Also please note that c string functions are error-prone and deprecated. Please interpolate ufl expressions instead. Regards, David On Sun, 19 Mar 2017 at 13:48 William Booker <scwb@leeds.ac.uk> wrote:

Yep that was it.

I feel silly now.

Thanks

Will ------------------------------ *From:* firedrake-bounces@imperial.ac.uk <firedrake-bounces@imperial.ac.uk> on behalf of Andrew McRae <A.T.T.McRae@bath.ac.uk> *Sent:* 19 March 2017 13:40:42 *To:* firedrake@imperial.ac.uk *Subject:* Re: [firedrake] Changing a vector valued function in time loop

-- I suspect you used the label 'a' to define a Function, but then reused it to define your bilinear form. This would explain the error you allude to.

On 19 March 2017 at 13:36, Andrew McRae <A.T.T.McRae@bath.ac.uk> wrote:

I'm not sure what error you're seeing (it would help if you'd posted a full stack trace).

However, I think the normal way to do this is to declare the expression separately:

a_expr = Expression(["0.5*F0*cos(omega*t)*x[1]", "-0.5*F0*cos(omega*t)*x[0]"], F0=F0, omega=omega, t=t) a.interpolate(a_expr)

Inside the time loop, you then update any quantities that you need, and reinterpolate:

a_expr.t = t (or maybe a_expr.t.assign(t), I've forgotten...) a.interpolate(a_expr)

This is better than creating a new Expression object each time you go around the loop, which is what it looks like you are doing at the moment.

On 19 March 2017 at 13:27, William Booker <scwb@leeds.ac.uk> wrote:

Dear Firedrakers,

I'm trying to update a vector function in my time loop

a.interpolate(Expression(["0.5*F0*cos(omega*t)*x[1]", "-0.5*F0*cos(omega*t)*x[0]"], F0=F0, omega=omega, t=t))

but if I use assign or interpolate, I get told the form does not have the respective property. How should I go about making this work, such that it'll change the value of a in my solver?

Thanks

will

-- Dr David Ham Department of Mathematics Imperial College London