Prior::get_value() is still float

[KrisThielemans]

stir::GeneralisedPrior::compute_value() calculations are using and returning double. We currently return this after casting to float

SIRF/src/xSTIR/pSTIR/STIR.py

Line 2323 in 1e2c5c1

v = pyiutil.floatDataFromHandle(handle)

SIRF/src/xSTIR/cSTIR/cstir.cpp

Lines 1332 to 1333 in 1e2c5c1

	float v = (float)prior.compute_value(image);
	return dataHandle<float>(v);

A similar problems occured for the log-likelihood but was recently fixed.

[evgueni-ovtchinnikov]

Is it not a STIR issue? - see recon_buildblock/GeneralisedPrior.h:

template <typename DataT>
class GeneralisedPrior : public RegisteredObject<GeneralisedPrior<DataT>>
{
public:
  inline GeneralisedPrior();

  //! compute the value of the function
  /*! For derived classes where this doesn't make sense, it's recommended to return 0.
   */
  virtual double compute_value(const DataT& current_estimate) = 0;

[KrisThielemans]

why? STIR returns double, SIRF casts to float.

[evgueni-ovtchinnikov]

I thought you wanted float, sorry, fixed now by #1291.