Open in TheBrain

Solving Polynomial Equations 1

Wild Egg Maths

3.87K subscribers

How do you solve a polynomial equation? Let's play around with this famous problem and see if we can tinker our way to some new insights. Along the way we strengthen our understanding of polynomial or polynumber arithmetic, and its extension to on-polynumbers or power series, and get some experience with the power of Scientific Workplace for writing and working on mathematics.

00:01

hello everyone i'm roman wilder

00:04

and thanks for joining me on this

00:06

adventure where we're going to be

00:08

looking at

00:09

polynomial equations so i'm going to

00:12

take you through

00:13

some ideas that we might have a look at

00:16

this is a very classical problem the

00:18

problem is

00:19

how do you solve polynomial equations

00:21

going back of course

00:22

uh you know to antiquity to the

00:24

babylonians who are interested in

00:25

quadratic equations

00:29

and then in the 15th 1600s

00:32

a lot of work on the cubic and quartic

00:35

equations

00:36

and then some sort of difficulties as

00:38

people started to realize that

00:40

degree five and higher equations were

00:42

somehow different and then that

00:44

the standard procedures that the

00:47

classical

00:48

mathematicians had developed weren't

00:49

working that well so

00:51

right now i would say that we're not in

00:53

a very satisfactory situation

00:55

and i'm going to look at this problem

00:58

and hopefully with some fresh eyes

01:00

so let's start by just doing some simple

01:03

things which will also allow me to

01:05

generate a little bit of experience for

01:07

you working with

01:08

scientific workplace okay let's say uh

01:11

the variable is x

01:12

and let's have a very simple equation

01:14

say a plus

01:16

b x equals zero okay there is a

01:18

polynomial equation of degree one

01:20

i've made it very simple let's see what

01:22

our computer

01:23

tells us is the solution so we're going

01:25

to ask it to solve

01:26

exact and we're going to solve for x

01:30

okay we find that x is equal to

01:34

right here x equals minus a over b

01:37

provided that b is not equal to zero

01:40

fine

01:44

okay let's be a little bit more

01:46

adventurous let's uh

01:48

look at a quadratic equation

01:52

but let's actually do a specific one

01:53

just to be a little bit more

01:55

concrete so let's look at two minus

01:57

three x

01:58

plus um say five x squared

02:02

okay there's a quadratic equation let us

02:04

try to solve it

02:07

solve it exact okay we get

02:10

two solutions and we see that they're

02:12

complex here why are they complex

02:14

there's some i's here it's because the

02:16

discriminant of this equation

02:18

is 3 squared or minus 3 squared

02:21

minus four times two times five so we

02:24

get

02:25

complex solutions

02:29

since the discriminant

02:33

which is uh minus three

02:36

squared minus four times a

02:40

times five and that that's equal to

02:43

minus 31

02:44

that is is negative

02:48

okay so let's try a variant on that

02:51

let's make the discriminant positive so

02:54

we could do that

02:55

say for example by bumping this three up

02:57

let's make it

02:58

say i don't know nine

03:02

and then what will happen is

03:06

solve exact okay now we're getting um

03:11

solutions in the standard form involving

03:14

some sirs or some square roots so

03:18

this involves thirds

03:22

which for us are not

03:26

exact numbers okay so we're going to

03:29

draw upon

03:30

our wealth of experience here you all

03:33

know that i don't believe in square

03:34

roots

03:35

that these are actual numbers these are

03:37

algorithms that can be run but they

03:39

don't actually finish so this is a

03:41

kind of a red herring and we can see

03:43

that

03:44

more clearly and directly by

03:47

by asking for a numerical solution

03:52

solve numeric okay now we get the two

03:55

solutions but they're of course only

03:57

approximate so here are two solutions

03:59

to five decimal places or so okay

04:03

so we see that even the quadratic

04:06

equation

04:07

has has come complications we're not

04:10

really able to solve it in some way

04:12

okay let's be now more adventurous and

04:14

let's go to a cubic equation

04:16

say three plus x minus a two

04:19

x squared okay plus okay uh

04:22

let's say four x cubed

04:26

equals zero all right what's going to

04:28

happen if we

04:30

try to solve it let's first solve it

04:32

numerically

04:34

okay so this thing here has three

04:36

solutions but two of them are complex

04:39

and one of them is

04:42

real okay and and all of them are

04:45

approximate

04:46

so this the graph of this thing which

04:48

seems to cross the

04:50

x-axis in only one point around x equals

04:53

minus 0.69 okay so but

04:57

that's still not really a solution i

04:59

mean those are approximate solutions but

05:01

they're not really exact solutions

05:03

so let's have a look at trying to solve

05:05

this thing exactly

05:06

solve exact okay and now we

05:10

run into some complicated formulas that

05:13

are really going back to cardano

05:15

and uh antar taglia and uh and

05:18

del faro and uh and the italians where

05:21

they realize that the formulas for the

05:23

cubic equations were rather complicated

05:26

they involve cube roots you can see some

05:28

cube roots here

05:29

and also some square roots inside

05:32

and of course you know this is very

05:34

unwieldy i mean to actually use this

05:37

is is almost impossible really at least

05:39

before the age of computers

05:41

and so this is really a major stumbling

05:43

block i mean people got

05:44

these cubic equations and they could

05:46

write down some formulas but

05:48

they are really not very useful so um

05:52

so it's a problem and and

05:55

if you go higher higher degrees well

05:57

let's say you go to four

06:00

four degrees let's see that okay let's

06:02

modify this a little bit

06:04

five and let's take a quick a quarter

06:07

term in here okay so now we have a

06:09

quartic equation

06:10

and let's solve it first

06:13

numerically uh we've got two

06:17

real solutions and two complex solutions

06:20

okay and now i'm going to just

06:21

solve it exact

06:26

oh in this case this is actually

06:28

relatively simple so maybe i've just

06:30

chosen some

06:31

really particularly simple ones here

06:33

okay you see

06:34

this solution is quite simple that's not

06:37

interesting

06:38

in that it only involves square roots so

06:42

that's a happy accident

06:43

let's modify it a little bit if i change

06:45

this x to the

06:48

fourth here to 7x to the fourth and i

06:50

don't think that's going to happen

06:52

again so now let's calculate the

06:55

solution of this

06:55

exactly okay and

06:59

you can see that it's somewhat analogous

07:02

to the cubic thing in fact the cubic and

07:04

the quartic equations are

07:05

are closely related um and again we have

07:08

some

07:09

rather complicated um you know formulas

07:11

involving

07:12

nested uh radicals okay

07:16

so this is somehow the the classical

07:18

story i mean it's a very

07:19

long and laborious story but um the the

07:23

basic idea

07:24

or direction is the classical idea

07:29

is to try to

07:32

reduce such equations

07:35

to simpler equations involving just

07:39

radicals

07:41

involving just radicals such as

07:44

i.e x

07:48

to the n equals a

07:51

okay so if you have an equation like

07:54

that then you express the solution

07:56

with an nth root and

07:59

then you feel confident that you've

08:01

gotten something even though

08:03

we would not accept that as a really

08:05

legitimate answer but anyway that was

08:07

the orientation

08:08

of the classical people

08:12

and so we want to do something different

08:15

okay we want to look at this entire

08:17

story from a new point of view

08:21

all right so let's try to be more

08:23

algebraic

08:24

okay so we're going to

08:27

we'll dispense with the degree one

08:29

equation and we're going to go to a

08:31

degree

08:31

two equation and see if we can kind of

08:34

think about you know how to solve a

08:36

quadratic equation maybe in a slightly

08:37

different way

08:39

so let's look at a general equation so

08:41

instead of looking at particular numbers

08:43

let's look at a general equation

08:44

a general quadratic what does a general

08:48

quadratic equation look like

08:50

well it's got a linear term and i'm

08:53

going to

08:54

express that'll be term in mathematics c

08:57

sub zero

08:58

let's say that's the constant term and

09:00

then plus say c

09:01

sub 1 times x plus c sub 2

09:05

x squared equals zero okay there's the

09:09

our kind of generic quadratic equation

09:11

where

09:12

we're not specifying particular values

09:15

of c0 c1 and c2

09:17

now note that

09:20

we really ought to assume

09:24

assume that uh the c2

09:29

first of all c even c zero that c zero

09:32

does not equal

09:33

does not equal zero because if it c zero

09:36

equals zero then

09:37

you can factor out an x and then it

09:39

becomes a linear equation so it's so

09:42

somehow already we've dealt with that

09:45

and we can also assume that uh c2

09:50

does not equal zero because if c2 equals

09:53

zero then we hit it back to a linear

09:54

equation okay so

09:55

we'll make these two assumptions but

09:57

otherwise we'll keep the nature of c0 c1

10:00

and c2 somewhat

10:02

open in particular we might even

10:05

eventually

10:06

think about c0 and c1 and c2 as

10:09

you know variables

10:13

now you will know those of you who have

10:16

watched my videos that i don't like to

10:17

think about variables i rather think

10:19

about

10:19

poly numbers okay and so we would want

10:22

to

10:23

think about that those variables as

10:25

being developed in a polynumber

10:27

framework

10:28

but we're not going to go in that

10:29

direction that's on the side that maybe

10:31

we'll talk about

10:32

at some other point in in this at this

10:34

stage here we're just going to sort of

10:36

think about this as a fundamental little

10:37

equation okay and we want to solve this

10:39

okay so we want to solve for x we want

10:42

to solve for x

10:46

well in terms of c0

10:50

c1 and c2

10:54

so what what kind of

10:57

thing are we looking for well we're not

11:00

not going to allow radicals we've just

11:02

said that because radicals are not

11:04

something that we could actually

11:05

calculate

11:06

so what are the kinds of things that we

11:08

can calculate

11:09

right we're just interested in

11:11

restricting ourselves to

11:13

to basic algebraic operations which are

11:16

addition

11:17

subtraction multiplication and division

11:21

so really what we're doing is we're

11:25

at least you know one approach would be

11:27

well what would happen if you

11:29

just tried to express x as a polynomial

11:32

in c zero c one and c two

11:36

well it's not too hard to see that uh

11:39

that's not going to work

11:40

so if x is a polynomial or let's say a

11:45

poly number if you allow me to do that

11:47

in c0 c1 and

11:50

c2 then

11:54

this is not going to work well

11:58

because there's going to be top degrees

12:00

and then when you plug in the top

12:01

degrees you can see that the top term is

12:03

not going to be 0.

12:05

but in a big but when

12:08

poly numbers or polynomials don't work

12:10

sometimes

12:12

poly numbers that keep on going on poly

12:15

numbers

12:16

polynomials that keep on going formal

12:18

power series

12:19

they sometimes have a better chance of

12:22

working

12:23

okay that's an important kind of

12:26

understanding that

12:29

if you can't make polynomials

12:32

fit what you're doing you should maybe

12:34

think in terms of power series

12:36

even though maybe what exactly a power

12:38

series is is problematic but we're just

12:39

playing around so let's let's see what

12:41

we can do

12:41

but um

12:45

how about how about a on

12:49

poly number or maybe

12:52

or maybe usually called a power series

12:57

okay um so let's try a power series

13:01

and uh so for that we need a kind of a

13:04

you know a variable and

13:07

where you're maybe going to choose one

13:09

of these

13:12

things to be variable why don't we say

13:14

try c0

13:15

let's set

13:18

c 0 equal to say

13:22

t and think of this as a a variable for

13:24

a power series

13:26

and and we're looking for and we look

13:30

for

13:32

a solution which is a

13:35

on poly number or a power series

13:39

in t so we look for a solution of the

13:40

form x

13:42

equals okay maybe a

13:45

sub zero plus a sub

13:49

1 1 times t

13:52

plus a sub 2 times t squared plus

13:57

3 dots carrying on in some fashion or

14:00

another

14:01

okay so that's that's interesting so

14:03

what would happen if we just sort of did

14:05

that

14:05

if we sort of tried to fit a poly number

14:09

um or on poly number solution to

14:13

our original equation well

14:17

one way of trying to make sense of this

14:20

so one way of giving meaning

14:24

to this is to restrict

14:28

oneself to just terms

14:31

of a certain fixed

14:34

degree or less that is

14:38

we ignore all terms beyond some point

14:42

so for example let's say um

14:46

let's say to be specific that we'll

14:49

let x be

14:56

we'll go a little bit more than

14:57

quadratic maybe uh a sub

14:59

3 t cubed plus

15:03

a sub 4 t to the fourth plus

15:06

a sub 5 t to the fifth

15:09

and we're going to go only up to

15:13

5

15:17

or degree 5. let's say up to degree

15:21

5 x is equal to a 0 plus a 1 t plus a

15:25

two t

15:25

squared plus a three t cubed plus a four

15:27

to the fourth plus a

15:28

fifth t to the fifth so in other words i

15:32

e

15:32

what we're doing is we're kind of set

15:35

set

15:36

t to the six equals zero

15:40

okay that's sort of what we're thinking

15:42

about

15:43

we're trying to just

15:46

approximately get a hold of something

15:48

and so we're just going to agree to

15:49

truncate our

15:50

our algebra and work in this fixed

15:54

domain up to say degree 6 just to see

15:56

what happens

16:01

all right so let's plug in and see what

16:03

we get

16:05

we get okay so our original expression

16:08

was

16:10

c0 plus c1 x

16:13

plus c2x squared and

16:16

we've already agreed that c0

16:20

is going to be replaced with a slightly

16:22

easier to deal with

16:23

t okay and so now we're going to

16:28

just let me put that t at the beginning

16:29

there we're going to replace

16:31

the x with

16:40

with the expression that we have for it

16:41

so c sub 1

16:43

times x what is x x is this combination

16:47

of powers of t

16:50

and similarly this x is also going to be

16:53

replaced

16:54

with that particular combination so now

16:55

the whole thing is a polynomial in t of

16:58

degree

16:59

  1. let's expand it out

17:02

to see what it looks like and let's

17:04

actually to make it a little bit

17:06

easier to understand we're going to

17:07

collect terms so

17:09

organize it in terms of t

17:13

all right there it is in decreasing

17:16

powers of tt the 10 to

17:18

the nine etcetera it down to to this

17:20

thing here so let's have a look at uh

17:24

at all this i

17:27

want to use the fact that we are

17:30

agreeing that t to the 6 is 0.

17:32

so all of these terms from t to the 10

17:34

down to the t to the 6

17:35

can be removed all right

17:39

so all of this is our original

17:43

polynomial and the whole point is to try

17:45

to set this equal

17:46

to zero and solve for x in this case it

17:49

means we have to solve for the

17:52

the a's now

17:55

this constant term the constant term

17:59

gives us

18:03

what's the constant term the constant

18:05

term is this thing right here

18:06

so if we factor this

18:13

that's supposed to be equal to zero so

18:15

we get two solutions there are two

18:16

possibilities so two possibilities

18:20

we could get a sub zero a

18:23

sub zero equals zero

18:27

and the other one is that a a

18:30

sub zero equals

18:33

uh i guess we that's just a linear

18:35

equation so that's minus

18:37

c sub one divided by c

18:40

sub two

18:43

and recall that we've agreed that c sub

18:45

2 is not

18:46

zero so that quotient does make sense so

18:49

these

18:50

two possibilities now sort of provide a

18:52

fork in the road there's going to be

18:54

two kinds of solutions that we're going

18:57

to be able to generate one from each one

18:59

of them

19:00

so let's look at the simpler one first

19:02

okay

19:03

so we'll start with

19:07

a sub zero equals 0.

19:10

and that's sort of a case so i'm going

19:12

to actually make a little

19:15

subsection here okay

19:21

okay what happens if a sub zero equals

19:23

zero

19:24

well that's just one of the terms so

19:26

that constant term will be zero but we

19:27

want this entire polynomial

19:31

in t to be zero so we also get

19:35

a number of conditions the

19:38

maybe i'll start with that one so a sub

19:40

0

19:42

equals 0 so we're agreeing on that and

19:44

then

19:46

the linear term here

19:51

that's going to be zero the quadratic

19:54

term

19:58

that's got to be zero

20:03

the cubic term that's going to be zero

20:08

quartic term that's going to be zero

20:12

and the degree five coefficient

20:18

all right let's stare at this for a

20:19

little while so what are we trying to do

20:20

we're trying to solve

20:22

for the a's in terms of the c's

20:29

okay uh we've got a zero here the second

20:33

equation

20:34

is a linear equation in a

20:37

one

20:42

a linear equation in a one involving a

20:46

zero okay but a0 we've said is 0. so

20:49

actually this

20:50

term here is going to be 0. so

20:54

let's simplify things all right i'm

20:55

going to take that term and i'm going to

20:57

kill it

20:57

because of our assumption that a 0 is 0.

21:00

similarly this thing here

21:02

will disappear anything else with a0 in

21:05

it

21:05

that will disappear

21:12

a0 so we also

21:16

get these relations

21:22

okay now we want to solve for

21:27

a1 a2 a3

21:30

a4 and a5 now scientific workplace

21:34

actually

21:35

is able in some cases to solve

21:39

a whole bunch of systems like this so

21:41

we're going to solve for a

21:43

sub 1 a sub 2

21:46

a sub 3 a sub 4 a

21:49

sub 5 maybe we'll put in a sub 0 just

21:52

to be complete even though we already

21:55

know what that one is

21:56

okay let's see if scientific workplace

21:59

will

22:00

figure out what the values are hey

22:03

it does so

22:06

we get

22:10

we get

22:17

these values for

22:23

our solution so what is this a zero is

22:26

zero

22:28

ie so now we can write down what x is x

22:30

is going to be

22:32

uh actually but the way i'm going to do

22:36

this

22:36

uh just to show you something else i'm

22:38

going to use a definition so i'm going

22:41

to

22:41

select this and i'm going to make this

22:42

definition up here okay that's a new

22:44

definition

22:45

and here are all the definitions there's

22:47

only one definition which is that a1

22:49

equals

22:49

minus 1 over c1

22:53

and then here um

22:57

i'm going to select this and make that

22:58

into a definition

23:03

make that into a definition

23:07

a4 into a definition

23:10

and a five in definition and i'll just

23:12

check my definition so i've got one for

23:14

each one of these

23:15

things a1 3a5 and so now i can go back

23:18

up to

23:19

x which is this thing here

23:31

and i can just evaluate this thing

23:36

okay that's kind of cool and a

23:40

zero um i didn't i should have also

23:43

selected that as something else so i'll

23:45

add that to the list

23:46

okay zero's on the bottom there so

23:48

that'll then

23:50

kill this thing and we'll get this as an

23:52

expression

23:53

for x

24:00

sorry try that again

24:04

display this

24:13

all right what have we got here it looks

24:15

like we've got something

24:17

that looks like a solution

24:20

to our original

24:24

quadratic expression or quadratic

24:27

formula

24:28

as a polynomial or poly number an

24:32

ongoing polynomial

24:33

in t and remember t was equal to

24:36

c zero where i remind you

24:40

t equals c zero

24:44

and there's uh notably there's powers

24:48

of c1 in the denominators

24:51

there are powers of c1

24:54

in the denominators increasing powers

24:58

so that means we better

25:02

ensure so c one

25:05

had better not equal zero

25:09

we better remind ourselves right we

25:11

can't allow multiplication

25:13

or division by zero and and otherwise

25:17

there are

25:18

some coefficients

25:21

with additional coefficients

25:28

one of these coefficients uh 1 1

25:31

ignore the minus sign 2 5 14.

25:39

now these are very famous numbers

25:43

these are very famous numbers

25:49

they are the first

25:54

five catalan numbers

26:01

so it looks like this elementary

26:05

kind of manipulation has allowed us to

26:08

find

26:09

a little formula at least the start of a

26:11

formula

26:13

for a solution to a quadratic equation

26:15

in which the

26:17

catalan numbers kind of miraculously

26:20

appear

26:21

and then there's then it going to be a

26:24

corresponding

26:25

other solution which this

26:29

uh the a0 will have to be replaced with

26:31

is more complicated minus c1 over

26:33

c2 okay but i think we can

26:36

already see here that we're kind of

26:38

interested we're getting somewhere here

26:40

right

26:40

we're making some progress where we're

26:43

seeing

26:44

the catalan numbers fall out it means

26:45

that we're actually

26:47

touching base with the classical

26:49

formulas because it turns out that the

26:50

catalan numbers are

26:52

very closely connected with the square

26:54

root of function like square root of one

26:56

plus four x or something has a has a

26:58

one over over that has a power

27:01

series involving catalan numbers so it

27:04

looks like this is maybe some kind of

27:06

algebraic way of

27:07

trying to you know get at the usual

27:10

square roots

27:11

so i think this is a good place to stop

27:13

um we've made an initial foray

27:15

into this and obviously we're

27:19

raring to go to see how we can go

27:21

further

27:22

in this direction so we'll be doing that

27:24

next time hope you'll join me for that

27:26

i'm norman wildberger thanks for listening

Attachments