Merge branch 'now_DescentMinimizer_and_bugfix' into 'master'

Now descent minimizer and bugfix

See merge request !97

nifty/energies/energy.py

@@ -39,10 +39,6 @@ class Energy(Loggable, object):
     def position(self):
         return self._position
 
-    @position.setter
-    def position(self, position):
-        self._position = position
-
     @property
     def value(self):
         raise NotImplementedError

nifty/minimization/__init__.py

@@ -18,7 +18,7 @@
 
 from line_searching import *
 from conjugate_gradient import ConjugateGradient
-from quasi_newton_minimizer import QuasiNewtonMinimizer
+from descent_minimizer import DescentMinimizer
 from steepest_descent import SteepestDescent
 from vl_bfgs import VL_BFGS
 from relaxed_newton import RelaxedNewton

nifty/minimization/quasi_newton_minimizer.py → nifty/minimization/descent_minimizer.py

@@ -26,7 +26,7 @@ from keepers import Loggable
 from .line_searching import LineSearchStrongWolfe
 
 
-class QuasiNewtonMinimizer(Loggable, object):
+class DescentMinimizer(Loggable, object):
     __metaclass__ = NiftyMeta
 
     def __init__(self, line_searcher=LineSearchStrongWolfe(), callback=None,

nifty/minimization/line_searching/line_search_strong_wolfe.py

@@ -148,8 +148,9 @@ class LineSearchStrongWolfe(LineSearch):
 
         # extract the full energy from the line_energy
         energy_star = energy_star.energy
-
-        return alpha_star, phi_star, energy_star
+        direction_length = pk.norm()
+        step_length = alpha_star * direction_length
+        return step_length, phi_star, energy_star
 
     def _zoom(self, alpha_lo, alpha_hi, phi_0, phiprime_0,
               phi_lo, phiprime_lo, phi_hi, c1, c2):

nifty/minimization/relaxed_newton.py

@@ -16,11 +16,11 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-from .quasi_newton_minimizer import QuasiNewtonMinimizer
+from .descent_minimizer import DescentMinimizer
 from .line_searching import LineSearchStrongWolfe
 
 
-class RelaxedNewton(QuasiNewtonMinimizer):
+class RelaxedNewton(DescentMinimizer):
     def __init__(self, line_searcher=LineSearchStrongWolfe(), callback=None,
                  convergence_tolerance=1E-4, convergence_level=3,
                  iteration_limit=None):
@@ -38,8 +38,3 @@ class RelaxedNewton(QuasiNewtonMinimizer):
         curvature = energy.curvature
         descend_direction = curvature.inverse_times(gradient)
         return descend_direction * -1
-        #norm = descend_direction.norm()
-#        if norm != 1:
-#            return descend_direction / -norm
-#        else:
-#            return descend_direction * -1

nifty/minimization/steepest_descent.py

@@ -16,10 +16,10 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
-from .quasi_newton_minimizer import QuasiNewtonMinimizer
+from .descent_minimizer import DescentMinimizer
 
 
-class SteepestDescent(QuasiNewtonMinimizer):
+class SteepestDescent(DescentMinimizer):
     def _get_descend_direction(self, energy):
         descend_direction = energy.gradient
         norm = descend_direction.norm()

nifty/minimization/vl_bfgs.py

@@ -18,11 +18,11 @@
 
 import numpy as np
 
-from .quasi_newton_minimizer import QuasiNewtonMinimizer
+from .descent_minimizer import DescentMinimizer
 from .line_searching import LineSearchStrongWolfe
 
 
-class VL_BFGS(QuasiNewtonMinimizer):
+class VL_BFGS(DescentMinimizer):
     def __init__(self, line_searcher=LineSearchStrongWolfe(), callback=None,
                  convergence_tolerance=1E-4, convergence_level=3,
                  iteration_limit=None, max_history_length=10):