Proceedings of the Platform for Advanced Scientific Computing Conference 2019
DOI: 10.1145/3324989.3325724
|View full text |Cite
|
Sign up to set email alerts
|

A Multivariate Approach to Ensure Statistical Reproducibility of Climate Model Simulations

Abstract: Effective utilization of novel hybrid architectures of pre-exascale and exascale machines requires transformations to global climate modeling systems that may not reproduce the original model solution bit-for-bit. Round-off level differences grow rapidly in these non-linear and chaotic systems. This makes it difficult to isolate bugs/errors from innocuous growth expected from round-off level differences. Here, we apply two modern multivariate two sample equality of distribution tests to evaluate statistical re… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
7
0

Year Published

2021
2021
2023
2023

Publication Types

Select...
3
2
1
1

Relationship

0
7

Authors

Journals

citations
Cited by 8 publications
(7 citation statements)
references
References 10 publications
0
7
0
Order By: Relevance
“…The verification methodology in this work shares some similarities with verification methodologies presented in previous studies, most notably Baker et al (2015Baker et al ( , 2016; Milroy et al (2018); Mahajan et al (2017Mahajan et al ( , 2019; Massonnet et al (2020).…”
Section: Verification Methodologymentioning
confidence: 90%
See 2 more Smart Citations
“…The verification methodology in this work shares some similarities with verification methodologies presented in previous studies, most notably Baker et al (2015Baker et al ( , 2016; Milroy et al (2018); Mahajan et al (2017Mahajan et al ( , 2019; Massonnet et al (2020).…”
Section: Verification Methodologymentioning
confidence: 90%
“…The methodology follows Livezey (1985): and combines Monte Carlo methods and subsampling to produce a distribution of rejection rates, which can be used to get the probability of having n rej rejections for two ensembles coming from the exact same model. Mahajan et al (2017) and Mahajan et al (2019) chose almost the same approach, but they produce the reference distribution by pooling two ensembles (from which they do not know yet whether they come from the same distribution) together and then applying the test to randomly drawn subsets from the pooled ensemble. This approach allows them to bypass the creation of a control ensemble and therefore save compute time.…”
Section: Verification Methodologymentioning
confidence: 99%
See 1 more Smart Citation
“…We refer to our resolution model as the "ultra-low resolution" (ULR) model, which corresponds to a 7.5° grid resolution in the atmosphere and a 240 km grid resolution for the ocean and sea ice. The ULR configuration of the E3SM was originally created primarily to enable rapid turnaround testing, and was recently used to develop an approach for ensuring statistical reproducibility of climate model simulations on a variety of conventional as well as hybrid architectures (Mahajan et al, 2019a(Mahajan et al, , 2019b. In contrast, our primary objective here was to investigate for the first time the ULR E3SM's skill as a physics-based surrogate of the fully coupled E3SM.…”
Section: Contributions and Organizationmentioning
confidence: 99%
“…They show that the climate extremes test based on GEV theory is considerably less sensitivity to changes in optimization strategies than the K-S test on mean values. Mahajan et al (2019) applied two relatively modern multivariate two sample equality of distribution tests, the energy test and the kernel test (see Mahajan et al, 2019, for details on the respective test statistics), on year-long ensemble simulations following Baker et al (2015) and Mahajan et al (2017).…”
Section: Current State Of the Artmentioning
confidence: 99%