and more generally, informational output from the model. Note that ModSpec's OutputMatrix facility is too limited at the moment. In principle, any quantity in the VA code can be defined to be an OP_something, which is an informational quantity.
So ModSpec has to have an informational_output capability. .InformationalOutputNames, .InformationalOutputs(vecX, vecY, vecU, parms).
And then, DAEAPI needs to be augmented, too, similarly - looping into the devices to do the actual calculations.
One problem: in Verilog-A, ddx with respect to branch voltages like V(a,b) are actually illegal right now; only V(a) is allowed. It should be the opposite. What's more, you can't have derivatives with respect to currents! There is an old discussion here: https://groups.google.com/forum/?fromgroups=#!topic/cmc-verilog-a/uH6MrKo3ryQ. Contact Colin and scream.
and there should be levels of informational output. 0 = none, 1 = basic, 2 = worse, 3 = terrible.
and more generally, informational output from the model. Note that ModSpec's OutputMatrix facility is too limited at the moment. In principle, any quantity in the VA code can be defined to be an OP_something, which is an informational quantity.
So ModSpec has to have an informational_output capability. .InformationalOutputNames, .InformationalOutputs(vecX, vecY, vecU, parms).
And then, DAEAPI needs to be augmented, too, similarly - looping into the devices to do the actual calculations.
One problem: in Verilog-A, ddx with respect to branch voltages like V(a,b) are actually illegal right now; only V(a) is allowed. It should be the opposite. What's more, you can't have derivatives with respect to currents! There is an old discussion here: https://groups.google.com/forum/?fromgroups=#!topic/cmc-verilog-a/uH6MrKo3ryQ. Contact Colin and scream.
and there should be levels of informational output. 0 = none, 1 = basic, 2 = worse, 3 = terrible.