Lookup Table Error: Values of 'Breakpoin​tsForDimen​sion1' in 'Model' must be strictly monotonically increasing.

There are already many question related to this topic, I rad them through and couldn't find a solution to this issue. Forgive me if the question is repeating.

As you can see in the picture the lookup table has monotonically increasing values of both breakpoints.

The column values are increasing with the stepsize(2) until the end, i.e. 40 to 154 (58 entries)

The row values are increasing with the stepsize(5) until the end, i.e. 115 to 280 (34 entries)

(is it because the column and row stepsize are different?)

Can someone please help to understand why does it keep showing the same error? Any advise or help would be great!

Error message : Values of 'BreakpointsForDimension1' in 'Model_xyz' must be strictly monotonically increasing. The problem occurs at element 2. Change the breakpoints data or use the fixed-point tools such as Auto Scaling or Fixed-Point Advisor to determine a different parameter dialog type.

[AKASH DEEP]

Hi,

I have the data which is decreasing, then what should I do it is showing the same error.

[4.1834 4.0550 3.9446 3.8467 3.7565 3.6814 3.6449 3.6154 3.5577 3.4818] this is my break point data

Answer 1

[Andy Bartlett]

Please check the data type specified for the breakpoints.

Quantization of the original value to the specified run-time data type could cause repeated values.

Example 1 Parameter type is int8

vOrig1 = 115:5:135
dtName1 = 'int8';
vQuantized1 = cast(vOrig1,dtName1)

outputs

vOrig1 =
   115   120   125   130   135
vQuantized1 =
  1×5 int8 row vector
   115   120   125   127   127

Notice that the last two values are equal and thus not monotonically increasing.

Example 2 Parameter type Fixed-Point

Orig2 = 40:2:46
nt2 = numerictype(1,8,2)
vQuantized2 = fi(vOrig2,nt2)

outputs

vOrig2 =
    40    42    44    46
nt2 =

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 2
vQuantized2 = 
   31.7500   31.7500   31.7500   31.7500

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 8
        FractionLength: 2

Notice that all the values are equal and thus not monotonically increasing.

Example 3 Parameter type Single

vOrig3 = 2^25 + (1:5)
dtName3 = 'single';
vQuantized3 = cast(vOrig3,dtName3)

outputs

vOrig3 =
    33554433    33554434    33554435    33554436    33554437
vQuantized3 =
  1×5 single row vector
    33554432    33554432    33554436    33554436    33554436

Notice that the first two values are equal and the last three values are equal, thus the vector of breakpoints is not monotonically increasing.

[Garvit Amipara]

You are right, thank you for the accurate explaination!

[Garvit Amipara]

For anyone having the similar problem as mentioned above, it is possible that there are better solutions.

Change the Input Data Type from the properties of the n-D Lookuptable block. The matrix and vector values in my files were 'double' . One way would be to change the data type. I had less time and was afraid that the original values might change while trying to change the data type with any commands. So I did not touch the matrix and vectors. Just by choosing 'int32' for the table and breakpoint 'input data type' worked.

[Taimoor Khan Mehmand]

@Garvit Amipara.

Hi Garvit! May I know which version of the MATLAB/Simulink were you using when you encountered this error? Actually, I encountered this problem when I used my Simulink model that I built in 2018a in the 2021a version.

Secondly, in your last comment where you mentioned and I quote ( Just by choosing 'int32' for the table and breakpoint 'input data type' worked. ). I am unable to find the " input data type " option in the properties of of n-D Lookuptable block. Can you please share the screen shot of the properties window etc ?

I have attached the screen shots of the windows of the Lookuptable block.

[Andy Bartlett]

Like most blocks, the Lookup Table block does not set the data types for its inputs.

It inherits them from the block upstream that outputs the signal that become the Lookup's input.

Hint 1: Turn on Display Port Data Types to make it easier investigate type related issues.

Hint 2: If you don't understand a type from it's name such as 'sfix16_En3', just feed the name to numerictype

nt = numerictype('sfix16_En3')
nt =

          DataTypeMode: Fixed-point: binary point scaling
            Signedness: Signed
            WordLength: 16
        FractionLength: 3

You can also look at some key attributes of the type

minMaxRepresentable = nt.range
representableValueSpacingIntegerOrFixedPoint = nt.Slope

minMaxRepresentable = 
                     -4096                  4095.875
      numerictype(1,16,3)

representableValueSpacingIntegerOrFixedPoint =
                     0.125