Скачать или смотреть Understanding Matlab and Fortran Precision Discrepancies in Scientific Calculations

Discover why you may encounter different results when running identical codes in `Matlab` and `Fortran`, and learn to fix precision issues in scientific calculations.
---
This video is based on the question https://stackoverflow.com/q/69575543/ asked by the user 'user3307971' ( https://stackoverflow.com/u/3307971/ ) and on the answer https://stackoverflow.com/a/69586647/ provided by the user 'user3307971' ( https://stackoverflow.com/u/3307971/ ) at 'Stack Overflow' website. Thanks to these great users and Stackexchange community for their contributions.

Visit these links for original content and any more details, such as alternate solutions, latest updates/developments on topic, comments, revision history etc. For example, the original title of the Question was: Matlab and fortran precision

Also, Content (except music) licensed under CC BY-SA https://meta.stackexchange.com/help/l...
The original Question post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license, and the original Answer post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license.

If anything seems off to you, please feel free to write me at vlogize [AT] gmail [DOT] com.
---
Understanding Matlab and Fortran Precision Discrepancies in Scientific Calculations

When coding for scientific computations, it's common to encounter discrepancies between different programming languages, particularly when working with Matlab and Fortran. Both languages are widely used among scientists and engineers, but slight differences in how they handle numerical precision can lead to unexpected results—even when the code appears identical.

In this post, we'll explore why these discrepancies occur and how to ensure that your computations in both languages yield consistent results.

The Problem: Inconsistent Results

Consider the following scenario: You have written the same code for both Matlab and Fortran, yet upon execution, you receive different outputs. For instance, in one execution, Matlab gives you:

[[See Video to Reveal this Text or Code Snippet]]

While in Fortran, you get:

[[See Video to Reveal this Text or Code Snippet]]

The unsettling realization is that the two programming environments can yield different results simply due to the ways they handle numerical precision.

Example Code

Here's a basic example that illustrates your scenario:

Matlab Code

[[See Video to Reveal this Text or Code Snippet]]

Fortran Code

[[See Video to Reveal this Text or Code Snippet]]

The Solution: Aligning Precision

Step 1: Understanding Data Types and Precision

One of the main reasons for discrepancies in output is the default precision settings for floating-point numbers in each language:

Matlab typically uses double precision by default for its computations.

Fortran allows you to specify precision using constructs such as REAL*8.

To align the precision settings in both languages, make sure you declare your variables for consistent types:

In Fortran, using REAL*8 ensures that your calculations are in double precision.

Step 2: Adjusting Precision in Matlab

If you want to achieve the same results as in Fortran, you can use the double(single(...)) function in Matlab to control the precision. Here’s how you do that:

Change assignments to bulk_strength, KAPPA, and other floating-point constants or variables by using double(single(...)). This converts values explicitly to a potential lower precision matching that of Fortran.

Avoid using 1.D0 format unless explicitly managing numerical precision in Fortran. Keep the calculations straightforward.

Updated Code Samples

Updated Fortran Code

[[See Video to Reveal this Text or Code Snippet]]

Updated Matlab Code

[[See Video to Reveal this Text or Code Snippet]]

Conclusion

Understanding and managing precision in programming is crucial for reliable scientific computations. By aligning the precision settings in Matlab and Fortran, you can eliminate discrepancies and achieve consistent results. Whether working with real*8 in Fortran or utilizing conversion functions in Matlab, always pay attention to how your data types might affect your computational results.

By applying these techniques, you should now be able to run your codes in both languages successfully and get the same output.