Axel

Memory effect: like a spring: interpolate between formulae
-->

division by zero: improve
-->

Eq.(1.37), J = ... exp[-pi m^2/e|Q|E]
what mass do we need to suppress Schwinger
do simulation with mass
-->

Plot final value (E,B) vs time for different alpha/f
-->

E/B ~ vs time or alpha/f
-->

time dependence
coll, non
rho vs alpha

indepence of final results on BD cutoff.

How far from Ohm's law, in mixed picture

rho_E, rho_B same at end of inflation
rho_E/rho_total and rho_E/rho_B

Bfinal vs 

is J/E linear? as scatter plot, different times as different colors

fix sigE, sigB when E=0 or B=0

linear regime without backreaction.