MRArbGrad 1.1.4

Gradient waveform design tool for arbitrary k-space trajectories.

Magnetic Resonance Arbitrary Gradient Toolbox (MRArbGrad, MAG)

Introduction

This toolbox is a pip package with C++ backend. The pip package can be called via Python interface to generate non-Cartesian gradient waveforms for built-in and external trajectories. The C++ source code (in mrarbgrad_src/ext/) can be ported to other pulse sequence project like UIH's Adept project for gradient waveform calculation.

Install

Optionally, to create a new conda environment (in case the dependencies in this package break your current environment), please run:

$ conda create -n magtest -y
$ conda activate magtest
$ conda install python==3.10 -y

This package is NOT restricted to use Python 3.10 or numpy 1.26 (as specified in the requirements.txt). Feel free to adjust at your convenience, just if the package works.

To install the pip package of the proposed algorithm (including the trajectory library built on it), and also the dependencies:

$ pip install -r requirements.txt
$ bash install.bash

It's the best practice to use my script install.bash for installation. You can also install via pip install . but remember to delete *.egg-info or pip will run into bug when uninstalling this package in current folder (see comments in install.bash).

Examples & Usages

Examples for generating gradient waveforms for either built-in trajectory (trajectory library) or external trajectory (expressed by trajectory function or trajectory samples) can be found in the example folder.

Citation

If this project helps you, please cite our paper:

[1] R. Luo, H. Huang, Q. Miao, J. Xu, P. Hu, and H. Qi, “Real-Time Gradient Waveform Design for Arbitrary k-Space Trajectories,” Sep 9, 2025, arXiv preprint arXiv:2507.21625.