Quick start
=====================================================

The main function of the ``Pychop`` is the method ``Chop``, which is loaded by 

.. code:: python

    from pychop import Chop



    
``Pychop`` supports NumPy (default), JAX, Torch, and TensorFlow as backend for simulation. Before performing the quantization, to set backend:

.. code:: python

    pychop.backend('numpy') # use NumPy as backend, other options: 'torch', 'jax', and 'tensorflow'



.. note::

    Users do not need to specify the backend for precision emulation. By using ``pychop.backend("auto")`` (which is the default setting), ``Pychop`` will automatically detect the required backend

    The difference between setting backend and without setting backend is that the speed of the first run is different, others are almost identical:

    .. code-block:: python

        import pychop
        from pychop import Chop
        import numpy as np
        from time import time

        pychop.backend('numpy', 1) # Specify different backends, e.g., jax and torch
        np.random.seed(0)
        X = np.random.randn(5000, 5000) 

        ch = Chop(exp_bits=5, sig_bits=10, rmode=3) # half precision

        st = time()
        X_q = ch(X)
        et = time()

        print("runtime: ", et - st)
        
    .. code-block:: bash

        Load NumPy backend.
        runtime:  0.65281081199646

    .. code-block:: python

        import pychop
        from pychop import Chop
        import numpy as np
        from time import time

        X = np.random.randn(5000, 5000) 

        ch = Chop(exp_bits=5, sig_bits=10, rmode=3) # half precision

        st = time()
        X_q = ch(X)
        et = time()

        print("runtime: ", et - st)
        print(X_q[:10, 0])


    .. code-block:: bash

        runtime:  0.9031667709350586