MATLAB: Do I receive the warning “Parameter precision loss occurred for ‘InitialValue'” when I set the ‘DataType’ parameter of a signal to ‘single’ and I specify a non-zero ‘Initial value’ parameter of the signal

simulink

When I specify the 'DataType' property of a signal as 'single', I must enter a variable of type 'double' as the 'Initial value' parameter for this signal. This causes Simulink to convert the variable specified for the 'Initial value' from type 'double' to type 'single'. To reproduce this issue, follow these steps:

1 – Create a Simulink model with a Data Store Read block

2 – Open the Model Explorer and create a new signal

3 – Set the 'DataType' property of the signal to 'single'

4 – Set the 'Initial value' parameter to any non-integer value, for example you can use the value 73.3.

4 – Execute the model and notice the warning displayed in the MATLAB command window.

Best Answer

This enhancement has been incorporated in Simulink 7.2 (R2008b). For previous product releases, to work around this limitation, two solutions are available:

One solution is to set the 'Detect precision loss' diagnostic option to 'none'.

Another solution is to use a variable which will not lose its precision when converted from 'double' to 'single'. In the previous example, the value 73.3 loses precision. When it is converted to type 'single', it becomes 73.3000031. This result can be converted to type 'double' to be used as 'Initial value' parameter of the signal.

In the following line of MATLAB code, the example value of 73.3 is converted to type 'single'. The result (73.3000031) is then converted to 'double' to be used as 'Initial value' parameter of a signal:

double(single(73.3))

Related Solutions

MATLAB: How to specify that the DataType of the Simulink.Parameter object should be “double” in Simulink 7.3 (R2009a)

This enhancement has been incorporated in Release 2012a (R2012a). For previous product releases, read below for any possible workarounds:

Not being able to set the DataType of a Simulink.Parameter to ‘double’ is a known limitation with Simulink 7.3 (R2009a).

You can set the DataType of a Simulink.Parameter directly, but if you set it to ‘double’, you get the warning message “Switching 'double' data type to 'auto' for parameter object”, and DataType remains ‘auto’:

 >> p=Simulink.Parameter;
 >> p.Value=5;
 >> p.DataType='double';
 Warning: Switching 'double' data type to 'auto' for parameter object 
 >> sllastwarning
 ans = 
          Type: 'warning'
     MessageID: 'Simulink:SL_Param_SwitchDoubleDataTypeToAuto'
       Message: 'Switching 'double' data type to 'auto' for parameter object'
        Handle: [0x1 double]
 >> p.DataType
 ans =
 auto

If you get a different error, please let us know the Release, message text, and message id. You can avoid the warning by casting the Value instead of setting the DataType:

>> q=Simulink.Parameter;
>> q.Value=double(5);

You can also suppress the warning:

>> warning('off', 'Simulink:SL_Param_SwitchDoubleDataTypeToAuto')

and execute:

>> r=Simulink.Parameter;
>> r.Value=5;
>> r.DataType='double';

We have not seen any case in which a parameter Value of type double and parameter DataType ‘auto’ causes incorrect simulation results or incorrect code generation. Please let us know if you see such a problem.

Since R2007b, all blocks honor specifying the DataType of a Simulink.Parameter with a Simulink.AliasType:

>> warning('on', 'Simulink:SL_Param_SwitchDoubleDataTypeToAuto')
>> force_double=Simulink.AliasType;
>> force_double.BaseType
>> s=Simulink.Parameter;
>> s.Value=5;
>> s.DataType='force_double';

MATLAB: Providing Maximum Value as Initial Value for Single Precision Signal Causes Error due to Quantization

I had similar issues with this. I have requested the Mathworks to do an enhancement but have not seen it in place yet.

The problem arises when casting a number from double to single. The "precision loss" occurred and Simulink reports it as warning or error. It is true that the "precision loss" could happen but I think the software should give an option whether the user wants to ignore the double-to-single type cast precision loss.

What version of MATLAB are you using? The message seems to be different than what I saw in several versions of MATLAB/Simulink. Typically, you can go to Simulink configuration parameters, Diagnostics, Data Validity, Parameters, change the setting for "Detect precision loss". If you set it as "warning" or "none", you could pass the model update with or without even the warning messages.

You could also try to set the Initial value as single(10.3). In my case, the warning disappeared by doing this.

Best Answer

Related Solutions

MATLAB: How to specify that the DataType of the Simulink.Parameter object should be “double” in Simulink 7.3 (R2009a)

MATLAB: Providing Maximum Value as Initial Value for Single Precision Signal Causes Error due to Quantization

Related Question