Freshmark
← Back to all writing

Class Review 1

Example 1: Analytical Algebra

It is known that the ellipse C:x24+y23=1C:\frac{x^2}{4}+\frac{y^2}{3}=1 and the function y=mx+1+n(m2)y=m^{x+1}+n(m\ge 2) intersect the ellipse at two points A,BA,B, and find the maximum value of nn.

Assume A(x0,y0),B(x0,y0),x0(0,2),y0(0,3)A(x_0,y_0),B(-x_0,-y_0),x_0\in(0,2),y_0\in(0,\sqrt{3}), there are: {y0=mx0+1+n,(1)y0=mx0+1+n,(2)x024+y023=1,(3)\begin{cases} y_0=m^{x_0+1}+n,(1)\\ -y_0=m^{-x_0+1}+n,(2)\\ \frac{x_0^2}{4}+\frac{y_0^2}{3}=1,(3) \end{cases} Considering that there are a total of four variables x0,y0,m,nx_0,y_0,m,n and three constraints, there are 43=14-3=1 degrees of freedom.

Obviously if the scope of one of the variables can be determined, the scope of the other variables will be easily determined:

In order to use symmetry while obtaining identical deformation, consider (1)+(2) and (1)-(2), then the sufficient and necessary conditions are:

{2n=m(mx0+mx0),(4)2y0=m(mx0mx0),(5)x024+y023=1,(3)\begin{cases} -2n=m(m^{x_0}+m^{-x_0}),(4)\\ 2y_0=m(m^{x_0}-m^{-x_0}),(5)\\ \frac{x_0^2}{4}+\frac{y_0^2}{3}=1,(3) \end{cases}

Make a few simple observations:

  • It is easy to see from (4) that n<0n\lt 0.
  • In (4), mx0>1m^{x_0}\gt 1, then mm is larger, 2n-2n is smaller, and nn is larger
  • In (5), the larger m is, the smaller 2y_0 is, and m2m\ge 2 can be eliminated by scaling mm

So, we further perform lossless scaling (m=2):

{2n2(2x0+2x0),(6)2y02(2x02x0),(7)x024+y023=1x024+22x0+22x023,(α)\begin{cases} -2n\ge 2(2^{x_0}+2^{-x_0}),(6)\\ 2y_0\ge 2(2^{x_0}-2^{-x_0}),(7)\\ \frac{x_0^2}{4}+\frac{y_0^2}{3}=1\ge \frac{x_0^2}{4}+\frac{2^{2x_0}+2^{-2x_0}-2}{3},(\alpha) \end{cases}

(α\alpha) is the key to solving the problem! It is not difficult to see that x024+22x0+22x023\frac{x_0^2}{4}+\frac{2^{2x_0}+2^{-2x_0}-2}{3} is monotonically increasing with respect to x0x_0, and the solution is x01x_0\le 1.

So according to (6) we have:

n(2x0+2x0)52\begin{gathered} n\le -(2^{x_0}+2^{-x_0})\le -\frac{5}{2} \end{gathered}

Mathematical background: hyperbolic trigonometric functions

alt text

alt text

Hyperbolic trigonometric functions are derived from hyperbola x2y2=1x^2-y^2=1, so they are called hyperbolic functions

{sinhx=exex2,coshx=ex+ex21,tanhx=sinhxcoshx=exexex+ex=e2x1e2x+1<1\begin{cases} \sinh x=\frac{e^x-e^{-x}}{2},\\ \cosh x=\frac{e^x+e^{-x}}{2}\ge 1,\\ \tanh x=\frac{\sinh x}{\cosh x}=\frac{e^x-e^{-x}}{e^x+e^{-x}}=\frac{e^{2x}-1}{e^{2x}+1}\lt 1 \end{cases}

Similar to trigonometric functions, hyperbolic functions have corresponding identity transformation formulas:

cosh2xsinh2x=1,cosh2x+sinh2x=cosh2x,sinh(x+y)=sinhxcoshy+coshxsinhysinh(xy)=sinhxcoshycoshxsinhycosh(x+y)=coshxcoshy+sinhxsinhycosh(xy)=coshxcoshysinhxsinhytanh(x+y)=tanhx+tanhy1+tanhxtanhytanh(xy)=tanhxtanhy1tanhxtanhy\begin{gathered} \cosh^2 x-\sinh^2 x=1,\\ \cosh^2 x+\sinh^2 x=\cosh 2x,\\ \sinh(x+y)=\sinh x\cosh y+\cosh x\sinh y\\ \sinh(x-y)=\sinh x\cosh y-\cosh x\sinh y\\ \cosh(x+y)=\cosh x\cosh y+\sinh x\sinh y\\ \cosh(x-y)=\cosh x\cosh y-\sinh x\sinh y\\ \tanh(x+y)=\frac{\tanh x+\tanh y}{1+\tanh x\tanh y}\\ \tanh(x-y)=\frac{\tanh x-\tanh y}{1-\tanh x\tanh y}\\ \end{gathered}

For details, see Wikipedia.

When dealing with Example 1 (4) (5), you can also use the analogy of cosh2xsinh2x=1\cosh^2 x-\sinh^2 x=1 and consider the squares and then add them.

Example 2

If f(x)=ax1f(x)=a^{x-1} and g(x)=logax+1g(x)=\log_a x+1 have three intersection points, find the value range of aa:

When a>1a\gt 1, f(x)=ax1f(x)=a^{x-1} increases monotonically, and f(x),g(x)f(x),g(x) are inverse functions of each other, then if the intersection point is not on y=xy=x:

Suppose the intersection point is (x,y)(x,y), then there must be a pair of intersection points (y,x)(y,x), and as long as xyx\neq y, it is inconsistent with monotonicity.

This means that y=ax1y=a^{x-1} as a downward convex function has two intersection points with y=xy=x, which obviously leads to a contradiction:

Therefore, 0<a<10\lt a\lt 1, at this time, there is obviously an intersection point (1,1)(1,1), and the other two intersection points cannot be on y=xy=x, then the two paired intersection points are not on y=xy=x, and there is a pair of (x,y),(y,x)(x,y),(y,x) that is an intersection point that meets the meaning of the question:

{ax1=y,(1)ay1=x,(2)\begin{cases} a^{x-1}=y,(1)\\ a^{y-1}=x,(2) \end{cases}

ax1+ay1=x+y,ax1ay1=yx\begin{gathered} a^{x-1}+a^{y-1}=x+y,\\ a^{x-1}-a^{y-1}=y-x \end{gathered}

Let k=lna<0,x1=u,y1=vk=\ln a\lt 0,x-1=u,y-1=v, then:

eku+ekv=u+v+2,(3)ekuekv=vu,(4)\begin{gathered} e^{ku}+e^{kv}=u+v+2,(3)\\ e^{ku}-e^{kv}=v-u,(4) \end{gathered}

Order u+v=S,uv=Du+v=S,u-v=D again

{ekS2(ekD2+ekD2)=S+2,(5)ekS2(ekD2ekD2)=D,(6)\begin{cases} e^{\frac{kS}{2}}(e^{\frac{kD}{2}}+e^{-\frac{kD}{2}})=S+2,(5)\\ e^{\frac{kS}{2}}(e^{\frac{kD}{2}}-e^{-\frac{kD}{2}})=-D,(6) \end{cases}

(5)/(6):

tanh(kD2)=DS+2\boxed{\tanh(\frac{kD}{2})=-\frac{D}{S+2}}

The above are the necessary conditions for the existence of x,yx,y.

alt text

When SS is fixed, y=1S+2Dy=-\frac{1}{S+2}D is a straight line, considering that it has three intersection points with y=tanh(kD2)y=\tanh(\frac{kD}{2}).

Obviously, it can be seen from the image that the slope of y=tanhkD2y=\tanh \frac{kD}{2} at D=0D=0 is k2\frac{k}{2}.

Then if 1S+2>k2-\frac{1}{S+2}\gt \frac{k}{2}, then the straight line and the hyperbolic tangent function will have three intersection points.

If 1S+2k2-\frac{1}{S+2}\le \frac{k}{2}, then there is only one intersection point between the straight line and the hyperbolic tangent function.

From (3):

S+2=eku+ekv2ekS2\begin{gathered} S+2=e^{ku}+e^{kv}\ge 2e^{k\frac{S}{2}} \end{gathered}

If S<0S\lt 0, then 2>S+22ekS2>22\gt S+2\ge 2e^{k\frac{S}{2}}\gt 2, resulting in a contradiction.

So S0S\ge 0.

Critical condition S0,k1,a1eS\to 0,k\to -1,a\to \frac{1}{e}

Replace "ratio slope at S0S\to0" with a single variable, global argument to completely avoid the trouble of entanglement of two unknown quantities. The cleanest thing is to go back to a fixed point language.

Assume f(x)=ax1f(x)=a^{x-1} (0<a<10<a<1), the axis outer pair is the fixed point of g(x)=f(f(x))g(x)=f(f(x)) except x=1x=1. inspection

ψ(x)=f(f(x))x,ψ(1)=0.\psi(x)=f(f(x))-x,\qquad \psi(1)=0.

**Step 1: ψ\psi Up to 3 zero points. ** To calculate the derivative, record w(x)=f(x)+x2w(x)=f(x)+x-2, then

ψ(x)=f(f(x))f(x)1=(lna)2aw(x)1.\psi'(x)=f'(f(x))f'(x)-1=(\ln a)^2\,a^{\,w(x)}-1.

Moreover, w(x)=ax1lna+1w'(x)=a^{x-1}\ln a+1 is strictly increasing, so ww first decreases and then increases and has a unique minimum. Therefore, when the base is less than 1, aw(x)a^{w(x)} first increases and then decreases. Thus ψ\psi' is also unimodal: although it tends to 1-1 at both ends, it may become positive in between. Hence ψ\psi' changes sign at most twice, and ψ\psi has a decrease–increase–decrease profile with at most three zeros. By inverse-function symmetry, nontrivial zeros occur in pairs, so there is at most one off-axis pair.

**Step 2: The threshold is exactly ψ(1)=0\psi'(1)=0. ** At the known zero point x=1x=1,

ψ(1)=f(1)21=(lna)21.\psi'(1)=f'(1)^2-1=(\ln a)^2-1.

  • If (lna)21(\ln a)^2\le1 (i.e. a1ea\ge\frac1e): then ψ(1)0\psi'(1)\le0. Combined with the single-peak structure, it can be verified that ψ\psi no longer crosses zero on both sides of x=1x=1 (it crosses down at 11, and both ends are also facing -, and the positive peak in the middle cannot reach to create a new intersection point if it exists), so there is only one zero point** and only one intersection point for x=1x=1.
  • If (lna)2>1(\ln a)^2>1 (that is, a<1ea<\frac1e, because lna<0\ln a<0 is lna<1\ln a<-1): then ψ(1)>0\psi'(1)>0, ψ\psi crosses the zero point at x=1x=1. However, ψ(x)+ (x)\psi(x)\to+\infty\ (x\to-\infty) and ψ(x) (x+)\psi(x)\to-\infty\ (x\to+\infty) (use 0<a<10<a<1 to directly test the limits at both ends), combined with the "decrease, increase and decrease" shape caused by a single peak, x=1x=1 forces a new zero point on both sides**, which is exactly a pair of off-axis solutions.

**Step Three: Continuity Closure (replacing your S0S\to0). ** The second step is sufficient and necessary, but if you want to clarify "why this pair of solutions was born at a=1ea=\frac1e": ψ(1)=(lna)21\psi'(1)=(\ln a)^2-1 continues with aa, and when a1ea\uparrow\frac1e is ψ(1)0\psi'(1)\downarrow0, the pair of zero points are continuously merged into x=1x=1 (that is, D0D\to0). This is the correct origin of the "S0S\to0" phenomenon - it is the ultimate behavior of the conclusion, not a criterion used as a premise.

Therefore, it is important to:

three intersections    (lna)2>1 and 0<a<1    0<a<1e.\text{three intersections}\iff(\ln a)^2>1\ \text{and}\ 0<a<1\iff \boxed{0<a<\tfrac1e}.

Example 3

(2021 Tsinghua Strong Foundation) defines xy=x+y1+xyx*y=\frac{x+y}{1+xy}, then (...(23)4)...)21=(...(2*3)*4)...)*21=__.

It is not difficult to see that the background of this question is the hyperbolic tangent function.

Let x=u1u+1,y=v1v+1,xy=uv1uv+1x=\frac{u-1}{u+1},y=\frac{v-1}{v+1},x*y=\frac{uv-1}{uv+1}, where:

{u=x+1x1,v=y+1y1\begin{cases} u=-\frac{x+1}{x-1},\\ v=-\frac{y+1}{y-1} \end{cases}

Note g(x)=x+1x1g(x)=-\frac{x+1}{x-1}, then:

(xy)z=(g(x)g(y)1g(x)g(y)+1)z=g(x)g(y)g(z)1g(x)g(y)g(z)+1\begin{gathered} (x*y)*z=(\frac{g(x)g(y)-1}{g(x)g(y)+1})*z=\frac{g(x)g(y)g(z)-1}{g(x)g(y)g(z)+1} \end{gathered} (...(23)4)...)21=g(2)g(3)...g(21)1g(2)g(3)...g(21)+1(...(2*3)*4)...)*21=\frac{g(2)g(3)...g(21)-1}{g(2)g(3)...g(21)+1} Among them g(2)g(3)g(4)...g(21)=314253...2220=21×221×2=231g(2)g(3)g(4)...g(21)=\frac{-3}{1}\frac{-4}{2}\frac{-5}{3}...\frac{-22}{20}=\frac{21\times22}{1\times2}=231

What you want =230232=115116=\frac{230}{232}=\frac{115}{116}