Main Content

Replace a Custom Function with a Lookup Table

This example shows how to replace a custom function with a lookup table approximation function using the Fixed-Point Converter app.

Prerequisites

This example requires the following products:

Create Algorithm and Test Files

In a local, writable folder:

  1. Create a MATLAB function, custom_fcn.m which is the function that you want to replace.

    function y = custom_fcn(x)
        y = 1./(1+exp(-x));
    end
  2. Create a wrapper function, call_custom_fcn.m, that calls custom_fcn.m.

    function y = call_custom_fcn(x)
        y = custom_fcn(x);
    end
  3. Create a test file, custom_test.m, that uses call_custom_fcn.

    close all
    clear all
    
    x = linspace(-10,10,1e3);
    for itr = 1e3:-1:1
       y(itr) = call_custom_fcn( x(itr) );
    end
    plot( x, y );
    

Open the Fixed-Point Converter App

  1. Navigate to the work folder that contains the file for this example.

  2. On the MATLAB Toolstrip Apps tab, under Code Generation, click the app icon.

Select Source Files

  1. To add the entry-point function call_custom_fcn to the project, browse to the file call_custom_fcn.m, and then click Open. By default, the app saves information and settings for this project in the current folder in a file named call_custom_fcn.prj.

  2. Click Next to go to the Define Input Types step.

    The app screens call_custom_fcn.m for code violations and fixed-point conversion issues. The app opens the Review Code Generation Readiness page.

Review Code Generation Readiness

  1. Click Review Issues. The app indicates that the exp function is not supported for fixed-point conversion. You can ignore this warning because you are going to replace custom_fcn, which is the function that calls exp.

  2. Click Next to go to the Define Input Types step.

Define Input Types

  1. Add custom_test as a test file and then click Autodefine Input Types.

    The test file runs. The app determines from the test file that x is a scalar double.

  2. Click Next to go to the Convert to Fixed Point step.

Replace custom_fcn with Lookup Table

  1. Select the Function Replacements tab.

    The app indicates that you must replace the exp function.

  2. Enter the name of the function to replace, custom_fcn, select Lookup Table, and then click .

    The app adds custom_fcn to the list of functions that it will replace with a Lookup Table. By default, the lookup table uses linear interpolation and 1000 points. The app sets Design Min and Design Max to Auto which means that app uses the design minimum and maximum values that it detects by either running a simulation or computing derived ranges.

  3. Click the Analyze arrow , select Log data for histogram, and verify that the test file is call_custom_test.

  4. Click Analyze.

    The simulation runs. The app displays simulation minimum and maximum ranges on the Variables tab. Using the simulation range data, the software proposes fixed-point types for each variable based on the default type proposal settings, and displays them in the Proposed Type column. The Convert option is now enabled.

  5. Examine the proposed types and verify that they cover the full simulation range. To view logged histogram data for a variable, click its Proposed Type field. The histogram provides range information and the percentage of simulation range covered by the proposed data type.

Convert to Fixed Point

  1. Click Convert.

    The app validates the proposed types and generates a fixed-point version of the entry-point function, call_custom_fcn_fixpt.m.

  2. In the Output Files list, select call_custom_fcn_fixpt.m.

    The conversion process generates a lookup table approximation, replacement_custom_fcn, for the custom_fcn function. The fixed-point conversion process infers the ranges for the function and then uses an interpolated lookup table to replace the function. By default, the lookup table uses linear interpolation, 1000 points, and the minimum and maximum values detected by running the test file.

    The generated fixed-point function, call_custom_fcn_fixpt.m, calls this approximation instead of calling custom_fcn.

    function y = call_custom_fcn_fixpt(x)
        fm = get_fimath();
    
        y = fi(replacement_custom_fcn(x), 0, 16, 16, fm);
    end

    You can now test the generated fixed-point code and compare the results against the original MATLAB function. If the behavior of the generated fixed-point code does not match the behavior of the original code closely enough, modify the interpolation method or number of points used in the lookup table and then regenerate code.

Related Topics