Consider all triples of real numbers ( x , y , z ) such that x , y , z ≥ 2 1 and x + y + z = 3 .
If the minimum and maximum values of x x y y z z are A and B , respectively, what is A + B ?
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
My solution is almost the same! . Bernoulli's and Jensen's inequality are my favourites
@rohit kumar nice problem!
Log in to reply
i can't even begin to understand this problem's solution ! and never heard of Bernoulli's and Jensen's inequality !
Brilliant ! And considering you wrote the solution of phone, it's pretty neat too. By the way, how did come up with f (x, y, z) < f (x+y-0.5,0.5, z)
Log in to reply
BTW How was workshop?
Log in to reply
Easy. One of the questions used integration(that was not one of easy ones. Only Divij could solve it.)rest was easy.
Log in to reply
@Rohit Kumar – That is expected no exception
@Rohit Kumar – And do you know how much are XO1 People are scoring in AITS (Advanced)
Log in to reply
@Prakhar Bindal – No one is scoring more than you. Piyush is 30 marks less than you.
Log in to reply
@Rohit Kumar – Ohh cool! .I Am very excited for the KVPY Camp now
@Rohit Kumar – @Rohit Kumar . What happened . you coming to Kolkata?
Log in to reply
@Prakhar Bindal – Most probably yes.if tomorrow's train is not more than 7 hours late then yes.
Log in to reply
@Rohit Kumar – Cool . it wont be late !
Log in to reply
@Prakhar Bindal – I do hope so. By the way I've turned turned off email notifications for brilliant. Ii might be in convenient for you to get a reply 25 to 30 mins later
I had a ton of difficulty understanding what you are doing.
I suspect that when showing the maximum, you used a change of variables, but didn't change the notation. I think you have "your x" = "problem x - 2 1 ".
Can you clean up the solution?
Let u = x x y y z z . We want to find when u attains its minimum and maximum values.
Check the edge of the domain: x = y = 2 1 , z = 2 . Then u = ( 1 / 2 ) 2 ⋅ 2 2 = 2 .
Also on the edge of the domain: x = 2 1 , y > 2 1 and z = 2 2 1 − y . View u as a function of y , we have ln u = 2 1 ln 2 1 + y ln y + z ln z , d ln u = ( 1 + ln y ) d y + ( 1 + ln z ) d z ; since d z = − d y , d ln u = ( 1 + ln y − 1 − ln z ) d y = ( ln y − ln z ) d y = ln z y d y , which is zero if numerator and denominator are equal, i.e. y = z = 1 4 1 . Then u = 5 5 / 2 1 1 = 3 1 2 5 / 2 0 4 8 , which lies between 1 and 2.
With x , y > 2 1 and z = 3 − x − y , we view u as a function of x and y : ln u = x ln x + y ln y + z ln z , d ln u = ( 1 + ln x ) d x + ( 1 + ln y ) d y + ( 1 + ln z ) d z ; with d z = − d x − d y : d ln u = ( ln x − ln z ) d x + ( ln y − ln z ) d y ; at an extremum, both coefficients of d x and d y are zero, implying x = y = z = 1 and u = 1 .
Thus we find that the minimum value of u is A = 1 and the maximum value of u is B = 2 . The total is 3 .
I got a few questions:
Intuitively, I know that we should check the edge/boundaries of the domain to determine whether a min/max exists or not. But why is that so? All I can come up with are some handwaving argument.
In your second last line, you said "at an extremum, both..." , you have only shown that an extremal point is u = 1 . But how did you know it's a minimal point in the first place? Don't you have to finish it off via Hessian matrix ?
Log in to reply
Both the domain and its boundary are compact sets, and u is a continuous function. Therefore the values of u are bounded. Thus there exist a minimum and a maximum, and the only possible candidates for minimum and maximum are the three extreme values I calculated. The smallest of these values is 1 , which must therefore be the minimum.
Saying this differently, if 1 were not the minimum value, there should exist a point ( x , y , z ) for which u ( x , y , z ) = A < 1 , and that point should have shown up in our analysis of extreme values.
Log in to reply
You're awesome!!
So how do we solve this problem using your method?
WLOG, x ≤ y ≤ z
Case 1: Boundary on x = 0 , then we're left we 2 variables, and we get desired value (of function = x^5 y + y^5 z + z^5 x) of 5^5/6^6.
Case 2: Boundary on z = 1 , then we got desired value of 0.
Case 3: x = 0 and z = 1 <<< how do carry on from here?
u = x^5 y + y^5 z + z^5 x
==> du = ????
Log in to reply
@Pi Han Goh – For case 3, let z = 1 − x − y so that d z = − d x − d y . Then u can be viewed as a function of x and y , and d u = ( 5 x 4 y − y 5 − 5 z 4 x + z 5 ) d x + ( x 5 + 5 y 4 z − y 5 − 5 z 4 x ) d y . To find the extreme values, equate both coefficients to zero, so that simultaneously 5 x 4 y − y 5 − 5 z 4 x + z 5 = 0 ; x 5 + 5 y 4 z − y 5 − 5 z 4 x = 0 . Solve this e.g. by equation the two, ( 5 y − x ) x 4 = ( 5 y 4 − z 4 ) z , etc.
Log in to reply
@Arjen Vreugdenhil – Ah of course. For some reason, I couldn't get started on this other problem.
Thanks again!
Problem Loading...
Note Loading...
Set Loading...
Minimum
Let f ( x , y , z ) = x x y y z z . Let g ( x ) = x ln x .
We can show that g ( x ) is a convex function by considering its second derivative.
By jensen's inequality ,
l n ( f ( x , y , z ) ) = g ( x ) + g ( y ) + g ( z ) ≥ 0
Hence, f ( x , y , z ) ≥ e 0 = 1 .
Maximum
First, we show that f ( x , y , z ) ≤ f ( x + y − 0 . 5 , 0 . 5 , z ) . This is equivalent to proving
( a + b + 2 1 ) ( a + b + 2 1 ) ≥ 2 ( ( a + 2 1 ) ( a + 2 1 ) ( b + 2 1 ) ( b + 2 1 ) )
For nonnegative a , b
Now, assume WLOG a ≥ b ≥ 0 . Since b ≥ 0
( a + b + 2 1 ) ( a + b + 2 1 ) = ( a + b + 2 1 ) a ( a + b + 2 1 ) b ( a + b + 2 1 ) 2 1 ≥ ( a + 2 1 ) a ( a + b + 2 1 ) b ( a + b + 2 1 ) 2 1
We just then need to prove that
( a + b + 2 1 ) ( b + 2 1 ) ≥ 2 ( ( a + 2 1 ) ( 2 1 ) ( b + 2 1 ) ( b + 2 1 ) )
Or that
2 b + 1 ( 1 + b + 0 . 5 a ) b ( 1 + a + 0 . 5 b ) 0 . 5 ≥ 1
Which was attained by dividing the LHS by the RHS. But then, by Bernoulli's inequality, ( 1 + a ) n ≥ 1 + n a , we get
2 b + 1 ( 1 + b + 0 . 5 a ) b ( 1 + a + 0 . 5 b ) 0 . 5 ≥ 2 b + 1 ( 1 + 2 b + 1 2 a b ) ( 1 + 2 b + 1 b )
Now we need to prove
( 2 a + 1 ) ( 2 b + 1 ) 1 . 5 b ( 2 a + 2 + b 1 ) ( 2 a + 1 + y ) ≥ 1 .
Since equality occurs when b = 0 and for all positive b, the LHS is an increasing function in a, we can say that the minimum of the LHS occus when a is zero, and thus we just need to prove that
2 b + 1 b + 1 ≥ 1
Which is true for all nonnegative b. (One can prove it by Bernoulli's inequality!) Thus the original inequality is proven, and equality occurs when b = 0 .
Thus,
f ( x , y , z ) ≤ f ( x + y − 0 . 5 , 0 . 5 , z ) ≤ f ( x + y + z − 1 , 0 . 5 , 0 . 5 ) = 2 . So the maximum is 2 and the minimum is 1 which gives our answer to be 1 + 2 = 3 .