Thanks d_stranz, sorry for replaying so late I did not get a notification of your answer to my mail. You are right, the applet runs really fast. For the calculation I use the TLM method which simplifies a lot everything. You basically actualize the values of each pixel having a source (field[source_position]=field[source_position]+sin(xxxx)) and then you go from pixel to pixel and their value is "spreaded" to the North, west, east and south pixels. You don't simulate each "ray" independently so it is basically not heavier having reflections or more sources. You did scare me with the one second
I'll check Qwt then.