HI, Thank you so much for reply. BTW, is "fp.numerix" typo? And for the second method, it does not works well. Because, here LVV is a matrix depends on position, instead of a scalar. The problem comes out when doing the product of a Matrix and a vector: When I use: LVV*(alphaV*epsilon).faceGrad, it shows: "TypeError: unsupported operand type(s) for *: 'binOp' and '_FaceGradVariable'". ; and If I use: LVV*(alphaV*epsilon).grad, or LVV*v (v=the gradient of (alphaV*epsilon) by derive it explicitly.), it shows: "IndexError: id1 does not have the same size as b".
best, Zebo 2014-05-06 9:20 GMT-05:00 Guyer, Jonathan E. Dr. <[email protected]>: > > On May 5, 2014, at 6:53 PM, Zebo LI <[email protected]> wrote: > > > Thank you so much. I have fixed the boundary condition problem by > introduce an implicit source term, and the code works well. > > However, now I am thinking to write a fipy script with 4 equations, two > of them are ordinary diffusion equations, and the other two > > are not differential equation. But, I don't know how to deal with those > non-differential equations in fipy. The detailed things are shown > > in the attached file. > > I suppose you would write these terms essentially as they appear, e.g, > > eq3 = ImplicitSourceTerm(coeff=1., var=muV) == alphaV * epsilon + kB * T > * fp.numerix.log(CV) > > I don't recommend you do this, though, as there is no way to implicitly > relate muV with CV when you do this. By that, I mean that $k_B T \ln(C_V)$ > cannot be expressed as a matrix operating on $C_V$. > > You are better off carrying the derivatives through to get DiffusionTerms > in CV and in epsilon. If you aren't solving your elastic strain equations, > you can just write the strain part as (LVV * (alphaV * > epsilon).faceGrad).divergence. > _______________________________________________ > fipy mailing list > [email protected] > http://www.ctcms.nist.gov/fipy > [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ] >
_______________________________________________ fipy mailing list [email protected] http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
