Hallo Andreas,
Once you said relating importing starting solutions to hedge:
"What you need to provide hedge is a set of nodal values at the
interpolation nodes it generates for you. If you can somehow interpolate
your existing solution to those nodes, you should be good to go. For
example, on a triangle, first order nodes will sit at the vertices, and
second-order nodes will sit at vertices and edge centers. If you can
make those locations for your grid coincide with data you already have,
you don't even need to interpolate."
Is this supposed to be done as well as when I interrupt a run or a run is
crashed
by any reason or is there any shorter and faster way ?
Regards,
Islam
--
NEU: FreePhone - kostenlos mobil telefonieren und surfen!
Jetzt informieren: http://www.gmx.net/de/go/freephone
Hi Andreas,
thanks for your response. I'd like to know, whether you are able to run the
square.py example. I'm just wondering, because you told me, maybe I am not
feeding hedge with the right date types. But I am not feeding anything, I
am just trying to run the square.py example and then this error comes up.
I took your advice and debugged the square script and all arguments of arg
were of type numpy array float 64, except the first two of them, which were
something like uniform element or something.
I even reinstalled hedge and took another version of boost, but it still
does't work out.
I'm stuck and I really need help to move on with my calculations. Thanks
regards,
Islam
PS: I think I might found a mistake in square.py. make_box returns 4
variables but there are only 3 given to accept the transferparameter in
line 45. When I remove one return value in make_box, it works and no error
message shows up.
>
> -------- Original-Nachricht --------
> Datum: Tue, 18 Jan 2011 13:04:21 -0500
> Von: Andreas Kloeckner <kloeckner(a)cims.nyu.edu>
> An: Islam El Salawi <islamelsalawi(a)gmx.de>, hedge(a)tiker.net
> Betreff: Re: [Hedge] python c++ mismatch
>
> Hi Islam,
>
> apologies for the long wait.
>
> On Wed, 12 Jan 2011 11:24:30 +0100, "Islam El Salawi"
> <islamelsalawi(a)gmx.de> wrote:
> > first I wish you all a happy new year.
> > I tried to run the square.py example and received the following error
> > message:
> >
> > 4020 elements
> > [snip]
> > File "/home/elsalawi/src/hedge/hedge/tools/flops.py", line 32, in
> > wrapped_f
> > return func(*args, **kwargs)
> > Boost.Python.ArgumentError: Python argument types in
> >
> >
> codepy.temp.b7d05f40fbebe2e37390364ffd0cadd4.module.diff(UniformElementRanges,
> > UniformElementRanges, numpy.ndarray, numpy.ndarray, numpy.ndarray,
> > numpy.ndarray, numpy.ndarray)
> > did not match C++ signature:
> > diff(hedge::uniform_element_ranges,
> hedge::uniform_element_ranges,
> > pyublas::numpy_array<double>, boost::numeric::ublas::matrix<double,
> > boost::numeric::ublas::basic_row_major<unsigned long, long>,
> > boost::numeric::ublas::unbounded_array<double, std::allocator<double> >
> >,
> > boost::numeric::ublas::matrix<double,
> > boost::numeric::ublas::basic_row_major<unsigned long, long>,
> > boost::numeric::ublas::unbounded_array<double, std::allocator<double> >
> >,
> > pyublas::numpy_array<double>, pyublas::numpy_array<double>)
>
> What the error means is that the arguments being passed to a
> just-in-time compiled function don't have the right types. In this case,
> it's pretty likely that it's a mismatch with respect to the floating
> point type--say single vs. double precision.
>
> Please check that all the data you are feeding hedge is in double
> precision--in particular the initial conditions. Failing that, run the
> code in a debugger and examine the 'args' array at the failing call
> site. (flops.py: 32) You should check which of the 'args' is a numpy
> array, and in turn which of the numpy arrays have a dtype that doesn't
> match the function signature in the error message (look for float32
> instead of float64, say).
>
> If you need a Python debugger, I'd recommend PuDB [1].
>
> Hope this helps,
> Andreas
>
> [1] http://pypi.python.org/pypi/pudb
>
> --
> Andreas Kloeckner
> Room 1311 (Warren Weaver Hall), Courant Institute, NYU
> http://www.cims.nyu.edu/~kloeckner/
> +1-401-648-059
>
--
NEU: FreePhone - kostenlos mobil telefonieren und surfen!
Jetzt informieren: http://www.gmx.net/de/go/freephone