I got to see a lot of matlab code where the following types of initialization were used like param.L and so on ....why cant we just use L ?

param.L = 3;   % number of elements in each linear combination.
param.K = 50; % number of dictionary elements
param.numIteration = 50; % number of iteration to execute the K-SVD algorithm.
N = 1500; % number of signals to generate
n = 20;   % dimension of each data
SNRdB = 20; % level of noise to be added
[param.TrueDictionary, D, x] = generateSyntheticDictionaryAndData(N, param.L, n, param.K, SNRdB);

Answer 1

[Birdman]

In this context, param refers to a struct and L and K represents its fields. By this way, it is easier to maintain the variables. Fields can have different type of datas. For instance, you can define yourself as a struct, and your properties like height, weight, hair color, name, surname, ... as its fields.

[Stephen23]

Moreover, using a struct is a really simple and neat way to hold lots of parameter values, and pass them to any functions as required: this is because a struct is convenient and simple to handle (just one variable), but allow clearly identifiable fieldnames for each parameter. You should use structs.

[Birdman]

Exactly.