ShmoopTube

00:00

Thank you We sneak and here's your smoke Use your

00:05

brought to you by double variables double the precision Double

00:10

the things you have to worry about All right Well

00:12

what can you expect about the outcome from running this

00:14

code This coat right here and here The pencil answers

00:19

shmoop overflowed running All right so let's go let's Start

00:25

with a look at what This coat actually does Step

00:27

by step we'll first we create a double variable called

00:30

numb with the value is your appoint one Then we

00:33

begin a four loop that starts with an injured eye

00:35

at zero stops when eyes no longer less than one

00:38

hundred and increment i by one Each time the loop

00:41

is run Essentially this loophole run one hundred times each

00:44

son loop cycles we make numb equal the value of

00:47

one divided by num minus nine point nine Phillips put

00:50

real numbers to it This means for the first generation

00:53

one divided by num a zero point one would equals

00:57

ten Got it one by one then ten minus nine

01:01

point nine his point one And we're back where we

01:03

started so Luke will run 99 more times and since

01:06

none will always equal point one when we cycle the

01:09

result will always be point one again In theory we're

01:13

dealing with a double variable here double means double precision

01:17

floating point variable got it and floating point variables aren't

01:22

the rigid representation of riel numbers we might expect In

01:26

fact the floating point bearable is more like a very

01:28

good guess at a very precise number It's never quite

01:32

exact even the very simple point one plus point two

01:35

in floating point variables wouldn't give you a point Three

01:38

you get something like point three zero zero for or

01:44

thereabouts hearts let's try something if you have your compiler

01:47

handy add the following lines to this code and run

01:51

it and we'll see what happens if you don't have

01:53

your compiler handy well and just we'll put system out

01:57

front line number right here in the loop That way

02:00

it'll print the current value of numb each time the

02:03

loop cycles and we'll be able to see how it

02:05

changes over time and for those not ableto participate Well

02:09

here's a picture of wild turkey these things you're huge

02:11

all right and here's what The result looks like weird

02:15

Right What happened Well the imperfections in the floating point

02:20

operations kept magnifying themselves over and over and over getting

02:24

worse and worse and loris with each iteration before finally

02:27

settling in at negative ten Those faras our potential answers

02:31

go option a Yep they're being overflowed No no overflow

02:35

ares occurred Option b there will be under flow so

02:39

after running numb will equal zero Also no option c

02:43

after running numb will equals your appointment Well sure the

02:46

code may look like things will remain in zero point

02:48

one on paper but using a floating point variable means

02:51

the exact value of zero point one is not gonna

02:54

happen Option d after running none will not equal zero

02:58

point one Well that's right For a moment it even

03:01

seemed like num was equal to just about everything but

03:03

zero point one option e both andy well no ass

03:07

for overflow and there was no So our answer is

03:09

d and we're number one or rather we're number one 00:03:12.768 --> [endTime] point lobo too

Full Transcript