Problems with backflow optimization

[braulio “braulio gabriel” gabriel]

I'm trying to optimize the backflow term in the case of the Cu atom. I tried to follow the manual's suggestion and included the backflow terms as follows in the *.opt.xml file. However, it didn't work. Could someone help me?

 -3.70242854000000e-01 -2.46706938000000e-01  1.73301848000000e-01 -1.79983390000000e-01
</coefficient>

        </determinant>
      </slaterdeterminant>
      <backflow>
<!--The One-Body term with independent e-He and e-H terms. IN THAT ORDER -->
       <transformation name="eIonB" type="e-I" function="Bspline" source="ion0">
       <correlation cusp="0.0" size="8" type="shortrange" init="no" elementType="Cu" rcut="3.0">
       <coefficients id="eCuC" type="Array" optimize="yes"> 0 0 0 0 0 0 0 0 </coefficients>  
       </correlation>
       </transformation>
<!--The Two-Body Term with Like and Unlike Spins -->
       <transformation name="eeB" type="e-e" function="Bspline" >
       <correlation cusp="0.0" size="7" type="shortrange" init="no" speciesA="u" speciesB="u" rcut="1.2">
       <coefficients id="uuB1" type="Array" optimize="yes"> 0 0 0 0 0 0 0 </coefficients>  
       </correlation>
      </transformation>
     </backflow>
    </determinantset>
    <jastrow name="J2" type="Two-Body" function="Bspline" print="yes">
      <correlation rcut="10" size="10" speciesA="u" speciesB="u">
        <coefficients id="uu" type="Array"> 0.1781016121 0.04948325799 0.01120319345 -0.01954109079 -0.04585425141 -0.05704377154 -0.05243149109 -0.0148633159 0.01803572837 0.00500966
3233</coefficients>
      </correlation>
      <correlation rcut="10" size="10" speciesA="u" speciesB="d">
        <coefficients id="ud" type="Array"> 0.2182750791 0.03501999768 -0.01932060039 -0.05234541098 -0.08676606223 -0.1004876422 -0.0990968547 -0.06085188381 -0.003347623669 0.007683
263815</coefficients>
      </correlation>
    </jastrow>
    <jastrow name="J1" type="One-Body" function="Bspline" source="ion0" print="yes">
      <correlation rcut="10" size="10" cusp="0" elementType="Cu">
        <coefficients id="eCu" type="Array"> -3.692526917 -2.776497446 0.3730731331 0.8616260796 1.360412446 1.567355441 1.401043801 0.8434927552 0.1608138514 -0.03927361862</coeffici
ents>
      </correlation>
    </jastrow>
    <jastrow name="J3" type="eeI" function="polynomial" source="ion0" print="yes">
      <correlation ispecies="Cu" especies="u" isize="3" esize="3" rcut="5">
        <coefficients id="uuCu" type="Array" optimize="yes"> 0.00145783557 0.0007944261277 -0.001177929366 -0.002023424642 -0.000550239128 0.0009799645077 -0.004642449824 0.0022990561
27 0.007292320419 0.002784101118 -0.00118536769 -0.0009109467117 -0.005157161085 0.006340359977 0.007384264106 -0.0004322496787 -0.01572809167 0.02136326998 -0.003234312764 0.00327645
1027 0.0002871464785 -0.006955540576 0.002043697395 0.003966542845 0.003521160617 -0.001809220705</coefficients>
      </correlation>
      <correlation ispecies="Cu" especies1="u" especies2="d" isize="3" esize="3" rcut="5">
        <coefficients id="udCu" type="Array" optimize="yes"> -0.002841031921 0.002124166708 0.01265257811 -0.0005741159477 -0.003961199817 0.0006560352199 0.006638958786 -0.0054766386
43 0.02176567317 0.000883390562 0.001058234261 -0.00802595157 -0.01468879163 0.001606654272 0.01053227924 0.0007685815326 -0.007912634433 0.0110918315 0.002175616998 0.01379995928 -0.
0006106932376 -0.007071558866 -0.0005425415595 0.006548556162 0.001291890791 -0.0006485241395</coefficients>
      </correlation>
    </jastrow>

 <override_variational_parameters href="Cu.escolhido.vp.h5"/></wavefunction>
</qmcsystem>

[Paul R. C. Kent]

Backflow optimization was removed from the development version of the code, as noted in https://github.com/QMCPACK/qmcpack/blob/develop/CHANGELOG.md#unreleased and https://github.com/QMCPACK/qmcpack/pull/4688 . The development version will run with backflow wavefunctions, just not optimize them. However, all functionality including optimization should work with the last release v3.17.1 -- please try that and let us know how you get on. Within the last few years the backflow implementation has been successfully used in several electron gas studies, and should be fine for atoms and solids. It would be interesting to hear more about your interests in and planned uses of backflow.