[Scipy-tickets] [SciPy] #390: scipy.optimize.fmin_bfgs fails without warning

SciPy scipy-tickets@scipy....
Fri Mar 23 09:12:32 CDT 2007


#390: scipy.optimize.fmin_bfgs fails without warning
--------------------+-------------------------------------------------------
 Reporter:  ycopin  |       Owner:  somebody
     Type:  defect  |      Status:  new     
 Priority:  normal  |   Milestone:          
Component:  Other   |     Version:          
 Severity:  normal  |    Keywords:          
--------------------+-------------------------------------------------------
 On the particular attached test case (buggy.py using dataset from
 buggy.dat), the fitting routine scipy.optimize.fmin_bfgs will fail without
 warning, while it definitely did not converge:

 True parameters: (5.0, 1.0, -5.0)

 Optimization terminated successfully.
          Current function value: 1494.085082
          Iterations: 19
          Function evaluations: 31
          Gradient evaluations: 31

 Bfgs: [-2.5702969   1.1423796  -0.00866672]

 When using random dataset (uncomment line 47), it converges *most* of the
 time, but for this report, I saved a particular dataset for which the fit
 failed. I tested the jacobians, which are OK.

-- 
Ticket URL: <http://projects.scipy.org/scipy/scipy/ticket/390>
SciPy <http://www.scipy.org/>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list