Skip to content

Commit

Permalink
updates to paper to address issues raised in #12
Browse files Browse the repository at this point in the history
  • Loading branch information
deronsmith committed Nov 18, 2024
1 parent d6717ca commit b090911
Showing 1 changed file with 5 additions and 5 deletions.
10 changes: 5 additions & 5 deletions paper/paper.md
Original file line number Diff line number Diff line change
Expand Up @@ -44,11 +44,11 @@ collection or modeling.
Environmental Source Apportionment Toolkit (ESAT) is an open-source Python package that provides a flexible and
transparent workflow for source apportionment using NMF algorithms, developed to replace the EPA's Positive
Matrix Factorization version 5 (PMF5) application [@PMF5:2014; @Paatero:1999]. `ESAT` recreates the source apportionment
workflow of PMF5 including pre-post processing analytical tools, batch modeling, uncertainty estimations and customized
workflow of PMF5 including pre- and post-processing analytical tools, batch modeling, uncertainty estimations and customized
constraints. `ESAT` offers a simulator for generating datasets from synthetic profiles and contributions, allowing for
model output evaluation. The synthetic profiles can be randomly generated, use a pre-defined set of profiles, or be a
combination of the two. The random synthetic contributions can follow specified curves and value ranges. Running `ESAT`
using the synthetic datasets one is able to accurately assess ESAT's ability to find a solution that recreates the
combination of the two. The random synthetic contributions can follow specified curves and value ranges. By running `ESAT`
using the synthetic datasets, users are able to accurately assess `ESAT`'s ability to find a solution that recreates the
original synthetic profiles and contributions.

# Statement of Need
Expand Down Expand Up @@ -77,7 +77,7 @@ $$
Q = \sum_{i=1}^n \sum_{j=1}^m \bigg[ \frac{V_{ij} - \sum_{k=1}^K W_{ik} H_{kj}}{U_{ij}} \bigg]^2
$$
here $V$ is the input data matrix of features (columns=$M$) by samples (rows=$N$), $U$ is the uncertainty matrix of the
input data matrix, $W$ is the factor contribution matrix of samples by factors=$k$, $H$ is the factor profile of
input data matrix, $W$ is the factor contribution matrix of samples by factors=$K$, $H$ is the factor profile of
factors by features.

The `ESAT` versions of NMF algorithms convert the uncertainty $U$ into weights defined as $Uw = \frac{1}{U^2}$.
Expand Down Expand Up @@ -116,6 +116,6 @@ impacts a source apportionment solution.
This paper has been reviewed in accordance with EPA policy and approved for publication.
ESAT development has been funded by U.S. EPA. Mention of any trade names, products, or services does not convey, and
should not be interpreted as conveying, official EPA approval, endorsement, or recommendation. The views expressed in
this paper are those of the authors and do not necessarily represent the views or policies of the US EPA.
this paper are those of the authors and do not necessarily represent the views or policies of the U.S. EPA.

# References

0 comments on commit b090911

Please sign in to comment.