Hi Troels, For speed, maybe you should send in the single value parameters together with the array versions into the lib.dispersion modules? The check:
+ if np.allclose(dw, np.zeros(dw.shape)) or np.allclose(pA, np.ones(dw.shape)) or np.allclose(kex, np.zeros(dw.shape)): will be very expensive. Regards, Edward On 8 June 2014 19:48, <[email protected]> wrote: > Author: tlinnet > Date: Sun Jun 8 19:48:35 2014 > New Revision: 23737 > > URL: http://svn.gna.org/viewcvs/relax?rev=23737&view=rev > Log: > Re-implemented safety checks in lib/dispersion/cr72.py. > > This is now implemented for both rank-1 float array and of higher dimensions. > > This makes the unit tests pass for multi dimensional computing. > > Task #7807 (https://gna.org/task/index.php?7807): Speed-up of dispersion > models for Clustered analysis. > > Modified: > branches/disp_spin_speed/lib/dispersion/cr72.py > > Modified: branches/disp_spin_speed/lib/dispersion/cr72.py > URL: > http://svn.gna.org/viewcvs/relax/branches/disp_spin_speed/lib/dispersion/cr72.py?rev=23737&r1=23736&r2=23737&view=diff > ============================================================================== > --- branches/disp_spin_speed/lib/dispersion/cr72.py (original) > +++ branches/disp_spin_speed/lib/dispersion/cr72.py Sun Jun 8 19:48:35 > 2014 > @@ -128,9 +128,15 @@ > rank_1 = False > > # Catch parameter values that will result in no exchange, returning flat > R2eff = R20 lines (when kex = 0.0, k_AB = 0.0). > + # For rank-1 float array. > if rank_1: > if dw == 0.0 or pA == 1.0 or kex == 0.0: > back_calc[:] = array([r20a]*num_points) > + return > + # For higher dimensions, return same structure. > + else: > + if np.allclose(dw, np.zeros(dw.shape)) or np.allclose(pA, > np.ones(dw.shape)) or np.allclose(kex, np.zeros(dw.shape)): > + back_calc[:] = r20a > return > > # The B population. > @@ -165,16 +171,23 @@ > > # Catch math domain error of cosh(val > 710). > # This is when etapos > 710. > - if rank_1: > - if max(etapos) > 700: > + if max(etapos) > 700: > + if rank_1: > back_calc[:] = array([r20a]*num_points) > + return > + # For higher dimensions, return same structure. > + else: > + back_calc[:] = r20a > return > > # The arccosh argument - catch invalid values. > fact = Dpos * cosh(etapos) - Dneg * cos(etaneg) > - if rank_1: > - if min(fact) < 1.0: > + if min(fact) < 1.0: > + if rank_1: > back_calc[:] = array([r20_kex]*num_points) > + return > + else: > + back_calc[:] = r20_kex > return > > # Calculate R2eff. > @@ -182,8 +195,10 @@ > > # Catch errors, taking a sum over array is the fastest way to check for > # +/- inf (infinity) and nan (not a number). > - if rank_1: > - if not isfinite(sum(R2eff)): > + if not isfinite(sum(R2eff)): > + if rank_1: > R2eff = array([1e100]*num_points) > + else: > + R2eff = np.ones(R2eff.shape) * 1e100 > > back_calc[:] = R2eff > > > _______________________________________________ > 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
