Styjun

Redox reactions redefined

Crab Nebula short story from 1960s or '70s

Filtering fine silt/mud from water (not necessarily bacteria etc.)

Are L-functions uniquely determined by their values at negative integers?

Align by center of symbol

Too many spies!

Why do they not say "The Baby"

Is a public company able to check out who owns its shares in very detailed format?

Is killing off one of my queer characters homophobic?

Alternatives to using writing paper for writing practice

(algebraic topology) question about the cellular approximation theorem

Why does the Earth have a z-component at the start of the J2000 epoch?

What is the closed form of the following recursive function?

Do native speakers use ZVE or CPU?

Is this more than a packing puzzle?

What exactly is the Tension force?

Does entangle require vegetation?

What are some symbols representing peasants/oppressed persons fighting back?

Remove intersect line for one circle using venndiagram2sets

How can I legally visit the United States Minor Outlying Islands in the Pacific?

Why hasn't the U.S. government paid war reparations to any country it attacked?

What is the English equivalent of 干物女 (dried fish woman)?

What's the phrasal verb for carbonated drinks exploding out of the can after being shaken?

How long do Apple retain notifications to be pushed to iOS devices until they expire?

How to calculate sums across matrix diagonals in Tensorflow?

Calling a function of a module by using its name (a string)How to merge two dictionaries in a single expression?How to remove an element from a list by index?What is a clean, pythonic way to have multiple constructors in Python?How to leave/exit/deactivate a Python virtualenvDifference between numpy.array shape (R, 1) and (R,)How to print the value of a Tensor object in TensorFlow?Get the diagonal of a matrix in TensorFlowBlock Diagonal Matrices in Tensorflowtensorflow - multiply a vector of matrices against each matrix in another vector

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty margin-bottom:0;

2

Say, I have matrix 4x4 like:

1 2 3 4
5 6 7 8
4 3 2 1
8 7 6 5`

I want to get matrix 2*4-1 with elements like:

8
4+7
5+3+6
1+6+2+5
2+7+1
3+8
4

How can I do that in Tensorflow? With tensors, of course - I have tensor with shape [a,b,c,..,l,n,n] and want to get tensor with shape [a,b,c,...,l,2*n-1]. Is there any single function to do this or looping is the only way?

python tensorflow

share|improve this question

edited Mar 26 at 6:35

rx303

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

add a comment | 

2

Say, I have matrix 4x4 like:

1 2 3 4
5 6 7 8
4 3 2 1
8 7 6 5`

I want to get matrix 2*4-1 with elements like:

8
4+7
5+3+6
1+6+2+5
2+7+1
3+8
4

How can I do that in Tensorflow? With tensors, of course - I have tensor with shape [a,b,c,..,l,n,n] and want to get tensor with shape [a,b,c,...,l,2*n-1]. Is there any single function to do this or looping is the only way?

python tensorflow

share|improve this question

edited Mar 26 at 6:35

rx303

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

add a comment | 

Say, I have matrix 4x4 like:

1 2 3 4
5 6 7 8
4 3 2 1
8 7 6 5`

I want to get matrix 2*4-1 with elements like:

8
4+7
5+3+6
1+6+2+5
2+7+1
3+8
4

How can I do that in Tensorflow? With tensors, of course - I have tensor with shape [a,b,c,..,l,n,n] and want to get tensor with shape [a,b,c,...,l,2*n-1]. Is there any single function to do this or looping is the only way?

python tensorflow

share|improve this question

edited Mar 26 at 6:35

rx303

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

1 2 3 4
5 6 7 8
4 3 2 1
8 7 6 5`

8
4+7
5+3+6
1+6+2+5
2+7+1
3+8
4

share|improve this question

edited Mar 26 at 6:35

rx303

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

share|improve this question

edited Mar 26 at 6:35

rx303

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

asked Mar 26 at 6:26

rx303rx303

3144 bronze badges

1 Answer
1

active

oldest

votes

1

You can use tf.py_func to wrap a numpy function.

import tensorflow as tf
import numpy as np

def np_all_trace_sum(a):
 n = a.shape[-1]
 all_trace_sum = [a.trace(i,axis1=-1,axis2=-2) for i in range(n-1,-n,-1)] # shape = (2*n-1,a,b,c,..,l)
 return np.moveaxis(all_trace_sum,0,-1) # shape = (a,b,c,..,l,2*n-1)

A = tf.placeholder(shape=[None,None,4,4],dtype=tf.float32)
result = tf.py_func(np_all_trace_sum, [A], tf.float32)

a = np.array([[1,2,3,4],[5,6,7,8],[4,3,2,1],[8,7,6,5]])

with tf.Session() as sess:
 print(sess.run(result,feed_dict=A:[[a,a,a],[a,a,a]]))

[[[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]

 [[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]]

share|improve this answer

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

add a comment | 

Your Answer

StackExchange.ifUsing("editor", function ()
StackExchange.using("externalEditor", function ()
StackExchange.using("snippets", function ()
StackExchange.snippets.init();
);
);
, "code-snippets");

StackExchange.ready(function()
var channelOptions =
tags: "".split(" "),
id: "1"
;
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function()
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled)
StackExchange.using("snippets", function()
createEditor();
);

else
createEditor();

);

function createEditor()
StackExchange.prepareEditor(
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader:
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
,
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
);



);

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);

Sign up using Google

Sign up using Facebook

Sign up using Email and Password

Post as a guest

Name

Email

Required, but never shown

StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f55350988%2fhow-to-calculate-sums-across-matrix-diagonals-in-tensorflow%23new-answer', 'question_page');

);

Post as a guest

Name

Email

Required, but never shown

1

You can use tf.py_func to wrap a numpy function.

import tensorflow as tf
import numpy as np

def np_all_trace_sum(a):
 n = a.shape[-1]
 all_trace_sum = [a.trace(i,axis1=-1,axis2=-2) for i in range(n-1,-n,-1)] # shape = (2*n-1,a,b,c,..,l)
 return np.moveaxis(all_trace_sum,0,-1) # shape = (a,b,c,..,l,2*n-1)

A = tf.placeholder(shape=[None,None,4,4],dtype=tf.float32)
result = tf.py_func(np_all_trace_sum, [A], tf.float32)

a = np.array([[1,2,3,4],[5,6,7,8],[4,3,2,1],[8,7,6,5]])

with tf.Session() as sess:
 print(sess.run(result,feed_dict=A:[[a,a,a],[a,a,a]]))

[[[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]

 [[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]]

share|improve this answer

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

add a comment | 

1

You can use tf.py_func to wrap a numpy function.

import tensorflow as tf
import numpy as np

def np_all_trace_sum(a):
 n = a.shape[-1]
 all_trace_sum = [a.trace(i,axis1=-1,axis2=-2) for i in range(n-1,-n,-1)] # shape = (2*n-1,a,b,c,..,l)
 return np.moveaxis(all_trace_sum,0,-1) # shape = (a,b,c,..,l,2*n-1)

A = tf.placeholder(shape=[None,None,4,4],dtype=tf.float32)
result = tf.py_func(np_all_trace_sum, [A], tf.float32)

a = np.array([[1,2,3,4],[5,6,7,8],[4,3,2,1],[8,7,6,5]])

with tf.Session() as sess:
 print(sess.run(result,feed_dict=A:[[a,a,a],[a,a,a]]))

[[[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]

 [[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]]

share|improve this answer

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

add a comment | 

You can use tf.py_func to wrap a numpy function.

import tensorflow as tf
import numpy as np

def np_all_trace_sum(a):
 n = a.shape[-1]
 all_trace_sum = [a.trace(i,axis1=-1,axis2=-2) for i in range(n-1,-n,-1)] # shape = (2*n-1,a,b,c,..,l)
 return np.moveaxis(all_trace_sum,0,-1) # shape = (a,b,c,..,l,2*n-1)

A = tf.placeholder(shape=[None,None,4,4],dtype=tf.float32)
result = tf.py_func(np_all_trace_sum, [A], tf.float32)

a = np.array([[1,2,3,4],[5,6,7,8],[4,3,2,1],[8,7,6,5]])

with tf.Session() as sess:
 print(sess.run(result,feed_dict=A:[[a,a,a],[a,a,a]]))

[[[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]

 [[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]]

share|improve this answer

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

import tensorflow as tf
import numpy as np

def np_all_trace_sum(a):
 n = a.shape[-1]
 all_trace_sum = [a.trace(i,axis1=-1,axis2=-2) for i in range(n-1,-n,-1)] # shape = (2*n-1,a,b,c,..,l)
 return np.moveaxis(all_trace_sum,0,-1) # shape = (a,b,c,..,l,2*n-1)

A = tf.placeholder(shape=[None,None,4,4],dtype=tf.float32)
result = tf.py_func(np_all_trace_sum, [A], tf.float32)

a = np.array([[1,2,3,4],[5,6,7,8],[4,3,2,1],[8,7,6,5]])

with tf.Session() as sess:
 print(sess.run(result,feed_dict=A:[[a,a,a],[a,a,a]]))

[[[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]

 [[ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]
 [ 8. 11. 14. 14. 10. 11. 4.]]]

share|improve this answer

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges

answered Mar 26 at 9:44

giser_yuganggiser_yugang

4,14722 gold badges99 silver badges3030 bronze badges