Problem in mini batch gradient descent

#1

I am not able to understand the algo of mini batch gradient descent. The limits in loops and summation.
Please help me to understand that.

#2

i am assuming that u understand normal gradient descent agorithm.
Mini batch gradient descent is in which instead of sending entire data to update our parameters we send only some portions of it.for example if we have 100,000 data samles it will be really computationally expensive to update the parameters after calculating error on entire datasets so instead we send data in form of batches lets just say we make batch of 200 datapoints we calculate error and upgrade are paramaeters in the bases of these paramaters we keep sending these batches until are required number of iterations of sending data is complete or until we get required accuracy.

As to how to get these batches generally we take random batches from our entire data and then send
it.
or we can shuffle our data and send in our whole data into form of batches like send first two hundred then the next.