Lecture 32

IIT Kanpur July 201
20 followers
1 Video View·Mar 1, 2024
[Music]
uh welcome back my dear students uh and
participants for this course titled
multicriterial decision making I'm very
good morning good afternoon and good
evening to all of you and my good name
is ragun s Gupta from the ime Department
atanur in
India so the broader Concepts which we
are still continuing to discuss is
multicriterial decision making
multiobjective decision making which we
continue in this uh
lecture then going into multiattribute
decision making and multiattribute
utility
Theory so this is the 32nd lecture out
of the 60 lectures whole course which is
spread over 12 weeks and as we know that
each week we have five classes each
class being for half an
hour the concept which we'll cover uh in
this 302 lecture would be I'll try
obviously if there is a spillover as I
said we'll go into the the subsequent
lectures we'll consider the concept of
par optimality in 2D space concept of
par optimality in 3D space and try to
highlight that with
examples so the problem was rich farmer
in India growing four crops Ry sweet
bajra me selling price cost price
given then output per acreage for these
four crops given and all the constraints
are
given so based on that we had the
revenue cost
function and revenue Minus cost will
give you the profit remember that now we
come to the constraints the constraints
are if if you remember 2% of the land
has to be left for movement of tractor
and so on and so forth the
total land is th000 Acres 2% of that
goes so you have to basically have
98 percentage of that th000 which can be
utilized so this is the first constraint
which I'll put a dark yellow tick mark
the second one was related to that the
due to water restriction rice cannot
exceed the combined acreage of Basra and
wheat where bazra and wheat are the
third and the fourth so it's a third and
the fourth or let me see no is it's the
second and the third maze is the fourth
so rice cannot exceed the sum of the
mazra
and
wheat so it is
basically X1 is less than equal to X2 +
X3 second constraint third one
is that wheat is the staple diet hence
it production should be more than 50% of
the combined of rice bajra
wheat so rice is X1 bajra is X3 maze is
X4 wheat was
X2 so if I consider 50% more than 50%
are combined
so combined is for 1 third fourth and
more than 50 50% of that so it's 1.5 and
X2 is basically greater than equal to
that and we have mentioned that
each uh crop has to be in minimum of 100
acres but cannot exceed 400 so this is
the
fourth so these are the four constraints
now let us come
to the problem the problem is we want to
maximize the revenue which is true and
remember we are only going to maximize
the revenue we have not considered
anything of the cost till now so you
want to maximize the revenues Revenue
function was if you remember 35,000 X1 +
28,750 X2 + 36,000 X3 + 26,400 X4
constraints which we had
considered are all put
here this is a linear programming
problem with linear objective functions
linear constraints when we solve it the
value of x the X1 X2 X3 X4
correspondingly comes as X1 is
zero X X2 is 400 X3 is 2ou 266 and
23 and the value
of um the fourth one is zero so
technically the amount uh which
is obtained for
the revenues is 35,000 into 0 + 28,750
into 4 400 4 for the second variable X2
plus 3 36,000 into into 266 into 2/3
where 2 266 and 2/3 is the third
variable and 26,400 into 0 because the
4th variable variable is
zero the total functional value for the
revenues comes out to
be
21 2
2,1
110
lakhs now let us just double
check the constraints are being met so 4
400 + 266 and 2/3 is less than equal to
980
perfect then if I
have uh the second one so this is also
being met because the first variable X1
is
0o - 400 - 260 and 2/3 is less than
equal to 0 then similarly for the third
constant it is - 400 plus a value which
is um less than um 400 so- 400 and a
value less than 400 is less than equal
to zero and all of them if you see X1 X2
X3 X4 are between values
of so technically this value of 0 and0
should have been met but we are taking
we are trying to basically maximize the
functional value when I come to the cost
now cost Factor has a different
implication the cost Factor implication
is that we are not considering any
revenues we are only concentrating on
the Trying to minimize the cost so
technically if I consider
everything is a cost so that means
33,200 is the cost factor for producing
one unit or one acreage of the first
variable 30,500 for the second 37 35750
for the 3 and 27,000 for the 4th so
correspondingly the values which we
will put now take technically if I see
if all of them have to be produced at
minimum 100 so obviously the cost would
be
increasing in the theoretical sense
theoretical sense the cross are
increasing like
this I have to produce 100 acres for uh
rice so that would be 33
200 so there's a cost already so red
color me negative
even though I I did use red color in the
other combination but I didn't mention
whether it's a positive negative it is
here specifically negative similarly
for if I produce at least 100 for the
second variable on the
second product which was
wheat the cost will come out to be
this for the third one again 100
[Music]
minimum the cost cost will come out to
be 100 into
35750 and the last one which is X4 it
will come out to be
this so this is the total cost component
so considering if I don't produce
anything is is the minimum amount of
of cost I can have that means don't do
anything the moment you start producing
this is a cost Factor because I not
considered revenues as such so generally
we with a value of 0 0 0 0 for all X1 X2
X3
X4 the value comes out to be zero cost
Factor now consider
the net loss function so obviously you
can consider from the net loss function
and the net profit function why I'm
using the concept of net loss function
and the net profit function is like this
if you see the two equations which is
revenues and cost separately here
this Factor
so I'll
use red for negative value so these are
there are no sign here so actually there
are cost so these are
all negative
flow and if I
consider positive one as blue or let us
use the green
one so these are all
positive very
interestingly if I follow for the first
variable which is
rice if I take the values so X1 is
common if I take the
differences this will be positive I'm
using green color for
positive if I
consider the next variable which is
wheat this is negative because 30,500 is
a cost which is which is more than
28,750 which is the
revenues if I consider the
third which is bazra it is positive
because 36,000 is more than
35750 and finally if I consider the maze
which is the fourth variable here 27,000
which is corresponding the cost is more
than 26,400 which is the revenue so
these are
positive negative positive negative so
if I consider the profit
function and the r and the net profit
and the less loss function the signs
would be revers so this part positive
which is green becomes red red becomes
green B based on how you try to analyze
so obviously you'll try to maximize your
net profit and when you're considering
the net loss you'll try to basically
minimize your net loss
function so if I if I consider the net
loss as the word is written
here so it will be basically the loss
function minus the revenues of the cost
function minus the
revenues so it was basically positive
negative positive negative as given so
now this becomes net loss means the
first one rather than positive now
becomes
negative the second one rather than
positive um negative being
positive the third one rather than being
positive will be negative because
there's a negative sign
here and
finally the fourth value rather than
negative being
positive so if you consider that also in
any combination the
values would come out to be zero because
the moment you produce it will basically
be or
loss and furthermore all the constraints
are
met if I consider the net profit
function so the values are just reverse
the negative becomes positive hence I'm
putting a green
sign the positive becomes negative a
great
sign again a green
sign then a red sign and the
corresponding values for tro C the net
profit function is also
zero if we solve the maximization of
only revenues with the cost as a
constraint now this is the point if you
remember we had considered that for the
gold programming or the concept of
multiobjective problem the idea was this
this I'll write it here and then solve
the problem
accordingly
so in case you had the maximization
which I use the green
color Su
Max of
fub1 and some
mean of
G1 and the constraints were given as
sver example H1 I'm not writing X is
greater than equal to
B1 then H2 being equal to B2 and H3 is
less than equal to
B3 so here what we'll do is like this
we'll maximize F1
and bring
this this G1 function
which was the minimization one as a
constraint so the G1 function has to be
minimized so that means it cannot exceed
a value of C for example
B4 this B1 B2 B3 which are shown in
black colors are for the already
existing constraint this B4 which is
shown in red colors is basically the
socalled
value which cannot be exceeded
considering is a minimization problem so
if I consider a value of the cost say
for example zero consider zero it can be
consider that it can be other values
also it can be say for example loss has
to be less than equal to 200 or 500
rupees or 5,000 rupees So based on the
fact that I'm taking the
constraint of the minimization as in the
from the objective function brought into
minimization here the problem is
the revenue function is already there
I'll put a green Mark so these are X1 X2
X3 Revenue
functions the constraints are already
there I'm putting a violet
tickmark
and the minimization has been brought
back into the constraint I'll put a red
mark so it is
33,200 X1 30,500 X2
35750 X3 27,000 X4 and this is less than
zero so that means I'm considering this
as
B4
similarly for the less than
type so less than type is basically I'm
considering
this this so they are all less than
type and this being a combination of of
less than greater than because actually
if I consider any X X's so the actual
equation would be x i is greater than
100 is less than
400 so if I consider the concept of less
than they would go
here and if I consider this um so this
would basically become the greater than
type because X1 is more so this would go
into the greater than
function so this manipulation of greater
than less than equal to can be done very
easily now when I consider
this I solve it considering the
constraint is zero I get a value of zero
for all the X's which is fine the main
question which I have not solved here
but we will discuss later on for other
problems
is what
if the value Which is less than zero can
be considered at say for example less
than equal to
500 or less than equal to say for
example 5
lakhs and so on and so forth would we be
getting answers accordingly for that
that means a minimum loss has to be
there because due to some reason you
have to produce due to some policy
decision by the government then or you
have to basically think that you should
produce they can be different
reasons So based on that you'll consider
whether you want to maximize your
revenue and what is the minimum level of
loss you can
sustain now the question
is can this problem be considered where
you want to minimize the loss and
consider the constraint also include
includes the
maximization objective fun function
which was there so if I write
it it is
now
minimization of G1 because G1 was the
minimization as per the concept of the
nure you
using you already have the existing
three sets of
constraints which
at H1 is greater than equal to
B1 H2 is equal to
B2 and H3 is less than equal to
B3 and the objective function which was
there on the maximation type that comes
here it was F so it becomes fub1 is
greater than equal to sum
B4 because profit has to be Revenue has
to be more and and losses were less than
that B4 there which was shown in the red
color B4 so in this case the problems
would be which minimization would
basically bring the objective function
as minimization and the constants
accordingly now the question is that we
have not solved this for the case when
you are bringing the objective function
as maximization in one of the
constraints the question is that uh how
would you choose this red B4 and green
B4 depending on the problem formulation
I using the same symbol but I'm
mentioning the color scheme so if you
have watched the video you would
understand what I mean by red and green
red was basically corresponding to the
cost and green was corresponding to the
revenue now
technically you can solve this problem
and get an intuitive start
accordingly like I'll I'll mention and
we'll consider that later on like
solve the objective function first with
what we written that solve the
minimization problem as a single
objective function solve the
maximization problem at the single
objective function solve them get the
answer now when we come and and trying
to basically put the so-called
multiobjective case which we have
written
for the case when it
is Max of
fub1 constants X H1 greater than
B1 H2 = to
B2 H3 is greater than
B3
and
there's a part of
G1 is less than equal to B4 this
B4 which will circle with violet color
is the what not the exact value it would
be a relationship with the single
objective function which we have solved
so see for example in the single
objective function obviously in this
problem it was Zero the final value when
you're trying to solve the minimization
problem consider it was 2 lakhs minimum
value so we will consider some value of
more than 2 lakhs here in place of uh R
before solve the problem and try to
check that when you consider in the
constraint of one of this objective
function which was already there what is
the change in the first objective
function which is shown in green here
green means
maximization in the similar
way when you Trying to minimize the
cost minimization of the cost which is
G1 and is red in color and the objective
function which is initially which is the
maximization comes in the
constraints so this would
be
fub1 is now greater than equal to that
was less than equal to so here you have
already solved F1 initially separately
as I mentioned bring that uh objective
function maximization means it will be a
little bit less than what you maximum
you can obtain and solve this problem so
in one case constraints are same in one
case you have this is the minimization
problem I'm repeating it time and again
please bear with
me your
maximization
fub1
H1
B1
H2
B2
H3
B3
and
G1 being less than equal to
B4 so the main idea is if you
consider
socalled problem one and problem
two which I'm writing case one and case
two there's a subtle difference between
them how they look like one is a
maximization another minimization
considering that we are including in the
constraints the other part of the object
functions
accordingly now
we solve the problem
considering which I mentioned when we
solve the problem considering the cost
minimization with the revenue in the
constraint so this what I was talking
about this I spend a lot of time in
discussing in the last slide so which
will highlight
here so
here is a minimization one so I'm
putting a
red underlying for this
the constraints are all
same so these are the constraints the
constraint existing one only the new one
which is being brought is basically the
value
of the objective function which was
maximization which will Mark in
green only thing
is this value which is
21 lakhs that can can be changed per TB
and based on that I can basically find
out what is the minimization of the cost
which can be obtained so rather than
having 21 lakhs it could have been some
value of
21
lakhs minus a Delta
value so Delta can be very large here
just Delta I'm using as a symbol and
based on that I solve when I solve it
the values here it is 0 0 for X1 X4
that's a different question so but the
values which we get it is basically 400
and
266 and 2/3 for the values of X2 and X3
so if you remember when we solving the
problem for the maximization problem we
got that based on then that when we put
the values accordingly we get the
corresponding output as
217 that means is equal to basically 217
L
33,333 and
1/3 now the question
is before
we wrap up this class the question is
what happens if you have three objective
functions already Max Max Min something
like that so let me
put I'll only consider one already
existing constraint which I'll use the
red color which I'll use the the black
color sorry
so you
have Max of
fub1 Max of
FS2 mean
of
G1 constraint was
H1 greater than B1 H2 less than
B2 so this is equal to to
sorry = to
H3 less than B3 so this was the problem
so what we can do is like this solve
them separately like we
solve Max F1 as a single problem without
considering Max FS2 and Min G G2 then we
solve Max F2 without considering Max
fub1 and Max Min G2 and we solve
separately Min G1 without considering
Max F1 and Max F2 constraints remain
same for all this statement which I made
just now when we want to solve the soall
multiobjective
case it will become like this the
objective function first remains as it
is
and we bring FS2
being greater than equal to B4 B4 is a
value which you the idea which you can
get once we have solved case two which
we I just wrote down and I
erased and this case G1 would be less
than equal
to so this values less than equal to are
the minimum values for G1 and max value
for B4 this is B5 and this we have got
the idea after solving case 3 which I
have just erased and similarly we can do
the combinations accordingly I'll solve
the problems
accordingly and with this I'll end this
uh class which is the 32nd class have a
nice day and thank you very
[Music]
much
[Music]
[Music]
