Bugfix: Do not pass argument f_grad to gradient-free optimization alg…

…orithms
andreArtelt · Sep 4, 2019 · 45f8416 · 45f8416
1 parent 399ccaf
commit 45f8416
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/ceml/optim/optimizer.py b/ceml/optim/optimizer.py
@@ -149,7 +149,7 @@ def prepare_optim(optim, f, x0, f_grad=None, tol=None, max_iter=None):
         else:
             raise ValueError(f"Invalid value of 'optim'.\n'optim' has to be 'nelder-mead', 'powell', 'cg' or 'bfgs' but not '{optim}'")
     elif isinstance(optim, Optimizer):
-        args = {'f': f, 'x0': x0, 'f_grad': f_grad, 'tol': tol, 'max_iter': max_iter}
+        args = {'f': f, 'x0': x0, 'tol': tol, 'max_iter': max_iter}
         if is_optimizer_grad_based(optim):
             args['f_grad'] = f_grad