MaplePrimes - answers and comments on Question, unable to store HFloat

Imaginary

Preben Alsholm — Thu, 08 Nov 2012 07:26:31 Z

The term n*(-diff(f(eta),eta$2))^(n-1)*diff(f(eta),eta$3) is likely to cause problems. Is this written exactly as you want it?

Try

evalc((-1*p)^(0.4-1)) assuming p>0;

You see that if diff(f(eta),eta,eta) is real and positive then (-diff(f(eta),eta$2))^(n-1) is imaginary.

Re((-diff(f(eta),eta$2))^(n-1) )?

J4James — Thu, 08 Nov 2012 12:13:24 Z

@Preben Alsholm thanks for your explanation.

The system is exactly the one for which I need a solution. You can see the attached research

paper Magnetohydrodynamic_.pdf, where they give the solution of exactly the same equation (eq1).

Infact, I adopted the eq1 from there including one extra term which I needed for my model.

As you rightly mentioned that, "if diff(f(eta),eta,eta) is real and positive then (-diff(f(eta),eta$2))^(n-1) is imaginary". What I am thinking is that can we consider the real part and find the solution?

How we will do it in Maple?

Thanks

Re

PatrickT — Thu, 08 Nov 2012 13:34:32 Z

@J4James

haven't read your question in detail, still if your imaginary numbers are spurious, you could try to set the option 'complex"=false in dsolve or you could use Re(x(t)) to trim the imaginary bits out. The former is presumably a better approach.

http://www.maplesoft.com/support/help/Maple/view.aspx?path=Re

How?

J4James — Thu, 08 Nov 2012 16:03:00 Z

@PatrickT, I tried to follow ur suggestions but no luck.

1. complex=false

 dsolve(subs(para,epsilon=1.6, {bc, eq1, eq2}), numeric,complex=false):
Error, (in dsolve/numeric/bvp) bvp keyword was complex, optional keyword must be

 one of 'output', 'value', 'range', 'abserr', 'maxmesh', 'initmesh','continuation', 'mincont',

'approxsoln', 'adaptive', 'interpolant', 'optimize'

2. Re

eq1:=Re(n*(-diff(f(eta),eta$2))^(n-1)*diff(f(eta),eta$3))+(2*n/(n+1))*f(eta)*

diff(f(eta),eta$2)-diff(f(eta),eta$1)^2-M*diff(f(eta),eta$1)+M*epsilon+epsilon^2+

lambda*theta(eta)=0;

eq2:=remain the same with bcs;

dsolve(subs(para,epsilon=1.6, {bc, eq1, eq2}), numeric):
Error, (in fproc) unable to store 'HFloat(0.251188643150958)*

RootOf(2000000000*Re(_Z)-HFloat(6.5e10))' when datatype=float[8]

Thanks

bvp v. ivp

PatrickT — Fri, 09 Nov 2012 13:10:12 Z

you're right, 'complex' is not an option when dsolve takes a bvp as input, I hadn't noticed that. Can you convert your system to an ivp? if you have a known endpoint, you could start from there and run time backwards (i.e. t=-10..0). Have you tried variations on your problem? i.e. different parameter values. Do you have problems for this specific formulation or more generically?

the same issue with another model

J4James — Tue, 13 Nov 2012 09:04:09 Z

@PatrickT

I am facing the same issue of "unable to store 'HFloat(1.005717340914193)*y'".

Even I tried to to use approx solution but no luck.

Please have a look. P_new.mw

Thanks