(Feature Request)It's time for the Deep Learning Toolbox to support a broader set of dlarray operations
郦歌
on 5 Nov 2025
Latest activity Edit by xingxingcui
on 6 Nov 2025
From my experience, MATLAB's Deep Learning Toolbox is quite user-friendly, but it still falls short of libraries like PyTorch in many respects. Most users tend to choose PyTorch because of its flexibility, efficiency, and rich support for many mathematical operators. In recent years, the number of dlarray-compatible mathematical functions added to the toolbox has been very limited, which makes it difficult to experiment with many custom networks. For example, svd is currently not supported for dlarray inputs.
This link (List of Functions with dlarray Support - MATLAB & Simulink) lists all functions that support dlarray as of R2026a — only around 200 functions (including toolbox-specific ones). I would like to see support for many more fundamental mathematical functions so that users have greater freedom when building and researching custom models. For context, the core MATLAB mathematics module contains roughly 600 functions, and many application domains build on that foundation.
I hope MathWorks will prioritize and accelerate expanding dlarray support for basic math functions. Doing so would significantly increase the Deep Learning Toolbox's utility and appeal for researchers and practitioners.
Thank you.
4 Comments
Time Descending
For the backpropagation of SVD function, I think we should figure out how to obtain 
based on 
. Maybe it is feasible to decompose the svd procedure as elemental operations and construct the dynamic computation graph of svd. Can mathworks transfer the codes of svd from pytorch to MATLAB utilizing the existing dlarray ?
based on . Maybe it is feasible to decompose the svd procedure as elemental operations and construct the dynamic computation graph of svd. Can mathworks transfer the codes of svd from pytorch to MATLAB utilizing the existing dlarray ?
I've seen some examples showing how to define custom operators that don't support automatic differentiation. For simple derivatives you can compute them yourself, but deriving complex gradients by hand can be much harder(tricky).😂
Thank you for the feedback. I'll pass this on to development. What can help is to understand which functions are the most important for you. This can help development prioritise what to add support for next.
You mention svd, for example. Did you need this for anything in particular?
Similarly, many commonly used basic functions such as interp2, eig, find, rank,diff,ismember,unique, etc. are not supported. I once wanted to perform image interpolation but there was no suitable function available. The only function with similar functionality, interp2, does not support dlarray, so I had to implement the image interpolation algorithm myself.
Sign in to participate
