dielectric_hjz...@126.com wrote: > My question is: which value of grid should we use during the > discretizing process? Should we decrease the "grid" from a larger > value to a smaller value, monitoring the band structure and > conductance behavior, until the band structure and conductance > behavior remain almost unchanged when further decreasing the "grid"?
Hi, it is in general (independently of Kwant) a good idea to setup numeric simulations such that resolutions (there can be several, e.g. in space, time, energy) are explicit parameters that can be modified and thus controlled. Increasing a resolution will generally provide more accurate results at the cost of more time and resource usage. (But you can also hit a wall where accuracy is limited by something else.) You need to find a compromise between the accuracy you desire and the resources required. To this end, it is often useful to plot the computed quantity as a function of resolution (starting at really low resolutions) and try to assess whether and how it converges.
smime.p7s
Description: S/MIME cryptographic signature
