Complex System of Equations

Algebra Level 5

For how many ordered pairs ( a , b ) (a,b) of complex numbers does the system of equations { x 2 y x = a x y 2 + y = b \begin{cases} x^2y-x=a\\ xy^2+y=b \end{cases} have exactly one complex number solution?

Details and assumptions

A complex number solution is an ordered pair of complex numbers ( x , y ) (x,y) that satisfies both equations.

Clarification: a , b , x a, b, x and y y are all complex numbers. As an explicit example, ( x , y ) = ( i , i ) (x,y) = (i, i) is a solution to ( a , b ) = ( 2 i , 0 ) (a,b) = (-2i, 0 ) .


The answer is 1.

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.

5 solutions

Laura G
May 20, 2014

First consider the cases a=0 or b=0.

If a=0, the first equation is x(xy-1)=0, which has 2 solutions: x=0 or xy=1. With x=0, second equation is y=b. With xy=1, second equation is 2y=b. The only way that the equation has only one solution is that b=0 (the solutions are (0,b) and (0,b/2)).

So, there is a pair (0,0), which has only one solution.

When b=0, second equation is y(xy+1)=0. If y=0, then -x=a. If xy=-1, then -2x=a. Again, the only possibility which has a unique solution is a=0.

So, we can consider that a and b are not 0. In this case, x and y are also different from 0. I can write y=k/x. Then the equations are:

kx-x=a k 2 / x + k / x = b k^2/x+k/x=b

From the second equation, x = k ( k + 1 ) b x=\frac{k(k+1)}{b} , and putting it on the first equation:

(k-1)k(k+1)=ab.

For each k, we have a x and a y that are solutions to the system of equations. But this last equation has 3 different complex solutions, so the system has 3 solutions.

Then, the only pair of parameters that have only one solution is (0,0).

All known correct solutions involve the cubic equation on the product x y xy . While it is tempting to use the theory of quadratic equations and discriminants, for x x with given y y and vice versa, it really cannot be used to prove non-uniqueness of the solution for ( a , b ) ( 0 , 0 ) . (a,b)\neq(0,0). There would be two such equations, and one having multiple roots x x for a fixed y y does not guarantee that there is more than one solution to the system.

Calvin Lin Staff - 7 years ago

Log in to reply

Considering this is now untrue, perhaps you should delete this comment. I added my solution as a comment just below your solution. Thanks.

James Wilson - 3 years, 8 months ago
Calvin Lin Staff
Aug 23, 2017

[This solution is essentially identical to Laura's. It explains how I was approaching it.]

Note that the system can be rewritten as { x ( x y 1 ) = a y ( x y + 1 ) = b \begin{cases} x(xy-1)=a\\ y(xy+1)=b \end{cases} If x y = t , xy=t, then multiplying the two equations, we obtain t ( t 1 ) ( t + 1 ) = a b , t(t-1)(t+1)=ab, which can be rewritten as t 3 t a b = 0. t^3-t-ab=0. Conversely, if t t is a root of this equation, not equal to 1 1 or 1 -1 , then (with some work) x = a t 1 x=\frac{a}{t-1} and y = b t + 1 y=\frac{b}{t+1} gives a solution to the original system of equations.
Note: Because we manipulated the system of (non-linear) equations, it is not guaranteed that any final solution will work. For example, for a given t t , we likely do not have x = y = t x = y = \sqrt{t} as a solution.

For any a a and b b , by the Fundamental Theorem of Algebra, the polynomial t 3 t a b t^3-t-ab has three roots, if counted with multiplicity. If a b 0 , ab\neq 0, then these roots cannot be 0 0 , 1 1 or 1. -1. Also, by Vieta's formula, the sum of the roots is zero, so, because 0 0 is not a root, the three roots cannot be all the same. Therefore, there are at least two distinct values of t t that satisfy the equation t 3 t a b = 0 , t^3-t-ab=0, and, as explained above, each produces a solution of the original system. Note that these solutions cannot be the same because they have a different value of x y . xy. Hence a b = 0 ab = 0 , in order for there to be a unique solution.

If a = 0 a=0 and b 0 , b\neq 0, then both ( 0 , b ) (0,b) and ( 2 b , b 2 ) (\frac{2}{b},\frac{b}{2}) are solutions of the system. Similarly, if a 0 a\neq 0 and b = 0 b=0 , then ( a , 0 ) (-a,0) and ( a 2 , 2 a ) (-\frac{a}{2}, \frac{2}{a}) are solutions.

Finally, if a = b = 0 , a=b=0, then note that x y 1 xy-1 and x y + 1 xy+1 cannot be both zero. So at least one of the variables x x and y y is 0 0 . Therefore x y = 0 , xy=0, so the system implies that both x x and y y are equal to zero.

This shows the system has exactly one solution if and only if ( a , b ) = ( 0 , 0 ) (a,b)=(0,0) , implying the answer is 1. 1.


This is an example of a map from the affine complex plane to itself which is quasi-finite, but not finite.

Very fun tricky problem. Here is my solution. I decided to post up higher so people could see it. Multiply by y y in the first equation and by x x in second equation, and subtract them to obtain 2 x y = b x a y 2xy=bx-ay , or, equivalently, y = b x a + 2 x y=\frac{bx}{a+2x} , assuming a + 2 x 0 a+2x\neq 0 . It will be handy to note at this point that the function y = b x a + 2 x y=\frac{bx}{a+2x} is one-to-one, for a + 2 x 0 a+2x\neq 0 , iff a , b 0 a,b\neq 0 . (I'll leave it to the reader to verify additionally that if a + 2 x = 0 a+2x=0 , then a = 0 a=0 or b = 0 b=0 ). Substitute this into equation 1, and, after some algebraic manipulation, we get b x 3 2 x 2 3 a x a 2 = 0 bx^3-2x^2-3ax-a^2=0 . Next note the fact that a , b 0 a,b \neq 0 \Rightarrow If this cubic polynomial equation does not have a unique solution, then, because the aforementioned function is one-to-one, there will be multiple solutions to the system. Assume a , b 0 a,b \neq 0 . Since the system has a unique solution, the cubic polynomial equation has a unique solution, and therefore, assumes the factored form: b ( x + r ) 3 b(x+r)^3 . This leads to the system of equations: 3 b r = 2 , 3 b r 2 = 3 a , b r 3 = a 2 3br=-2,3br^2=-3a,br^3=-a^2 (justified by the fact that two polynomials with the same roots and leading coefficient are identical). Dividing the first two equations and then the second two equations yields r = 3 2 a = a r=\frac{3}{2}a=a , which implies a = 0 a=0 . Analyzing the logic of the conclusions (retaining the assumption that the solution to the system is unique), we get ( a , b 0 a,b\neq 0\Rightarrow a , b 0 a,b \neq 0 and the solution to the cubic polynomial equation is unique) and ( a , b 0 a,b \neq 0 and the solution to the cubic polynomial equation is unique \Rightarrow a = 0 a=0 ). By applying transitivity, and then simplifying, one can conclude a = 0 a=0 or b = 0 b=0 . From there it can be reasoned that a a and b b both must be zero or the solution will not be unique (as was done in Calvin's and Laura's solution).

James Wilson - 3 years, 8 months ago
Steve Guo
Dec 17, 2013

Let us assume that x x and y y are both non-zero. For the first equation, we will treat y y as a constant in a quadratic of x x and vice-versa for the second. Note that for any y y or x x , the first and second equations have at least one solution in complex numbers for x x and y y , respectively. Since we are looking for pairs a , b a,b that yield exactly one solution, this implies that each equation has exactly 1 solution. Using the quadratic formula on both equations, we get

x = 1 ± 1 + 4 a y 2 y , y = 1 ± 1 + 4 b x 2 x x=\frac{1\pm \sqrt{1+4ay}}{2y} , y=\frac{-1\pm \sqrt{1+4bx}}{2x}

Since there is exactly one solution, the discriminants are zero, so ignoring the discriminants we have x = 1 2 y x=\frac{1}{2y} and y = 1 2 x y=\frac{-1}{2x} , which is impossible. Thus at least one of x , y x, y must be zero.

Say x = 0 x=0 . It immediately follows from the first equation that a = 0 a=0 . We wish to find constants b b such that the system

x 2 y x = 0 x^2y-x=0 y 2 x + y = b y^2x+y=b

has only one solution for ( x , y ) (x,y) . Factoring out an x x from the first equation, we get x ( x y 1 ) = 0 x(xy-1)=0 . Therefore either x = 0 x=0 or x y 1 = 0 xy-1=0 . If x = 0 x=0 , then plugging that in to the second equation yields y = b y=b . If x y 1 = 0 xy-1=0 , then we have that x y 2 + y = y ( x y + 1 ) = y 2 = b xy^2+y=y(xy+1)=y \cdot 2=b . Since there can only be one solution, we can conclude that the two solutions generated are the same, and therefore b = 2 y = y y = 0 , b = 0 b=2y=y \Rightarrow y=0, b=0 .

Now say y = 0 y=0 , and it follows that b = 0 b=0 . We can proceed similarly to above and get that a = 2 x = x x = 0 , a = 0 a=-2x=-x \Rightarrow x=0, a=0 . However, we have already counted a = 0 , b = 0 a=0, b=0 , so this is redundant.

We thus find that the only solution is a = 0 , b = 0 a=0, b=0 and therefore the answer is 1 \boxed{1} .

In the first case, when x = 0 x=0 , how is it possible that x y 1 = 0 xy-1=0 , because plugging in x = 0 x=0 leads to 1 = 0 1=0 , which is false?

Am I missing something?

minimario minimario - 7 years, 5 months ago

Log in to reply

Remember that we are finding ( a , b ) (a,b) such that there is only one solution for ( x , y ) (x,y) , not vice-versa. We know that if this system has only one solution, then we either have x = 0 , a = 0 x=0, a=0 and/or y = 0 , b = 0 y=0, b=0 . However, we are not guaranteed that this system has only one solution - therefore we must find b b such that it does. When we let a = 0 a=0 , we are not assuming that x x must be zero as we are trying to find all solutions of the system and set b b such that there is only one.

As I know I am terrible at explaining things, if we let a , b a,b be, say, ( 0 , 4 ) (0,4) , then we have the following system of equations: x 2 y x = 0 x^2y-x=0 y 2 x + y = 4 y^2x+y=4 The solutions to this system are x = 0 , y = 4 x=0, y=4 and x = 1 2 , y = 2 x=\frac{1}{2}, y=2 . Notice that there is a solution where x x is not zero. We claimed that x x or y y must be 0 in order for the system to have one solution, but that does not mean that it will have one solution. The rest of the solution proves that b b must be 0 in order for the system to have exactly one solution.

Sorry for any confusion! I should have made the transition clearer.

Steve Guo - 7 years, 5 months ago

(Edit: This is completely wrong. I added a correct solution as a comment below Calvin Lin's solution.) Here's a more complete explanation for a solution based on Steve's. Since the quadratic formula doesn't work under the assumption of a unique solution (it gives an inconsistent pair of equations), as Steve showed, then we are left with only the possibility that equation 1 is not quadratic in x or equation 2 is not quadratic in y (i.e., the leading coefficient is zero of at least one). Therefore, either y = 0 because, in equation 1, y is the leading coefficient of the quadratic polynomial in x; or x = 0 because, in equation 2, x is the leading coefficient of the quadratic polynomial in y. So we know that the solution must either look like (x, 0) or (0, y). Case 1 (y = 0): If y = 0, then b = 0 and x = -a, producing a solution. Note that any other solution must take the form (0, y) (since we know (-a, 0) is the only solution of the form (x, 0)). So we must check for solutions of the form (0, y). When we substitute x = 0, we get a = 0 and a solution of (0, 0). Therefore, this must be the only solution. Hence, a = b = 0. Case 2 (x = 0): if x = 0, then, a = 0, and we can do a similar process, to show b = 0. Fin.

James Wilson - 3 years, 9 months ago

Log in to reply

Not quite. You seem to be assuming that " y y is a polynomial function of x x ", which isn't obvious to me. This leaves out the possibility (amongst others) that " y y is a rational function of x x ",

For example, applying the quadratic formula to the 2nd equation, we get that y = 1 ± 1 4 x b 2 x y = \frac { - 1 \pm \sqrt{ 1 - 4xb }} { 2x } . Now, with this substitution back into the first equation, we end up with x × 1 ± 1 4 x b 2 x = a x \times \frac { - 1 \pm \sqrt{ 1 - 4xb }} { 2 } - x = a . Naively, if we square and things cancel out, we have a cubic equation, and possibly up to 3 roots. However, we still have to verify that we didn't introduce extraneous roots, and so it (might be) possible that we end up with 1 solution to the original equation.

Be very careful that you have to treat ( a , b ) (a,b) as fixed, instead of treating y y as fixed (e.g. by setting y = 0 y = 0 ), since there could be other solutions ( x , y ) (x,y) .

Calvin Lin Staff - 3 years, 9 months ago

Log in to reply

I meant that the coefficient of the x^2 term had to be zero, because otherwise the pair of equations produced is inconsistent. If the coefficient of the x^2 term is not zero you will surely get an inconsistent system. It's irrelevant what other relationship we specify between x and y

ax^2 + bx + c = 0 and a ~= zero is equivalent to x = (-b+/-sqrt(b^2-4ac))/(2a). Therefore, x = (1+/-sqrt(1+4ay))/(2y) is equivalent to yx^2 - x - a = 0 and y ~= 0. Therefore, whenever x and y aren't zero, assuming the unique solution criterion, we get 2xy = 1 = -1. So from there you can conclude either x or y is zero

a, b, and c are free variables in the original expression, so I can just let a = y, b = -1, and c = -a (ignoring the fact that I used a twice by mistake).


I wasn't trying to be disrespectful. It was not my intention to upset you. But I do actually know what I'm talking about with this one... (edit: I was indeed WRONG. continue reading to see a new solution)

Your solutions are always better than mine, and I can tell your abilities and achievements are well above mine. So I'm not trying to prove I'm smarter or anything. I love your favorite quotation by the way.

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson Not upset. Just too busy with other stuff to respond. Still too busy to respond, and hoping to respond by tomorrow.

Calvin Lin Staff - 3 years, 9 months ago

Log in to reply

@Calvin Lin Okay not a problem!

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson (FYI I combined your comments so that it is easier for others to see the train of thought. In future, you can just edit the first / main comment directly.)

I meant that the coefficient of the x^2 term had to be zero, because otherwise the pair of equations produced is inconsistent.

If I'm reading you right, what you are claiming generally is that if f ( x , y ) f(x,y) is a quadratic in x x and linear in y y , and g ( x , y ) g(x,y) is linear in x x and quadratic in y y , and ( f ( 0 , 0 ) , g ( 0 , 0 ) ) ( a , b ) \left( f(0, 0), g(0,0) \right) \neq (a,b) , then f ( x , y ) = a , g ( x , y ) = b f(x,y) = a , g (x,y) = b would have multiple solutions in the complex numbers. If so, then that claim is incorrect. For example, the system of equations ( x + 1 ) 2 ( y + 1 ) ( x + 1 ) = 0 , ( x + 1 ) ( y + 1 ) 2 + y = 0 (x+1)^2(y+1) - (x+1) = 0, (x+1) (y+1)^2+y = 0 satisfies the above conditions, but has a unique solution ( 1 , 1 ) (-1,-1) .

If the claim has additional conditions, please make it clear what else I'm missing.
If I'm reading you wrong, please elaborate further.

Therefore, whenever x and y aren't zero, assuming the unique solution criterion, we get 2xy = 1 = -1.

It's not immediately clear to me how you reached this conclusion. Note that the contradiction should have been x y = 1 = 1 xy = 1 = - 1 , which is the key observation that the solutions make.

Here is an instance where it's important to track the implication signs of your statements. Whenever you manipulate a system of equations, you run the risk of introducing extraneous solutions. This doesn't happen with linear equations and so many people are not aware of this potential mistake. As an explicit contrived example, if we wanted to solve x 2 x 1 = 0 , x + 1 = 0 x^2 - x - 1 = 0 , x + 1 = 0 , we cannot just add them up and say that x 2 = 0 x^2 = 0 hence x = 0 x = 0 . In this case, the original system has no solutions, and we introduced an extraneous solution when manipulating the system. We have the forward implication (obviously) but do not have the backward implication. (Food for thought: Why isn't this an issue with linear equations?)

This is exactly what happens here. If you look at my solution, even with the value of t = x y t = xy , we do not yet know if there are explicit solutions for x x and y y that satisfy the original system. In particular, if t = 1 t = 1 (or -1), then the solution set is x = a t 1 , y = b t + 1 x = \frac{a}{ t-1}, y = \frac{ b}{ t+1} which is rejected. Hence, we do not get the contradiction of "there exist multiple solutions that satisfy x 1 y 1 = 1 x_1y_ 1 = 1 and x 2 y 2 = x_2 y_2 = (something else)", because there is no ( x 1 , y 1 ) (x_1, y_1) solution.

Note: In general, avoid having your notation do double duty. You may mentally know the difference, but it isn't easily conveyed. Can you clean up the expression? An easy fix is to use capital letters.
I'm guessing that ~= is \neq ? If so, the more common expression is !=. Alternatively, you could use the Latex code of "\neq".

Calvin Lin Staff - 3 years, 9 months ago

Log in to reply

@Calvin Lin (Edit: this is not the correct solution. Continue through the replies to see it.) Ok, allow me to explain it in more detail. To start, assume "a" and "b" are such that the solution (x, y) to the system is unique. Applying the quadratic formula to the two equations (noting that the leading coefficient cannot be zero in order for the formula to be applied), we get the result y 0 x = 1 ± 1 + 4 a y 2 y y \neq 0 \Rightarrow x = \frac{1\pm \sqrt{1+4ay}}{2y} and x 0 y = 1 ± 1 + 4 b x 2 x x \neq 0 \Rightarrow y = \frac{-1 \pm \sqrt{1+4bx}}{2x} . Because the solution is unique, we know the discriminant is zero in both equations. So we are left with x = 1 2 y \ x = \frac{1}{2y} and y = 1 2 x \ y = \frac{-1}{2x} , which result in 2 x y = 1 2xy = 1 and 2 x y = 1 2xy = -1 . These equations are clearly inconsistent, so we are left to conclude that the assumption " x 0 x \neq 0 and y 0 y \neq 0 " was false, since we are retaining the assumption of a unique solution. Therefore, it is necessary that either x = 0 x = 0 or y = 0 y = 0 for any given solution to the system. (Note that, at this point, if another contradiction arose, then the assumption that "a" and "b" are such that the solution of the system is unique is false, from which it will follow from A introduction that there are no such "a" and "b".) Now I consider the two cases: "the system's solution has x = 0 x = 0 " or the system's solution has y = 0 y = 0 ". Case 1 (the solution takes the form (x, 0)): Upon substituting y = 0 into the system, we readily obtain b = 0 and x = -a, producing a solution (-a, 0), clearly, the only solution of the form (x, 0). At this point we must ensure that there are no other solutions to the system to show there are no contradictions with the original assumption. Any other solution must take the form (0, y) (because at least one of x or y must be zero, as determined earlier). Substituting 0 for x in equation 2 leaves us with y = 0. So the only solution is (0, 0), which implies a = b = 0. However, if you fancy, here is an alternate approach. Assume, bwoc, a 0 a \neq 0 . Then, reexamining the system, noting that no other solutions can have y = 0 y=0 , we can divide by it in equation 2, obtaining xy = -1. Substituting into equation 1, we obtain x = -a/2, from which it follows that y = 2/a. This contradicts the fact that the solution is unique. Therefore, a = 0. Hence, by necessity a = b = 0. I leave case 2 for the reader.

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson In this case, I think you are assuming that "A solution to the first equation is also a solution to the second equation".

While it is true that x = 1 ± 1 + 4 a y 2 y x = \frac{ 1 \pm \sqrt{ 1 + 4ay } } { 2y } satisfies the first equation, it could be that only x = 1 1 + 4 a y 2 y x = \frac{ 1 - \sqrt{ 1 + 4ay } } { 2y} satisfies the second equation. If so, that gives us a unique solution to the system of equations, without requiring x y = 0 xy = 0 .

(Of course, the claim of the question is that except when ( a , b ) = ( 0 , 0 ) (a,b) = (0,0) , then either both or neither solutions are satisfied in the second equation. However, using this line of reasoning, I do not yet see why that is true.)

Calvin Lin Staff - 3 years, 9 months ago

Log in to reply

@Calvin Lin Ah, I see. When you put it that way, I see the problem. I guess I need to fill that gap then.

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson I appreciate you pointing that out. I'm going to work on it whenever I can.

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson I was wrong, and I've got to start being more careful because I really don't want that to keep happening. I hate being a fool. But I can laugh at my mistakes. If you make all these problems up, I am impressed! I tried to find a patch for the last proof, but I realized it couldn't be done, because the assumption that the discriminants are zero is totally erroneous! i.e., they are NOT zero! Now I understand what you were saying. So everything I did before was pretty much total trash. I feel stupid for not noticing that before. So here's the new solution I came up with. It turned out to be pretty similar to your solution. Multiply by y y in the first equation and by x x in second equation, and subtract them to obtain 2 x y = b x a y 2xy=bx-ay , or, equivalently, y = b x a + 2 x y=\frac{bx}{a+2x} , assuming a + 2 x 0 a+2x\neq 0 . It will be handy to note at this point that the function y = b x a + 2 x y=\frac{bx}{a+2x} is one-to-one for a + 2 x 0 a+2x\neq 0 iff a , b 0 a,b\neq 0 . (I'll leave it to the reader to verify additionally that if a + 2 x = 0 a+2x=0 , then a = 0 a=0 or b = 0 b=0 ). Substitute this into equation 1, and, after some algebraic manipulation, we get b x 3 2 x 2 3 a x a 2 = 0 bx^3-2x^2-3ax-a^2=0 . Next note the fact that a , b 0 a,b \neq 0 \Rightarrow If this cubic polynomial equation does not have a unique solution, then, because the aforementioned function is one-to-one, there will be multiple solutions to the system. Assume a , b 0 a,b \neq 0 . Since the system has a unique solution, the cubic polynomial equation has a unique solution, and therefore, assumes the factored form: b ( x + r ) 3 b(x+r)^3 . This leads to the system of equations: 3 b r = 2 , 3 b r 2 = 3 a , b r 3 = a 2 3br=-2,3br^2=-3a,br^3=-a^2 (justified by the fact that two polynomials with the same roots and leading coefficient are identical). Dividing the first two equations and then the second two equations yields r = 3 2 a = a r=\frac{3}{2}a=a , which implies a = 0 a=0 . Analyzing the logic of the conclusions (retaining the assumption that the solution to the system is unique), we get ( a , b 0 a,b\neq 0\Rightarrow a , b 0 a,b \neq 0 and the solution to the cubic polynomial equation is unique) and ( a , b 0 a,b \neq 0 and the solution to the cubic polynomial equation is unique \Rightarrow a = 0 a=0 ). By applying transitivity, and then simplifying, one can conclude a = 0 a=0 or b = 0 b=0 . From there it can be reasoned that a a and b b both must be zero or the solution will not be unique (as was done in previous solutions).

James Wilson - 3 years, 9 months ago

Log in to reply

@James Wilson No worries. I can guarantee you I have made more mistakes (in an absolute number) on this site than you have. Several of my comments are peppered with "Ah I see ...", in part because I make quick evaluations / try to do math when not fully awake.

Ah, very nice solution.

Calvin Lin Staff - 3 years, 9 months ago

@James Wilson Ok now I think all the details have been filled in correctly. See you around Mr. Calvin.

James Wilson - 3 years, 9 months ago

@Calvin Lin I can't do it right now because I am spending family time, but probably by Monday.

James Wilson - 3 years, 9 months ago

@Calvin Lin Thanks for putting my ideas together. This new explanation is better though. I have to go for quite a while. So I won't be able to respond right away.

James Wilson - 3 years, 9 months ago

First we multiply the first equation for y and second equation for y: [; \mathbf{ x^{2}y^{2} - xy = ay } ;] and [; \mathbf { x^{2}y^{2} + xy = bx } ;]

Now we add the first equation to second equation and then we subtract the first from the second equation: [; \mathbf {2xy = bx - ay} ;] and [; \mathbf {x^{2}y^{2} = ay + bx} ;]

Get the square of the first equation and set it equal the second equation multiplied by a factor 4: [; \mathbf {b^{2}x^{2} + a^{2}y^{2} - 2abxy = 4ay + 4bx} ;] [; \mathbf {x^{2}b^{2} - 2x \times (aby - 2b) + a^{2}y^{2} - 4ay = 0} ;]

For [; \mathbf {a \neq 0} ;], we find 2 solutions for y. For [; \mathbf {a = 0} ;] and [; \mathbf {b \neq 0} ;], we find 2 solutions for x.

For [; \mathbf {a = b = 0} ;], we find one solution.

Therefore, the complex system have one solution satisfying the predicted conditions.

"For [; \mathbf {a \neq 0} ;], we find 2 solutions for y." I do not see why ths is true. Everything after that is equally true but not justified.

Calvin Lin Staff - 7 years ago
Raymond Lin
May 20, 2014

There are really 3 cases that we must deal with in this problem: (1) x 0 x \neq 0 and y 0 y \neq 0 , (2) x = 0 x = 0 and y 0 y \neq 0 or x 0 x \neq 0 and y = 0 y = 0 , and (3) x = 0 x = 0 and y = 0 y = 0 .

Case 1: x 0 x \neq 0 and y 0 y \neq 0

We can re-write the two original equations as y x 2 x a = 0 yx^2-x-a=0 and x y 2 + y b = 0 xy^2+y-b=0 .

Since x 0 x \neq 0 and y 0 y \neq 0 and the coefficients of the quadratic equations are complex, the solutions to the equations can be given as x = 1 ± 1 + 4 a y 2 y x = \frac{1 \pm \sqrt{1+4ay}}{2y} and y = 1 ± 1 + 4 b x 2 x y = \frac{-1 \pm \sqrt{1+4bx}}{2x} .

For x and y to only have one solution, the discriminants must be 0, so x = 1 2 y x=\frac{1}{2y} and y = 1 2 x y=\frac{-1}{2x} , implying that 2 x y = 1 2xy=1 and 2 x y = 1 2xy=-1 .

But 1 1 1 \neq -1 , so the discriminants cannot be 0. Therefore, when x 0 x \neq 0 and y 0 y \neq 0 , there is not only one complex solution to the system.

Case 2: x = 0 x = 0 and y 0 y \neq 0 or x 0 x \neq 0 and y = 0 y = 0

When x = 0 x = 0 and y 0 y \neq 0 , we can substitute x into the original equation x 2 y x = a x^2y-x=a to get that a = 0 a=0 .

By rearranging x 2 y x = 0 x^2y-x=0 , we get that y = 1 x y=\frac{1}{x} .

Plugging this into the original equation x y 2 + y = b xy^2+y=b , we get that b = y b=y . This appears to imply that b can be any complex number. Let us call such an arbitrary complex number c.

We can solve the system of equations x 2 y x = 0 x^2y-x=0 and x y 2 + y = c xy^2+y=c to check if there is only one complex solution. We factor the first equation to become x ( x y 1 ) = 0 x(xy-1)=0 , so x = 0 x=0 or x = 1 / y x=1/y .

Substituting into the second equation, we solve for y and we get that either y = n y=n and x = 0 x=0 or y = c 2 y=\frac{c}{2} and x = 2 c x=\frac{2}{c} .

Since there is more than one solution to the system, there is not merely a single complex solution when x = 0 x = 0 and y 0 y \neq 0 .

Similar steps can be used to show that there is more than one complex solution when x 0 x \neq 0 and y = 0 y = 0 . It is shown from substituting y into the second original equation that b = 0 b=0 and that a = c a=c , where c is an arbitrary complex number. When solving the system, x 2 y x = c x^2y-x=c and x y 2 + y = 0 xy^2+y=0 to check for the number of solutions, it is found that the solution can be either x = n x=n and y = 0 y=0 or x = c 2 x=\frac{c}{2} and y = 2 c y=\frac{2}{c} .

Case 3: x = 0 x = 0 and y = 0 y = 0

We can plug in the values for x and y into the two original equations to get that a = 0 a=0 and b = 0 b=0 .

We then solve the system of equations x 2 y x = 0 x^2y-x=0 and x y 2 + y = 0 xy^2+y=0 , plugging in the values for a and b, to see how many solutions there are.

We can factor each equation to become x ( x y 1 ) = 0 x(xy-1)=0 and y ( x y + 1 ) = 0 y(xy+1)=0 .

Now, we see that (0, 0) is the only solution, since x y = 1 xy=1 and y = 0 y=0 , x = x= and x y = 1 xy=-1 , and x y = 1 xy=1 and x y = 1 xy=-1 are impossible.

Therefore, the only possibility where the original system of equations has one complex solution is when x = 0 x=0 and y = 0 y=0 , or when a = 0 a=0 and b = 0 b=0 .

There is one ordered pair (a,b), so the answer to the problem is 1 \boxed1

"For x and y to only have one solution, the discriminants must be 0, " The argument has to be very different: both equations depend on the other variable, which makes this argument wrong.

Calvin Lin Staff - 7 years ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...