:58 that's used frequently made by the following sample 47:01 think of the experiment where you keep flipping a coin if you wait until your 47:05 pain ads for the first time 47:08 quest the sample space of this experiment 47:10 you might have 47:11 you might have been depressed flip it might happen in the then flip 47:15 heads for the first time michael courting the medium flip 47:18 so the outcome of this experiment is going to be an integer and there's no 47:22 bob to that integer you might have to wait very much until that happens 47:26 so the nitro sample spaces the set of all possible integers 47:30 somebody tells you 47:33 some information about the probably did not 47:36 the probing that you have to wait 47:38 foot pamphlets is equal to add to the miners and 47:41 accorded this comp that's a separate story 47:44 quite did it come from 47:45 some but it does 47:46 to us 47:47 and they asked and those probabilities are cloth and here it is a function of 47:51 and 47:52 and your ass to find the probative that the outcome isn't even number 47:55 how do you both how do you go about calculating that's probably 47:59 so the probe into a being in even number is the probability of the subset that 48:04 consists of 48:06 justly even nombre es so it would be a subset some of these kind that includes 48:11 tool for and so on 48:13 so any reasonable person 48:16 would say well the pro build your folks taining in mouth combats either two or 48:20 four or six and so on 48:23 is equal to the probate you're hoping to get to 48:25 plus the probe into hoping in a four plaza probe into for pertaining to six 48:29 and so on 48:31 these probabilities are given to us 48:34 here have to do my algebra ikkada these geometric serious and they get the man 48:38

circle one-third 48:40 that's what made you recently person would do 48:43 but that there are some cool only knows the axioms the page 48:47 boasted 48:48 just a little earlier 48:50 may get stock 48:51 we would get stuck at this point called the cuddly justified this 48:58 we had this property for the union opened these joint 49:02 sets 49:03 and the corresponding property that does is that uh... 49:06 the total probability of 49:08 fy ninety many things 49:10 outcomes is this some of the individual probabilities 49:13 but here was using it 49:15 on meant infinite collection 49:17 the probate deal infinitely many points 49:21 is equal to the problem the some of the probative sophie x one of these 49:26 to justify this step 49:28 we need to introduce one of the sonoma rule 49:30 an additional axiom that tells us that these steps 49:34 is actually illegitimate 49:36 and this is the comparable i_b_p_ vp axiom quick israel into a stronger 49:40 or quite a bit stronger than the edit dvd action which had before 49:45 it tells us that the if we have a sequence offsets 49:48 that are disjoint 49:50 and we want to find their thought the probability 49:53 then we aren't allowed to obviate individual probabilities 49:58 so the picture might be 49:59 succesful knows we have a sequence of sets 50:03 and one p no v me 50:05 and so on 50:07 i guess in order to freak them inside the salmon space cassettes need to get 50:10 smaller and smaller perhaps 50:13 the ideas drawing 50:15 would have a sequence of subsets the bottle probably fuel falling any query

50:20 inside one of those sets is the sample of their individual probabilities 50:25 bakke subtlety that's involved here is that we're talking about the sequins off 50:31 his pants 50:32 you'd think by sequence we mean that these events can be contained in order i 50:38 can tell you 50:39 the first event for the second event the third event and so one so if you have 50:44 such a collection of events that can be ordered 50:47 as first second third and so on then you can their probabilities 50:52 you can find finds the probative you've their union so this pond is actually 50:56 flew more subtle that you might appreciate at this point and i'm going 51:00 to return to its at the beginning the physics lecture 51:04 for now enjoyed the first week of classes stands a good weekend

000000000000000000000