How to create an upper triangular matrix

Question

I need create upper triangular matrix given a set of values(the order is not importation). The matrix could be too large to input manually. np.triu only gives you the upper triangular of a existing matrix, not creating a new one.

I am doing some optimization to get the parameters of upper triangular cholesky root of covariance matrix. My parameters are upper triangular cholesky roots of some covariance matrix. To initializing, I need to put the parameter values in the position of upper triangular.

array([[ a,  b,  c],
       [ 0,  d,  e],
       [ 0,  0,  f]])

Sorry, what output are you hoping for? [From your [1,2,3],...[10,11,12] input, I mean.] — DSM, Apr 06 '17 at 00:57
I tend to make an empty matrix and fill it using the indices, as [here](http://stackoverflow.com/questions/20055493/numpy-convert-an-array-to-a-triangular-matrix). Would that work? — DSM, Apr 06 '17 at 01:08

score 4 · Accepted Answer · edited Jun 09 '19 at 03:01

I had the same question, so I didn't want to leave this one without an answer. Based on the suggestions by @DSM I created the function below:

import numpy as np

def create_upper_matrix(values, size):
    upper = np.zeros((size, size))
    upper[np.triu_indices(3, 0)] = values
    return(upper)

c = create_upper_matrix([1, 2, 3, 4, 5, 6], 3)
print(c)

This could be improved by adding functionality that infers the size of the required matrix from the data, but for now it relies on the user providing that information.

How to create an upper triangular matrix

1 Answers1

Linked