- Theano - Discussion
- Theano - Useful Resources
- Theano - Quick Guide
- Theano - Conclusion
- Theano - Trivial Training Example
- Theano - Functions
- Theano - Shared Variables
- Theano - Variables
- Theano - Data Types
- Theano - Computational Graph
- Theano - Expression for Matrix Multiplication
- Theano - A Trivial Theano Expression
- Theano - Installation
- Theano - Introduction
- Theano - Home
Selected Reading
- Who is Who
- Computer Glossary
- HR Interview Questions
- Effective Resume Writing
- Questions and Answers
- UPSC IAS Exams Notes
Theano - Data Types
Now, that you have understood the basics of Theano, let us begin with the different data types available to you for creating your expressions. The following table gives you a partial pst of data types defined in Theano.
Data type | Theano type |
---|---|
Byte | bscalar, bvector, bmatrix, brow, bcol, btensor3, btensor4, btensor5, btensor6, btensor7 |
16-bit integers | wscalar, wvector, wmatrix, wrow, wcol, wtensor3, wtensor4, wtensor5, wtensor6, wtensor7 |
32-bit integers | iscalar, ivector, imatrix, irow, icol, itensor3, itensor4, itensor5, itensor6, itensor7 |
64-bit integers | lscalar, lvector, lmatrix, lrow, lcol, ltensor3, ltensor4, ltensor5, ltensor6, ltensor7 |
float | fscalar, fvector, fmatrix, frow, fcol, ftensor3, ftensor4, ftensor5, ftensor6, ftensor7 |
double | dscalar, dvector, dmatrix, drow, dcol, dtensor3, dtensor4, dtensor5, dtensor6, dtensor7 |
complex | cscalar, cvector, cmatrix, crow, ccol, ctensor3, ctensor4, ctensor5, ctensor6, ctensor7 |
The above pst is not exhaustive and the reader is referred to the tensor creation document for a complete pst.
I will now give you a few examples of how to create variables of various kinds of data in Theano.
Scalar
To construct a scalar variable you would use the syntax −
Syntax
x = theano.tensor.scalar ( x ) x = 5.0 print (x)
Output
5.0
One-dimensional Array
To create a one dimensional array, use the following declaration −
Example
f = theano.tensor.vector f = (2.0, 5.0, 3.0) print (f)f = theano.tensor.vector f = (2.0, 5.0, 3.0) print (f) print (f[0]) print (f[2])
Output
(2.0, 5.0, 3.0) 2.0 3.0
If you do f[3] it would generate an index out of range error as shown here −
print f([3])
Output
IndexError Traceback (most recent call last) <ipython-input-13-2a9c2a643c3a> in <module> 4 print (f[0]) 5 print (f[2]) ----> 6 print (f[3]) IndexError: tuple index out of range
Two-dimensional Array
To declare a two-dimensional array you would use the following code snippet −
Example
m = theano.tensor.matrix m = ([2,3], [4,5], [2,4]) print (m[0]) print (m[1][0])
Output
[2, 3] 4
5-dimensional Array
To declare a 5-dimensional array, use the following syntax −
Example
m5 = theano.tensor.tensor5 m5 = ([0,1,2,3,4], [5,6,7,8,9], [10,11,12,13,14]) print (m5[1]) print (m5[2][3])
Output
[5, 6, 7, 8, 9] 13
You may declare a 3-dimensional array by using the data type tensor3 in place of tensor5, a 4-dimensional array using the data type tensor4, and so on up to tensor7.
Plural Constructors
Sometimes, you may want to create variables of the same type in a single declaration. You can do so by using the following syntax −
Syntax
from theano.tensor import * x, y, z = dmatrices( x , y , z ) x = ([1,2],[3,4],[5,6]) y = ([7,8],[9,10],[11,12]) z = ([13,14],[15,16],[17,18]) print (x[2]) print (y[1]) print (z[0])
Output
[5, 6] [9, 10] [13, 14]Advertisements