Hi,
I just ran the following script:
-----
from math import exp
from numpy import array
times = array([ 0.7, 1. , 0.8, 0.4, 0.9])
I = array([ 476.76174875, 372.43328777, 454.20339981, 656.87936253,
419.16726341])
errors = array([ 9.48032653, 11.34093541, 9.35149017,
10.84867928, 12.17590736])
R = - 500.
I0 = 1000.
params = [R, I0]
I_bc = []
chi2 = 0.0
for i in range(len(I)):
I_bc.append(I0*exp(-R*times[i]))
chi2 += (I[i] - I_bc[i])**2 / errors[i]**2
print(chi2)
-----
And I see:
-----
aaa.py:16: RuntimeWarning: overflow encountered in double_scalars
chi2 += (I[i] - I_bc[i])**2 / errors[i]**2
inf
-----
However optimisation will never reach this R = -500 value. When R =
-5, the chi2 value is 270307925.086. When -0.1, chi2 = 16974.5262241.
Therefore any optimisation algorithm will never let R go up to such a
high negative value to start with. This is not an issue that needs to
be worried about.
Regards,
Edward
On 31 August 2014 14:33, <[email protected]> wrote:
> Author: tlinnet
> Date: Sun Aug 31 14:33:33 2014
> New Revision: 25483
>
> URL: http://svn.gna.org/viewcvs/relax?rev=25483&view=rev
> Log:
> Inserted systemtest Relax_disp.test_finite_value, to illustrate the return of
> inf from C-code exponential, when R is negative.
>
> This can be an issue, if minfx takes a wrong step when no constraints are
> implemented.
>
> bug #22552(https://gna.org/bugs/index.php?22552): Chi2 value returned from
> C-code Curve-fitting return 0.0 for wrong parameters -> Expected influence on
> Monte-Carlo sim.
>
> Modified:
> trunk/test_suite/system_tests/relax_disp.py
>
> Modified: trunk/test_suite/system_tests/relax_disp.py
> URL:
> http://svn.gna.org/viewcvs/relax/trunk/test_suite/system_tests/relax_disp.py?rev=25483&r1=25482&r2=25483&view=diff
> ==============================================================================
> --- trunk/test_suite/system_tests/relax_disp.py (original)
> +++ trunk/test_suite/system_tests/relax_disp.py Sun Aug 31 14:33:33 2014
> @@ -44,6 +44,15 @@
> from status import Status; status = Status()
> from test_suite.system_tests.base_classes import SystemTestCase
>
> +# C modules.
> +if dep_check.C_module_exp_fn:
> + from specific_analyses.relax_fit.optimisation import func_wrapper,
> dfunc_wrapper, d2func_wrapper
> + from target_functions.relax_fit import jacobian, jacobian_chi2, setup
> + # Call the python wrapper function to help with list to numpy array
> conversion.
> + func = func_wrapper
> + dfunc = dfunc_wrapper
> + d2func = d2func_wrapper
> +
>
> class Relax_disp(SystemTestCase):
> """Class for testing various aspects specific to relaxation dispersion
> curve-fitting."""
> @@ -65,7 +74,8 @@
>
> "test_bug_21344_sparse_time_spinlock_acquired_r1rho_fail_relax_disp",
> "test_estimate_r2eff_err",
> "test_estimate_r2eff_err_auto",
> - "test_estimate_r2eff_err_methods"
> + "test_estimate_r2eff_err_methods",
> + "test_finite_value",
> "test_exp_fit",
> "test_m61_exp_data_to_m61",
> "test_r1rho_kjaergaard_auto",
> @@ -3288,6 +3298,26 @@
> self.assert_('test' not in cdp.clustering)
> self.assertEqual(cdp.clustering['free spins'], [':2@N'])
> self.assertEqual(cdp.clustering['cluster'], [':1@N', ':3@N'])
> +
> +
> + def test_finite_value(self):
> + """Test return from C code, when parameters are wrong. This can
> happen, if minfx takes a wrong step."""
> +
> + times = array([ 0.7, 1. , 0.8, 0.4, 0.9])
> + I = array([ 476.76174875, 372.43328777, 454.20339981,
> 656.87936253, 419.16726341])
> + errors = array([ 9.48032653, 11.34093541, 9.35149017,
> 10.84867928, 12.17590736])
> +
> + scaling_list = [1.0, 1.0]
> + setup(num_params=2, num_times=len(times), values=I, sd=errors,
> relax_times=times, scaling_matrix=scaling_list)
> +
> + R = - 500.
> + I0 = 1000.
> + params = [R, I0]
> +
> + chi2 = func(params)
> +
> + print("The chi2 value returned from C-code for R=%3.2f and I0=%3.2f,
> then chi2=%3.2f"%(R, I0, chi2))
> + self.assertNotEqual(chi2, inf)
>
>
> def test_hansen_catia_input(self):
>
>
> _______________________________________________
> relax (http://www.nmr-relax.com)
>
> This is the relax-commits mailing list
> [email protected]
>
> To unsubscribe from this list, get a password
> reminder, or change your subscription options,
> visit the list information page at
> https://mail.gna.org/listinfo/relax-commits
_______________________________________________
relax (http://www.nmr-relax.com)
This is the relax-devel mailing list
[email protected]
To unsubscribe from this list, get a password
reminder, or change your subscription options,
visit the list information page at
https://mail.gna.org/listinfo/relax-devel
