5

Per the method described at How to find asymptotes of implicit function? , I proceeded to find the asymptotes of $$ x^3 + 3x^2y - 4y^3 - x + y + 3 = 0 $$

Whilst, it correctly generates the asymptote $$ y=x $$ , the remaining asymptote(s):- $$ 2y + x = ±1 $$ can't be deduced.

Instead another erroneous line $ y = 0 $ is outputted.

Some help will be appreciated.Alternate methods of solution would work too:)

  • 2
    The problem with the linked method seems to stem from the fact that $(X,Y,Z) = (2,-1,0)$ is a singular point of the homogenized curve. I'm not familiar enough with projective coordinates to know how to get around this, though. – Michael Seifert Jul 13 '18 at 13:53

7 Answers7

2

If the slant asymptote exists, we may denote $$k=\lim_{x \to \infty}\dfrac{y}{x}.$$

Then $$0=\lim_{x \to \infty}\left(1+3\cdot \frac{y}{x}-4 \cdot \frac{y^3}{x^3}-\frac{1}{x^2}+\frac{y}{x^3}+\frac{3}{x^3}\right)=1+3k-4k^3=-(k-1)(2k+1)^2.$$

Thus,$$k=1,-\frac{1}{2}.$$


  • If $k=1$, we denote $y=x+b$ and put it into the equation, we have $$-4b^3-12b^2x-9bx^2+b+3=0.$$

Then $$0=\lim_{x \to \infty} \frac{-4b^3-12b^2x-9bx^2+b+3}{x^2}=-9b.$$

Thus,$$b=0.$$


  • If $k=-\dfrac{1}{2}$, we denote $y=-\dfrac{x}{2}+b$ and put it into the equation, we have $$-8b^3+12b^2x+2b-3x+6=0.$$

Then $$0=\lim_{x \to \infty} \dfrac{-8b^3+12b^2x+2b-3x+6}{x}=12b^2-3.$$

Thus, $$b=\pm \frac{1}{2}.$$


As a result, the slant asymptotes are $$y=x, y=-\frac{x}{2}\pm \frac{1}{2}.$$

mengdie1982
  • 13,840
  • 1
  • 14
  • 39
  • 1
    best approach is to combine David K answer(+1) (to find the inclinations) with yours (+1)( to find the intercepts). – G Cab Jul 13 '18 at 23:34
2

The question does not give details about how to apply the method of How to find asymptotes of implicit function?, to this particular problem. Here is an application of that method:

In homogeneous coordinates $(X:Y:Z)$, where $(x,y)$ corresponds to $(x:y:1),$ the equation is $$ X^3 + 3X^2Y - 4Y^3 - XZ^2 + YZ^2 + 3Z^3 = 0. \tag1 $$

On the line at infinity, $Z = 0,$ yielding $$ X^3 + 3X^2Y - 4Y^3 = 0, $$ which factors to $$ (X - Y) (X + 2 Y)^2 = 0.$$ This is solved by $X = Y$ or $X = -2Y,$ which says that the asymptotes are of the form $x = y + \text{constant}$ or $x = -2y + \text{constant},$ but this does not say what the constants are.

Continuing with the method of How to find asymptotes of implicit function?, the derivative of the left-hand side of Equation $(1)$ is $$ (3X^2 + 6XY - Z^2)dX + (3X^2 - 12Y^2 + Z^2)dY + (2YZ - 2XZ + 9Z^2) dZ.\tag2 $$

Evaluating $(2)$ at $(X:Y:Z) = (1 : 1 : 0)$ produces $ 9 dX - 9 dY + 0dZ, $ from which we eventually find the asymptote $x = y$ (whose constant term is zero).

But evaluating $(2)$ at $(X:Y:Z) = (-2 : 1 : 0)$ gives only $0 dX + 0 dY + 0 dZ,$ from which we cannot derive the equation of a line. (It is not clear how you got $y = 0$.) So we have to find the constants that give the asymptote lines some other way.

David K
  • 98,388
1

Giving

$$ f(x,y) = x^3 + 3 x^2 y - 4 y^3 - x + y - 3 = 0 $$

the asymptotic directions can be explored by substituting $ y = a x + b\;\; $ into $f(x,y)$ giving

$$ f(x,ax+b) = 3 x^2 (a x+b)-4 (a x+b)^3+a x+b+x^3-x-3 = (1+3a-4a^3)x^3+3b(1-4a^2)x^2 +(a(1-12b^2)-1)x+b(1-4b^2)-3 $$

Now the conditions for $f(x,ax+b)$ to have line behavior are

$$ \left\{ \begin{array}{rcl} 3 b-12 a^2 b=0 \\ -4 a^3+3 a+1=0 \\ \end{array} \right. $$

and solving we find two solutions:

$$ \{a =-\frac 12, \forall b\} \cup \{a = 1, b = 0\} $$

so we have

$$ L_1\to y = x\\ L_2\to y = -\frac x2 + b $$

Considering $f(x,y)$ at their asymptotic values it's value should be the same or

$$ \lim_{x\to 0}f(x,-\frac 12 x+b)=\lim_{x\to 0}f(x,x) = -3 $$

so this can be solved by choosing $b$ such that $1-4b^2 = 0\;\;$ giving

$$ L_1\to y = x\\ L_2\to y = -\frac x2 \pm\frac 12 $$

Attached a plot showing in blue $f(x,y) = 0$ and in red $L_1, L_2$

enter image description here

Cesareo
  • 33,252
  • Thanks, but a couple of doubts.....Firstly, when we solve the first equation that outputs a = +1/2 & -1/2 and/or b = 0.Similarly,the second equation leads to a = -1/2 and 1.....How are we rejecting a=1/2.....And,why shall b be calculated by choosing (1-4b^2)=0 – Winged Blades of Godric Jul 13 '18 at 13:43
  • Also,how are we tallying the a's and b's....? – Winged Blades of Godric Jul 13 '18 at 13:50
  • $f(a,ax+b) = c_1 + c_2x+c_3x^2+c_4x^3$ so the conditions are $c_3=c_4=0$. The value $a = -\frac 12$ obeys simultaneously $c_1=c_2=0$ and $a = \frac 12$ does not. – Cesareo Jul 13 '18 at 14:50
  • I guess that the last line of your comment has a typo as to the subscripts of (c).....At any case, how are we accepting that (a)=1, given that it does not satisfy both the equations, simultaneously?.......I think some more clarity on <1> the exact outputs of (a) and (b), per equation,<2> as to why a particular value of (a) is being rejected, <3>how are you tallying the (a)s with (b)s i.e. how are you choosing (a) as 1 and (b) as 0 in the same equation rather than (a) as -1/2 and (b) as 0, and <4> finally as to why you chose (1-4b^2)=0 ........would be immensely helpful. – Winged Blades of Godric Jul 14 '18 at 14:44
  • @WingedBladesofGodric Yes. There is a typo. Should be $c_3=c_4=0$ I will try to clarify your doubts ASAP. Just now I can't follow. Thanks. – Cesareo Jul 14 '18 at 17:00
  • @WingedBladesofGodric I added some explanations I hope will help. – Cesareo Jul 14 '18 at 18:59
0

In other answers we find various heuristics leading to the conjecture that there are asymptotes of the form $$y= x+C\ , \qquad y=-{1\over2}x +C\ .$$ Therefore we introduce new coordinates $(u,v)$ in the plane such that their directions become the directions of the axes: $$\left\{\eqalign{ u&=x+2y\cr v&=x-y\cr}\right.\qquad{\rm resp.}\qquad\left\{\eqalign{x&={u+2v\over3}\cr y&={u-v\over3}\cr}\right.\ .\tag{1}$$ The given function then appears as $$g(u,v)=f\left({u+2v\over3},{u-v\over3}\right)=3+(u^2-1)v\ .$$ The equation $g(u,v)=0$ is equivalent with $$v={3\over 1-u^2}\ .$$ In terms of $u$ and $v$ we therefore have the two vertical asymptotes $u=\pm1$ and the horizontal asymptote $v=0$. These can now be expressed again in terms of $x$ and $y$, using $(1)$.

0

$$x^3 + 3x^2y - 4y^3 - x + y + 3 = 0 \tag 1$$ Another variant of method consists in expending $y(x)$ to series of the form : $$y(x)\simeq ax+b+\frac{c}{x}+\frac{d}{x^2}+... \tag 2$$ Putting it into Eq.$(1)$ transforms it to the form : $$Ax^3+Bx^2+Cx+D+...\simeq 0 \quad\text{which implies}\quad A=B=C=D=…=0$$ where $$A=1+3a-4a^3=0$$ $$B=3b-12ba^2=0$$ Etc. But we don't compute immediately $C$ , $D$ , … in the general case because this will be much simpler latter in each particular case.

The roots of $1+3a-4a^2=0=(1-a)(1+2a)$ are $a=1$ and $a=-\frac12$.

Case $a=1$ :

In this case $B=3b-12ba^2=0$ implies $b=0$, thus $y(x)\simeq x+\frac{c}{x}+\frac{d}{x^2}+...$ Putting it into Eq.$(1)$ leads to : $$C=-9c=0\quad\text{thus}\quad c=0$$ $$D=3-9d=0\quad\text{thus}\quad d=\frac13$$ $$y(x)\simeq x+\frac{1}{3x}+...$$ So, for $x\to\infty\quad \textbf{the straight line} \quad y=x\quad\textbf{is an asymptote.}$

Case $a=-\frac12$ :

In this case $B=3b-12ba^2=0$ is satisfied any $b$, thus $y(x)\simeq -\frac12x+b+\frac{c}{x}+...$ Putting it into Eq.$(1)$ leads to : $$C=6b^2-\frac32 x=0 \quad\text{thus}\quad b=\pm\frac12$$ $$D=3-4b^3+12bc+b=0\quad\text{thus}\quad c=\frac{4b^3-b-3}{12b}$$

Case $a=-\frac12$ and $b=\frac12$ then $c=-\frac12$ : $$y(x)\simeq -\frac12x+\frac12-\frac{1}{2x}+...$$ So, for $x\to\infty\quad \textbf{the straight line} \quad y=-\frac12x+\frac12\quad\textbf{is an asymptote.}$

Case $a=-\frac12$ and $b=-\frac12$ then $c=\frac12$ : $$y(x)\simeq -\frac12x-\frac12+\frac{1}{2x}+...$$ So, for $x\to\infty\quad \textbf{the straight line} \quad y=-\frac12x-\frac12\quad\textbf{is an asymptote.}$

JJacquelin
  • 66,221
  • 3
  • 37
  • 87
  • This seems quite nice and the best of the lot,without meaning any disrespect for the other answers......At any case, any hints about the expansion of y(x)? – Winged Blades of Godric Jul 14 '18 at 14:42
  • At first sight, it appear nice considering the simplicity of the principle. But is is far to be the method requiring a minimum of work : very simple calculus of expansion and rearranging of series, but rather boring ! One have to put Eq.$(2)$ into Eq.$(1)$ and rearrange the terms in decreasing order of the powers of $x$. – JJacquelin Jul 14 '18 at 17:59
0

I'd try to give another answer. Recall from wikipedia and the most common definition of asymptotic line, to find an asymptotic line of $f(x)$ as $x\to\infty$, we need to compute

(1) $a=\lim_{x\to\infty}\frac{f(x)}{x}$

(2) $b=\lim_{x\to\infty}(f(x) - ax)$.

Then the asymptotic line would be $y=ax+b$.

-

By implicit function theorem when $x$ is large, $y$ can be written as a function of $x$. Let's begin with

$x^3 + 3x^2 y - 4y^3 -x + y + 3 = 0. \qquad(1) $

Then

$1+ 3\frac{y}{x} - 4(\frac{y}{x})^3 - \frac{1}{x^2} + \frac{y}{x}\frac{1}{x^2} + \frac{3}{x^3}=0$

if $\lim_{x\to\infty} \frac{y}{x}=:a$ exists, then it might be a solution of $1+3a-4a^3=0$. Moreover, the little consequence $\lim_{x\to\infty} \frac{y-ax}{x}=0$ is required.

-

Then we may try $x^3 + 3x^2(y-ax+ax) -4(y-ax+ax)^3 - x + (y-ax+ax) + 3 = 0$. After simplifying, we obtain

$3(y-ax) - 4(y-ax) (\frac{y-ax}{x})^2 - 12 a(\frac{y-ax}{x})^2 - 12( y-ax)a^2 - \frac{1}{x} + \frac{y-ax}{x}\frac{1}{x} + \frac{a}{x} + \frac{3}{x^2}=0$. Thus

$(y-ax)[3 -4(\frac{y-ax}{x})^2 -12a^2] = 12a(\frac{y-ax}{x})^2 + \frac{1}{x} -\frac{y-ax}{x}\frac{1}{x} - \frac{a}{x} -\frac{3}{x^2}$. Taking limit, we obtain $b(3-12a^2)=0$, where $b=\lim_{x\to\infty} y-ax$, existing.

user74489
  • 497
0

$$\newcommand{\AFF}{\mathbb{A}}$$ $$\newcommand{\PSP}{\mathbb{P}}$$

Let

$$ f(x,y) = 0 $$

be an affine plane curve in $\AFF^2_k$. How can one find all possible asymptotic lines $a x + b y + c = 0$?

Answer: One embeds $\AFF^2_k$ into $\PSP^2_k$ with the coordinates $(X:Y:Z)$ and computes the intersections of $f(x,y) = 0$ with the infinite line $Z = 0$. Therefore one puts

$$ Z^d f(X/Z, Y/Z) = g(X,Y,Z) = 0 $$

with $d$ chosen such, that $Z$ does not divide the polynomial $g(X,Y,Z)$.

From $g(X,Y,0) = 0$ one gets a list of intersection points $(X_i: Y_i: 0)$. We choose one of it, which we write wlog as $P = (1 : y_0 : 0)$.

So we assume, that $X = 1$ and compute the affine representation of $f(x,y) = 0$ in the patch $X = 1$, that is $g(1, Y/X, Z/X) = 0$ as $p(u, v) = 0$ with $u=Y/X$ und $v = Z/X$.

It is then $u = y_0, v = 0$ the intersection point $P$ from above. We put $u' = u - y_0$ and get $P = (u' = 0, v = 0)$. Therefore

$$ h(u',v) = p(u' + y_0, v) = h_m(u',v) + \cdots + h_d(u',v) $$

with $h_\nu$ homogeneus of degree $\nu$ and $d > 0$. We choose a linear factor $l = a u' + b v = 0$ of $h_d$ and transform back

$$ l = a \,(u - y_0) + b \, v = a \, (Y/X - y_0) + b \, (Z/X). $$

Projectively this is

$$ l = a Y - y_0 X + b Z $$

and so in the affine patch $Z = 1$ the sought after equation for the asymptotic line is

$$ l = a y - y_0 x + b = 0 $$

Example: Let

$$ f(x,y) = x^3 + y^3 - 3 a \, x y = 0 $$

be "Folium Cartesii".

We have

$$ g(X,Y,Z) = X^3 + Y^3 - 3 a X Y Z = 0. $$

From $Z= 0$ follows $P=(1:-1:0)$, therefore $y_0 = -1$.

Further we have $p(u,v) = g(1,u,v) = 1 + u^3 - 3 a u v$ and therefore with $u' = u + 1$ we get for $h$

$$ h(u',v) = p(u' - 1, v) = u'^3 - 3 u'^2 + 3 u' - 3 a (u' - 1) v = u'^3 - 3 u'^2 - 3a u' v + 3 u' + 3 a v. $$

So it is $h_1(u',v) = 3 (u' + a v)$ and $l = u' + a v = u + 1 + a v = Y/X + 1 + a Z/X = 0$, and therefore finally

$$ l = y + x + a = 0. $$

Finally your example, calculated with maple

enter image description here

enter image description here