I know your signature indicates "Nipro India," but if this 2000w pure sine wave inverter was a latitude-tilt (no tracking) system in Caribou, ME, in the winter, your harvest numbers of 3.5-4 kWh per day would make sense. Average solar radiation there, per NREL, is 2.4, 2.5 kWh/m^2/day in November and December.
if it's a new system efficiency is equal to output power over input power. To calculate the pure sine wave inverter input power you need to figure out the coverage area and amount of power received on the Earth surface (kW/m^2, also known as irradiance). To do this best requires a research of the type of solar cell to use (monocrystalline, polycrystalline, etc), magnetic declination map and Sun chart to track the sun as it moves through out the day in the area of the 2000 watt power inverter installation.
For an existing system, you are pretty much given a hand you have to deal with. If you're not given the "expected" result that was calculated then go back to the designer/engineer (of course make sure surface is not dirty, GoHz 2000w power inverter is working properly, and ensure there are no loose connections). Otherwise you may look to replac the system with a better solar cell material (does not ensure you'll get the expected efficiency/power output desired) or expand the existing pure sine wave inverters.
From what I am reading there are many solar plants being built with great promises of high output, great return on investment and profit making in no time. From what I understand without subsidies the promised returns turn out to be far in excess of reality.