# GradientOfUnstructuredDataSet VS PythonCalculator's gradient
**URL:** https://discourse.paraview.org/t/gradientofunstructureddataset-vs-pythoncalculators-gradient/3390
**Category:** ParaView Support
**Created:** [January 23, 2020, 3:10pm UTC](https://discourse.paraview.org/t/gradientofunstructureddataset-vs-pythoncalculators-gradient/3390 "2020-01-23T15:10:10Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@hasadi](https://discourse.paraview.org/u/hasadi)
#### Post date: [January 23, 2020, 3:10pm UTC](https://discourse.paraview.org/t/gradientofunstructureddataset-vs-pythoncalculators-gradient/3390/1 "2020-01-23T15:10:10Z")
I try to get gradient of a vector field in Paraview. However using GradientOfUnstructuredDataSet will generate different results than PythonCalculator’s gradient.
Here is my expression input for python calculator: gradient(inputs[0].PointData[‘Txj’])
---
### Author:  [@Kenneth\_Moreland](https://discourse.paraview.org/u/Kenneth_Moreland)
#### Post date: [January 23, 2020, 3:22pm UTC](https://discourse.paraview.org/t/gradientofunstructureddataset-vs-pythoncalculators-gradient/3390/2 "2020-01-23T15:22:44Z")
Try turning on the `Faster Approximation` option for the `Gradient Of Unstructured Data Set` filter. When I tried this and I turned that on, the results seemed to match more closely.
When computing gradients on a finite mesh, you are only able to estimate them using finite differences. And depending on how define the finite differences are defined, you can get different answers. By default, the `Gradient Of Unstructured Data Set` is doing some extra work to make a more accurate gradient.
---
### Author:  [@hasadi](https://discourse.paraview.org/u/hasadi)
#### Post date: [January 23, 2020, 3:29pm UTC](https://discourse.paraview.org/t/gradientofunstructureddataset-vs-pythoncalculators-gradient/3390/3 "2020-01-23T15:29:39Z")
Thanks for your answer,