excessive electron density convergence failures

[gary]

the attached sim was reported on the yahoo group by kawata daisuke on 2008 April 11. it fails on the c07 end of life version and in c08. Robin reported it worked in newmole. it would be good to fix if this can be done without too much effort but could be ignored and fixed by adopting newmole

[rjrw]

I can't remember reporting it working on newmole... It certainly isn't now (@ r3620).

This appears to be due to EdenError?( n1 ) being multi-valued: see the electron density convergence track on failure.

   ConvEdenIoniz: brent fails
1.365992163089566e+02 -2.305586007422164e+00
1.391247212545181e+02 5.701953904662901e+00
1.378619687817374e+02 4.577811068269341e+00
1.372305925453470e+02 3.879465195438314e+00
1.369149044271518e+02 3.397895511861321e+00
1.367570603680542e+02 3.042099354053931e+00
1.366781383385054e+02 2.763206407820803e+00
1.366386773237310e+02 2.533578199084502e+00
1.366189468163438e+02 2.337401962886531e+00
1.366090815626502e+02 2.165495726986165e+00
1.366041489358034e+02 2.012374529227998e+00
1.366016826223800e+02 1.874638603686435e+00

Adding "no molecules" means the problem does not occur.

This suggests that it will be difficult to fix this problem without merging the electron density equilibrium solution in with the other population solutions.

[rporter]

Works without a hitch on newmole r5249.