yo yo

hey

hey

hey

yo wazzup lol

wow

yo yo yo

yo yo

make yourself so good at what you do that they can't ignore you

maybe we'll be okay to be deficient here for a while

create urgency/scarcity

this is basically the point

--> the people who are checking out are the people you want to target, so that subset is smaller than say all the people mildly interested

MAC is low-end

need to know these

which is better? Manual or automatic? Won't fb fuck us on automatic?

#Qs

what is meant by this? conversion pixel-- a pixel made just for conversion?

#Qs

wdym automatic bidding?

#Qs

consider use

ayesha song is the one im looking for, hindi song daddy was listening to

pricey

so essentially it seems that this is about syncing, and considering that a facebook account is something for a little later we can postpone this until further discussion

this piece is just an FB ad push, but it basically says that FB is good at timing, devices, relevancy, and scale.

this might actually be something propietary, but it is something that can be done by hand regardless

from what i understand they're just asking facebook to do the work. they're not actually doing any work themselves

he only made about 20k... + whatever he sold it for

he worked through a niche, we should maybe try that if this doesn't work out

this is his niche

what nonterminals can derive epsilon

not derived from the B, B disappears, goes to epsilon in the follow set of B

rule 7 makes the capital B disappear

soon as you have the idea that capital leters can disappear you find that first is not enough

our new rule

first alone cannot cover all cases

first terminal expect to see if i have an S on the stack is AyB

first of s that i will see is the begin of file(?)

what is first all about--> use to fill up part of our predict table

have to match-- then you just pop it off

How do we know what to expand next???

That is the rest of the lecture.

summary slide

S' is non-terminal so expand

only one rule to expand S'

terminal is a blue letter

replace with AyB

not like natural deduction only one way

one of the most important questions

children

root

we do not want ambiguity when evaluating a parse tree

big n, small p

x is successes in hypergeometric

so just x failures followed by 1 success

same as negative binomial but k = 1 so x+1-1 = x of course where x is the number of failures

how many ways can we order successes over n trials

how many ways can we order the fucking failures given x+k-1 trials

last success is fixed

follows NB if you have S/F with independent trials, with replacement

k successes x is # of failures until k success

reqt of all probability functions

for a = THH

= (8C5)(3C3) = 561 which is also valid

flaw: you cannot make one 4 digit number in p ways and then turn around and make another 4 digit number in q ways and then claim that the there pxq ways to make a single 4 digit number. it doesnt make sense

only that particular even number is not available, so:

[3,7] and even is only 4 and 6--> subtract 2

there are multiple ways to do it and some are simpler than others

if you have einstellung with one, skip it and move on and you may find a simpler solution

--> same deal in (b)

same deal as in (a)

split into cases when desired/required

Position is unique, not the letters. The letters are indistinguishable.

i.e: S1=S2=S3 but position 1 != pos2!=pos3

the complement of all pins without 1s

so all pins with 1s which is the same as (9^(3)*2)/(10^(4)) since there are two (2) places for the 1 to go

every simple event is equally likely

.

must also consider/connais that all DFAs are NFAs

basically your start pos is multiple start pos's

solns?

yourew moving from any number of these states so of course the final state set is going to be the result of a union

accepting state

intrsxn w accepting states, right?

so here a subset is basically just concatenating instead of putting into a set

this is a simplistic diagram

you could have 1 taking you back to q0 from q1 and q2 since you want 2 0s at the end

alphabet

i.e: combinations of the alphabet

R^(i) = R⋅R^(i-1)--- remember: you already have an R as a "coefficient"

so: R^(0) = R⋅R^(-1) ??? Wait, then what's R^(-1) which then must exist by this definition?

ANS:?

concatenated to itself i-1 times (for each term in the sum)

superscript indicates length

basically like the '*' in regex

distribute

as you will recall, has infinite numbers of valid identifiers

language spec..

not recognize

token analysis

page 43:

i dont really understand this proof

page 43:

nonneg: 0 or positive nonpos: 0 or negative

page 37:

so basically its already known that Span B is a subset of S, basically by definition:

We already know that all the vectors in B are in S and since S is closed under linear combinations, Span B is a subset of S (i.e: can be written as an LC of the vectors in S). (you cannot go beyond)

the best you can do with n-1 vectors (even if you are touching all 4 dimensions) is to create a hyperplane assuming linear independence

page 31

thm 5 explanation:

• we cannot have any other coefficient set which will make the equation 0 besides the trivial solution

so we find that ci=di for every vector 1<=i<=k in S

the proof makes sense in example 14 since we show that

• R2 is a subset of Span B and that
• Span B is a subset of R2
• B is linearly independent

page 16

example 17 makes sense since you show that

• any vector in R2 is a linear combination of the vectors in B by offering a c1, c2 s.t. we have a soln for every x1,x2
• linear independence is had as we find that the only soln to the problem of x1=x2 =0 is c1=c2=0 (trivial

page27

essentially giving out all the bindings

prob that b occurs given a + prob that b occurs given a complement

this just reduces into p(b)

P(A)P(B|A) = P(A)P(AnB)/P(A) = P(A)

notice it keeps going so you would then add back in P(AiAjAkAlAm)

clever way to ensure distinctiveness

yes my method works (of choosing one of the four locations) so:

P(D)=4*9^(4)/10^(4) is right also

since we can only have one 1, otherwise you would need to use their method of finding the complement

makes sense

happens later, chapter 2 is genuinely soft

In this solution we simply flesh out the sample space and assign a prob'y dist'n to the events.

Then we simply count up the number of events which fulfill our criteria and produce the probability.

here, the same event is a compound event.

note that only one of these events will occur in a single trial in both cases, but there are multiple ways in which the event can occur (club)

as you can see with this example and the next,: the sample spaces need not be unique in order to produce the same result.

you do indeed have 5 options for the even digit, and then 9,8,7 for the rest of the digits

structural recursion with an accumulator is called accumulative recursion, stores subproblems

LIST OF TOKEN TYPES

so that: you can ID errors you can output the appropriate output when desired

basically a fancy way of saying "word"

MSB is a 1 so you dupe that and you will get a 2s comp (i.e: signed) -1, i.e.: 0xf

0x87654321 has the MSB as 1 which will trigger the different behaviours

recall that 8= 2^3 and is uniquely represented as the nibble 1000

printing from MSB to LSB (>>24 comes first!)

• cuts out MSB --> NOPE:it means what it means, see how char keeps cutting out the MSB until it is equal to the LSB?

it is saying that these are ALL EQUAL

THE FINALE!

consider that a 0 bit shift left would not add any additional 0s to the right.

1 bit shift left would put 11 zero there.

26? 26 zeroes

number is 15, so we're just going to keep the right nibble on

why is beyond on 24 when its just a label??

ANSWER: You don't want to be going back up to 20 and rerunning it every time. That would be wrong

builtin mechanism where if its not there it will get added

aka 32 bits

this is a good slide for pseudocoding your soln at least

relating to meaning

relating to form

actually, how big can the number be?

a what?

maybe they just mean some "type", basically

note that the scanner is providing the output as a decimal value

to stop program early, i suppose

fall off is lost

sas university is also an option

rethink the belows' use

idgi but sounds important for recursive+mutually-recursive

idgi but sounds important

what?

ote that the values of top-level expressions are printed after evaluation if they are not<void>.If you</void>

referring to a resrouce listed above (in ss 1)

if you got an 'a' then you go to the empty set

at the same time

if you end up in q1 and q2 then you will accept (at least 1)

not a single answer but a set of answers

you are here AND in q1

update state based on input

can think of as just a big lookup table

state symbol pair

in S and see a b go to b

in state b see an e then go to be,,

and so on

can be empty

subset of states

as stated before, know where im going next, for every step, i know here im going to next EX;; b + e = be-- > we know this, non-dfa would not know this

beginning in an accepting state

unbounded

Kleene Star

if input stopped here ---> in an nonaccepting or error state -------> not a sucessful termination of the program, will JUST hang

a or b

accepts this specific string only

no concentric circle state, accepts {} (nothing, epsilon)

idea of error state

equivalent of arrows in prev diagram

why is sleepy not an accepting state?

psbl ans: can't stay in "sleepy"

goal is to end up in an accepting state

as many as you want

BUT only one start state

meaning every move is determined

May 24th, 2018

obviously going to be infinite

^what the star means>>

means the empty string

most prog langs

combinations of our set

irl it would be ascii

easy to figure out whther say, youve been given a valid register (finite)

but this is an infinite sized langauge --> so we must specify and recognize them

how can i break strings into a series of tokens that i understand (?what?)

easily swung this way or that emotional given to sudden and unaccountable changes of mood or behaviour

why is the order reversed?

why do we update then load & restore then update?

stack pointer location

stereotype is generally one dimensional and negative archetype is the source from which copies are made

Stereotype is a "conceptual model" created by abstracting the key features of current examples. ---> abstracting means extracting so an investment banker stereotype may have the abstracted traits of selfishness and vanity

origin stories

author perspective

i.e: pulling it out from the archetype intos omoething else

contain elements across these boundaries we have just created

GIST

might be based in reality

completes, vs compliment which is just a nice thing to say

only skip on non-zero (i.e: summation is >0)

looks important

order matters!!!!!!!!!!!!

hazards: IMP

abbrev meanings

1 X is funct-field determined

crux

class 1 of instructions

idk about irl/ in the courseontes but this is imp/good to know

data memory is the fucking data which is not the fucking register

when?

used for branch instructions, ex: beq

32 bits wide

32bits wide

wired to be a perma-adder

alu or and for instance