Samuel Lazerson

@zhisong This is great work and underlies my suspicions about the code from day one, specifically the rank 4 arrays are inhibiting vectorization. @jloizu while reducing the total number of...

In the [STELLOPT virtual_casing_mod.f90](https://bitbucket.org/lazerson_princeton/stellopt/src/f0f17a887f96a6322a2e85e3324e933e341cb85b/LIBSTELL/Sources/Modules/virtual_casing_mod.f90?at=master) module used in DIAGNO, FIELDLINES, and BEAMS3D, we construct splines over the surface of the required quantities. This alleviates some of the overhead of evaluating the...

@SRHudson Regarding your debugging of DCUHRE here's the code used by vitual_casing_mod.f90 ```fortran SUBROUTINE bfield_virtual_casing_adapt_dbl(x,y,z,bx,by,bz,istat) IMPLICIT NONE ! INPUT VARIABLES DOUBLE PRECISION, INTENT(in) :: x, y, z DOUBLE PRECISION, INTENT(out)...

@SRHudson I believe @jloizu is doing the timing using the standard gnu profiling extension. This is useful as it automatically calculates timing for all routines.

@SRHudson , no it can't tell you time spent in those subroutine because we don't compile those libraries, we only link to them. However, while gprof is nice, there are...

@zhucaoxiang No the idea is that if you use `h5fc` or `h5pfc` then you're using the wrappered compilers and then don't need to worry about passing the compiler or linker...

@zhucaoxiang I attempted to comment out the two lines you suggested and now I get this error: ``` -----------INITIALIZE COILS---------------------------------- rdcoils : identified 2 unique coils in QASDEX.focus ; :...

New error with the CG ``` -----------INITIALIZE COILS---------------------------------- rdcoils : identified 2 unique coils in QASDEX.focus ; : 0 fixed currents ; 0 fixed geometries. : Parameter normalizations : Inorm=...

@zhucaoxiang I think there are still issues ``` -----------COIL DIAGNOSTICS---------------------------------- costfun : Reset target toroidal flux to 1.114269080647567E+00 diagnos : Bnormal ; Bmn harmonic ; tor. flux ; coil length...

The currents were zero in the TF/VF set. I set them to a finite value and it appears to be running through.