Migrating from v0.3 to v0.4
v0.4.0 rebuilds GSEE on a vectorized computation core: solar positions come from a restructured NREL SPA implementation that is shared across sites and timesteps, and everything downstream operates on (time, site) arrays. The single-site gsee.pv.run_model() keeps its v0.3 call signature but now runs about 20x faster and has slightly more accurate results.
At a glance
| v0.3 | v0.4 |
|---|---|
gsee.pv.run_model(...) |
Same call, with slightly shifted results, and the inverter modelled by default |
Looping run_model over many sites |
gsee.api.run_sites() / gsee.api.run_grid() |
gsee.trigon.sun_angles(...) |
gsee.api.sun_angles_frame(...) |
gsee.trigon.aperture_irradiance(...) |
run_model(..., include_raw_data=True) returns in-plane irradiance |
gsee.brl_model.run(...) |
gsee.core.diffuse.brl_diffuse_fraction(...) |
gsee.climatedata_interface.run_interface(...) |
gsee.climate.run_climate(...) |
ephem always installed |
Optional, only for gsee.legacy: pip install gsee[legacy] |
Installation and dependencies
- GSEE now requires Python ≥ 3.11, pandas ≥ 3 and xarray, and pvlib ≥ 0.15.
ephemis no longer a dependency. It is only needed for the frozen pre-0.4 code ingsee.legacy, installed via the new optional extrapip install gsee[legacy].- The climate data interface's built-in irradiance PDFs now come from the
gsee-climate-datacompanion package (pip install gsee[climate]) instead of a runtime download.