Surprisingly, I did it.

Using the method provided by Robert Israel above, the equation

tan(x*sqrt(cp^2/6450^2-1))*(cp^2/3220^2-2)^2+4*tan(x*sqrt(cp^2/3220^2-1))*sqrt((cp^2/3220^2-1)*(cp^2/6450^2-1)),

in the interval [3220,6450] can be transformed to

tanh((1/6450)*x*sqrt(41602500-cp^2))*((1/10368400)*cp^2-2)^2+(1/5192250)*tan((1/3220)*x*sqrt(cp^2-10368400))*sqrt(-431351361000000-cp^4+51970900*cp^2),

and then using implicitplot can get the curve in the interval [0, 3220]

but I don't know why it works.

can someone explain for that, thank you.

hello, my friends.

I got in trouble when compute cp<3220, maple can't give a reasonable initial condition.

can you help me?

Thank you very much.

If you have a nvidia display card and your maple version is 14 or 15, you can try cuda computation, more information is here:http://www.maplesoft.com/support/help/Maple/view.aspx?path=CUDA

For Markiyan Hirnyk, thank you for your reply and searching in Mapleprimes,

I have tried several methods provided in your reply, but none of them worked,thank you all the same.

For Preben Alsholm

thank you for your reply.That's exactly what I want.

This method is excellent and it can be extended to arbitrary higher derivative.

Thank you very much

One more question,

With the solution of dsolve command, how to get a plot of [t,diff(x(t),t$2)] curve for the first system?

Thank you in advance.

Thank you very much,Markiyan Hirnyk and Preben Alsholm

For Markiyan Hirnyk, thank you for help me to ask this question on other forum, thank you.

The answer is quite right.

For Preben Alsholm, thank you for keep on thinking with my question,thank you very much.

The answer is compact，thank you very much。

finally，thank you very much again，my friends。

Many thanks for you,Markiyan Hirnyk and Preben Alsholm.
The first problem is in my textbook,I think it can't be consistent,
I just want to know how to solve it by maple,thank you for your reply.
For Preben Alsholm,the result is excellent,
thank you very much.
I'm still waiting for the solution of the first problem.
can anyone help me,thank you in advance.

thank you for your reply,Markiyan Hirnyk.

I have read some books about conformal mapping,especially SC mapping.

Finally,I found that the SC mapping can't give an analytic relation with displace four or more points.

It can only be resolved by numerical computation.

Thank you all the same。

Thanks for your reply,pagan.
I used the command CurveFitting[LeastSquares] and specified the curve function (((((((((a*x+b)*x+c)*x+d)*x+e)*x+f)*x+g)*x+h)*x+i)*x+j)*x+k.
After that I got two functions,for x0-y0.txt it's 4176.179937351788382288395727590640646351-300.6660540493730448776826840175290258286*x+27521.39108097349785713682383847275900435*x^3-3.735406493253002990016789970254731216828*10^5*x^4+1.672189671937271806360308206963604946023*10^6*x^5-3.857371254620151064729121207146077966000*10^6*x^6+5.175950228893425696259290469039659777918*10^6*x^7-4.100441196197960012534870932764938467014*10^6*x^8+1.786010743645443341314221492701395902885*10^6*x^9-3.307281618623904306129232393702773370002*10^5*x^10+26.*x^2,

for x1-y1.txt it's 4176.179997041168008864990042291109931504-300.6665828273639226659942812670360484839*x+27521.49231942589937197915717094142898319*x^3-3.735410744245982936921697015714414169936*10^5*x^4+1.672190000196991333711308374328028613843*10^6*x^5-3.857369737309735398768456231880442933758*10^6*x^6+5.175945842072068693340540683402288678014*10^6*x^7-4.100436134041539005871098506785829141215*10^6*x^8+1.786007917733174840531550632751061268480*10^6*x^9-3.307275323052890598670089099106542035511*10^5*x^10+26.*x^2.
And when I use Statistics[PolynomialFit] command, two different results was derived.
After several computations,I think that two different results were optimum for x0-y0.txt and x1-y1.txt,but it didn't the optimum for (y1-y0)/y0.
With the results got from CurveFitting[LeastSquares], I got a better graph with the specified function.
It's not satisfied because at the start interval and the end interval it's distorted from the result got from the command CurveFitting[ThieleInterpolation].
Would anyone show me a better result or give me a constructive suggestion?
Thank you very much in advance.

Initially, I just want to know whether maple will support gpu symbolic computation.

Imagine that someone bought a computer,

and it can be accelerated by the gpu in maple, so, why not buy this type PC?

then I mean that's the trend.

Ansys 13 already include Nvidia and AMD accelerating function, although very limited.

After 1 or 2 years, there will be plenty of cards can do computation.

So, if the full computation resources can be utilized,

why not support that?

that's only my opinion.

Again, thank you for all answers.