How can I generate code for a Simulink model that uses exponents without enabling support for nonfinite numbers?

I have a Simulink model that requires the use of exponents for some of its calculations. I would like to generate code from the model, however for a variety of reasons I can't enable support for non-finite numbers. I've tried saturating the inputs to avoid negative numbers to non integer exponents, etc., creating my own math expression using the Fcn block but despite all of this Simulink seems to think that any exponent requires NaN support. Is there a way around this?

[Walter Roberson]

integer exponents or floating point?

[NJ2Cali]

Floating point.

I did some more digging into the model and it turns out that saturation helpeed, but there were other blocks in the code that required nonfinite number support.

I guess the follow up question is, is there a way to easily figure out which blocks are requiring nonfinite number support? As of now my approach is to generate code, look for NaN in the generated code, and work backwards from there. However, some functions (such as exponent) generate differently if NaN is supported even if they themselves don't require NaN support, so it's tricky to pin down the cause. I'm still working on it in my case.

[NJ2Cali]

I added the second half of my comment as an answer to my original question.

Answer 1

[NJ2Cali]

I did some more digging into the model and it turns out that saturation helped, but there were other blocks in the code that required nonfinite number support.

I guess the follow up question is, is there a way to easily figure out which blocks are requiring nonfinite number support? As of now my approach is to generate code, look for NaN in the generated code, and work backwards from there. However, some functions (such as exponent) generate differently if NaN is supported even if they themselves don't require NaN support, so it's tricky to pin down the cause. I'm still working on it in my case.

[Kiran K]

Hi,

>>However, some functions (such as exponent) generate differently if NaN is supported even if they themselves don't require NaN support,

Did you solve this problem or found any workaround?. I also face similar problem.

Also, how did you manage by saturating the inputs.

When I changed below options there was no impact in generated code of exp functions

• minimum and maximum values in Inport->Signal attributes
• using saturation block