<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/132872>132872</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            Multi-Output Operations Numerical Validation in MLIR Python Bindings
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            mlir
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          rockysingh-mcw
      </td>
    </tr>
</table>

<pre>
    I have implemented a VarMeanOp in MLIR that computes variance and mean as outputs. I have successfully lowered this operation to LLVM and am now testing it using MLIR Python bindings as part of the numerical validation process.

Problem:
When dealing with single-output operations, I can correctly validate the results using Python bindings by creating a memref and invoking the operation using ExecutionEngine.invoke(). This works by leveraging APIs such as:

```
get_memref_descriptor()
make_nd_memref_descriptor(rank, type)

```However, for multi-output operations, I am encountering issues.

What I Have Tried:

I created two separate memref descriptors for the two outputs (variance and mean).
I attempted to pass these memrefs to engine.invoke() to capture both results.
Unfortunately, I am unable to retrieve the correct outputs when invoking the operation.

Single output (works correctly)
```
res_memref = make_nd_memref_descriptor(rank, element_type)
engine.invoke("single_output_op", input_memref, res_memref)

```
Multiple outputs (not working)
```
var_memref = make_nd_memref_descriptor(rank, element_type)
mean_memref = make_nd_memref_descriptor(rank, element_type)
engine.invoke("multi_output_op", var_memref, mean_memref,input_memref)
```
Unable to get the correct results
How can I correctly handle multiple outputs in ExecutionEngine.invoke() when testing multi-output operations in Python?
Is there a recommended approach to create and pass multiple memrefs for capturing multiple outputs?
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJy0VU1v4zYQ_TX0ZRBDoaw4Pujg_TBiIGmCdjd7NChqLLGmSIGk7PrfF0PKH7veLQoUBRwEFMg3b968mRHeq8Yglqz4wIpPEzGE1rrSWbk7emWa9q6Th0ll62O5hlbsEVTXa-zQBKxBwLtwLyjMaw_KwMvz-ncIrQggbdcPAT3shVPCSARhauhQGBAe7BD6IfgpjJB-kBK93w5aH0HbAzqsIbTKg-3RiaCsgWDh-fn9JeKIDow9QEAflGlABRiIa4r_dgytNVApUyvTeIrXCxfAbiG0CGbo0CkpNOyFVnUC750lAlOWLVm2fHO20tixnA7fWjRQo9AU4KBCCxRK411K4sLQM_4R1iCFAWmdQxn08RQDY2iHftDBj2R_5FkdQToUMSMBHXYOtzFbZfZ2R18J4yJIQvn8F8qBzp9NowxO42Vk_JHxxRS-kIYH63YRXuMenWjo2fJt7Un2FoRPedLvIRt_2bLBsEkcNjV66VQfrEuwLFt2YocbU__0hhNmR1KEY4_p9jX0kz0QC7qwtQ66QQf1KylFB2ikHUxAF-vs_YCnIn0jm63hifzzxSmsz2msk45koYMFj71wVIFR0QtXHxmQqHRv9CQw_njjWdIyAosQsOsjtIVeeE_P_Qnb01e8rQN9lqIPg0OobGhPRiDMr2ZrXRiMCKiP57QHIyqN9M5hcAr3yUCjr85cD-TNn9tjVOmP6NXxAeWWzHA26Figq8I79GNZgeWf4N9UGtM82FwqfisCT12zSUw2tmec01tl6JjQ6XwJf-sdli1fyC_9OaNYLmND9LgyzW06e-H-ezrkgf9FlGj_G00unOl0FZzxj9_r9WO2X8--aTB8Z5nRcixbPtlDHFLrqzHVClNrTN14ra4y_zhgkgFPY_gXvUwgadaxfEVdFJvGIQhwKG3XoalplfS9s0K2sVli_8b2i0125nXqM2rc1FHnwFe0Wb6a1GVeL_KFmGB5P5_xx_lsNs8nbcmrWcEXcvvwgNv7WbEo8irfLu6r2SKX84fHbKJKnvEiy-lvNi_4NH8Qcz6bb3nG74si42yWYSeUnmq976bWNZM4l8r7nD_O-USLCrWP65TzTitHRS0-TVxJ9--qofFslmnlg78gBBU0ltHcd69JwNeLgL-dV9b7ZWWd1u24Rj6Ma2QyOF22IfRxrPMV46tGhXaoptJ2jK8o5Pjvrnf2T5SB8VWarIyvxiT2Jf87AAD__5GHy_g">