How to use exp() and powers ^ in uncertain matrixes?

I am trying to represent a thermistor using the Robust Control Toolbox.

I've created ureal objects to represent the thermistors base tolerance and beta and now want to create a matrix that represents the resistance.

ntcBase = ureal('ntcBase',2200,'Percentage',5)

ntcBeta = ureal('ntcBeta',3500,'Percentage',1)

ntcR = exp(ntcBeta)

I'm having issues getting exponentials and powers working in umats where the power is a ureal object and get the following error.

untitled

ntcBase =

Uncertain real parameter "ntcBase" with nominal value 2.2e+03 and variability [-5,5]%.

ntcBeta =

Uncertain real parameter "ntcBeta" with nominal value 3.5e+03 and variability [-1,1]%.

Check for incorrect argument data type or missing argument in call to function 'exp'.

Error in untitled (line 4)

ntcR = exp(ntcBeta)

Are these operators supported with these objects, or am I missing something else?

Thanks for the help

Answer 1

[Steven Lord]

ntcBase = ureal('ntcBase',2200,'Percentage',5)

Uncertain real parameter "ntcBase" with nominal value 2.2e+03 and variability [-5,5]%.

Let's see what operations are defined for this class.

methods(ntcBase)

Methods for class ureal:

actual2normalized  end                getBlockValue      inv                issiso             mrdivide           pid                series             squeeze            usample            
append             feedback           getLimits          iosize             issparse           mtimes             pid2               set                ss                 uscale             
blkdiag            fliplr             getNominal         ipermute           isstatic           nblocks            pidstd             setBlockValue      stack              uss                
cat                flipud             getTunedValue      isGeneralized      length             ndims              pidstd2            setTunedValue      tf                 usubs              
conj               frd                getValue           isLinear           lft                nmodels            plus               setValue           transpose          vertcat            
connect            gainsurfdata       gridureal          isParametric       lftdata            normalize          replaceBlock       showBlockValue     ufrd               voidModel          
ctranspose         genfrd             hasInternalDelay   isUncertain        makeStateSpace     normalized2actual  repmat             showTunable        umat               wcnorm             
display            genmat             hasdelay           isempty            minus              numel              reshape            simplify           uminus             
double             genss              horzcat            isfinite           mldivide           parallel           rsampleBlock       size               uplus              
dss                get                imp2exp            isreal             mpower             permute            sampleBlock        sminreal           ureal              

Static methods:

empty              matchChannelNames  

The list of methods for this type of object does not include exp so it is not defined for this class. To double-check, let's see what would be called if you ran the command you tried:

which -all exp(ntcBase)

'exp(ntcBase)' not found.

Compare with the following, which correctly identifies the exp function for double arrays:

which -all exp(1)

built-in (/MATLAB/toolbox/matlab/elfun/@double/exp)  % double method

The power function (the .^ operator) is not listed in the methods list, but the mpower function (the ^ operator) is. Since ntcBase is a scalar, the two would perform equivalent operations.

y1 = ntcBase^2

Uncertain matrix with 1 rows and 1 columns.
The uncertainty consists of the following blocks:
  ntcBase: Uncertain real, nominal = 2.2e+03, variability = [-5,5]%, 2 occurrences

Type "y1.NominalValue" to see the nominal value and "y1.Uncertainty" to interact with the uncertain elements.

y1.NominalValue

ans = 4840000

2200^2

ans = 4840000

y2 = ntcBase.^2

Operator '.^' is not supported for operands of type 'ureal'.

I'm not sure why the class doesn't define power. [Robust Control Toolbox is not my area of expertise.] This seems like a reasonable enhancement request to file with Technical Support. I could see exp being a little bit trickier to implement, though.

[Dylan Bartlett]

Thanks for this well explained answer Steven. The reason why I was asking about powers was to try and get around exp() not working in the first instance.

I tried the following and got another, much different error.

>> y = exp(1)^ntcBase

Check for incorrect argument data type or missing argument in call to function 'iosize'.

Error in ^ (line 26)

[ny,nu] = iosize(M);

I think all these issues are linked though, with some missing functionality holding me back. I'll raise a request and see what comes of it.

[Steven Lord]

From a little experimentation it looks like the mpower operator for ureal objects expects the base to be a ureal and the exponent to be an integer valued scalar number. The mpower function is asking questions (with the iosize utility function) about the first input that make sense for a ureal but not for a double value like exp(1).

ntcBase = ureal('ntcBase',2200,'Percentage',5);
ntcBase ^ 0.5

Error using ^
In the expression "M^K", the exponent K must be a scalar integer.