Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

USGS uses machine learning to show large lithium potential in Arkansas

usgs.gov

Post reply on HN
Print view View original

191–200 of 237 posts

Re: USGS uses machine learning to show large lithium potential in Arkansas

#191

Post by ZeroGravitas »

shmageggy wrote: post #179
MrBuddyCasino wrote: post #164

Earlier quoted context omitted.

[flagged]

Actually it looks like their arguments are presented entirely in terms of tradeoffs. They argue that the carbon benefit from electric cars (cited as very far down the list on e.g. https://drawdown.org/solutions/table-of-solutions ) isn’t worth the cost to biodiversity, water use and pollution, cultural values and history, peacefulness and tranquility, etc. https://www.protectthackerpass.org/mining-lithium-at-thacker.…

That list is only scale (e.g. 40 Gigatons saved by onshore wind or utility solar by 2050) and even on that measure EVs do pretty well at 10 Gigatons.

But they do even better if you consider cost since the TCO of many electric vehicle classes is lower than the alternative, so you save money and carbon.

These tradeoffs are displayed on a marginal abatement cost curve:

https://www.edf.org/revamped-cost-curve-reaching-net-zero-em...

> $0 per ton or less

> Technologies: Many measures in the power and transportation sectors are cost-effective right now, including several electric vehicle classes, electric efficiency, high-quality solar PV and onshore wind resources, and nuclear relicensing. The use of heat pumps in buildings is also available.

> Emissions: Together, the measures in this range represent more than 1 gigaton of potential annual emission reductions by 2050 or 22% of way toward net-zero carbon emissions by 2050.

Re: USGS uses machine learning to show large lithium potential in Arkansas

#192

Post by jncfhnb »

tomrod wrote: post #84
folli wrote: post #21

From the paper's method section, a bit more about which type of ML algo was used: An RF machine-learning model was developed to predict lithium concentrations in Smackover Formation brines throughout southern Arkansas. The model was developed by (i) assigning explanatory variables to brine samples collected at wells, (ii) tuning the RF model to make predictions at wells and assess model performance, (iii) mapping spa…

RF is a heavy hitter when it comes to tabular data. XGBoost is good as well, but more often than not needs and autotuner to really unlock it (e.g pycaret).

XGBoost models are random forest models. They’re also just consistently better for very little effort.

Re: USGS uses machine learning to show large lithium potential in Arkansas

#193

Post by prog_1 »

jncfhnb wrote: post #192
tomrod wrote: post #84

Earlier quoted context omitted.

RF is a heavy hitter when it comes to tabular data. XGBoost is good as well, but more often than not needs and autotuner to really unlock it (e.g pycaret).

XGBoost models are random forest models. They’re also just consistently better for very little effort.

you surely mean that both are ensemble models. RFs and GBMs differ in how they fit the data

Re: USGS uses machine learning to show large lithium potential in Arkansas

#194

Post by ZeroGravitas »

declan_roberts wrote: post #70
greenie_beans wrote: post #67

ugh i really don't want people to mine in the mobile basin. that's one of the most diverse ecosystems in north america. https://www.youtube.com/watch?v=8j9coyJeB4Q

It seems backwards, but pretty much the only fuel that protects ecosystems on a large scale are fossil fuels and nuclear. Global reforestation is almost entirely the result of households switching from wood to coal in the 20th century.

Contrarianism has definitely entered its self-parody era.

Re: USGS uses machine learning to show large lithium potential in Arkansas

#195

Post by bluedino »

EB66 wrote: post #48
silisili wrote: post #42

Earlier quoted context omitted.

Here's another list - Highest poverty rate? Lowest literacy rate? Last in opportunity? 8th worst in public safety? If you guessed California, you'd be right. Sweeping generalities and handpicked metrics do not tell an entire story.

You are citing the US News "best states" ranking. In that ranking, California is ranked #37 overall and Arkansas is ranked #47 overall. Even your own hand picked data source supports the OP...

When was the last time you heard someone clamoring to move to Nebraska or New Hampshire?

Re: USGS uses machine learning to show large lithium potential in Arkansas

#196

Post by Loic »

folli wrote: post #21

From the paper's method section, a bit more about which type of ML algo was used: An RF machine-learning model was developed to predict lithium concentrations in Smackover Formation brines throughout southern Arkansas. The model was developed by (i) assigning explanatory variables to brine samples collected at wells, (ii) tuning the RF model to make predictions at wells and assess model performance, (iii) mapping spa…

RF is random forest[0].

We had this discussion a couple of days ago: "Why do Random Forests Work? Understanding Tree Ensembles as Self-Regularizing Adaptive Smoothers".

https://arxiv.org/abs/2402.01502

https://news.ycombinator.com/item?id=41873968

[0]: https://en.wikipedia.org/wiki/Random_forest

Re: USGS uses machine learning to show large lithium potential in Arkansas

#197

Post by trimethylpurine »

corimaith wrote: post #169
trimethylpurine wrote: post #148

Earlier quoted context omitted.

Congress plays whack-a-mole with policy while importers shuffle goods or various parts of the manufacturing process to neighboring countries or follow tighter packaging constraints to avoid specific tax rules or earn specific tax incentives. Tariffs are political showmanship. It's not really a viable nor an enforceable option in the modern economy, at least not based on my experience in the industry.

And all that shuffling introduces more costs, creates negative externalities for exports does it not?

Maybe for certain complex manufacturing processes (microprocessors?), but for most goods, not really. India, Vietnam, etc. Same, same.

American labor and real estate (shelf space) are what make up the vast majority of the cost for goods purchased here.

Re: USGS uses machine learning to show large lithium potential in Arkansas

#198

Post by voidUpdate »

> "the amount of lithium present would meet projected 2030 world demand for lithium in car batteries nine times over"

Does that mean the entire field has enough lithium for the requirements of 2030, 9 times? Or in other words, it can supply the lithium needs of car batteries from 2030 to 2039? That's not particularly long...

Re: USGS uses machine learning to show large lithium potential in Arkansas

#199

Post by bastawhiz »

declan_roberts wrote: post #70
greenie_beans wrote: post #67

ugh i really don't want people to mine in the mobile basin. that's one of the most diverse ecosystems in north america. https://www.youtube.com/watch?v=8j9coyJeB4Q

It seems backwards, but pretty much the only fuel that protects ecosystems on a large scale are fossil fuels and nuclear. Global reforestation is almost entirely the result of households switching from wood to coal in the 20th century.

Lithium isn't a fuel

Re: USGS uses machine learning to show large lithium potential in Arkansas

#200

Post by bitmasher9 »

voidUpdate wrote: post #198

> "the amount of lithium present would meet projected 2030 world demand for lithium in car batteries nine times over" Does that mean the entire field has enough lithium for the requirements of 2030, 9 times? Or in other words, it can supply the lithium needs of car batteries from 2030 to 2039? That's not particularly long...

Lithium is infinitely recyclable. We cannot really predict lithium demand in 2039, because technology changes much.

Look at steel. Most of the steel used is recycled steel, we don’t mine a lot of it any more. If you asked someone 90 years ago, they would have assumed global steel demand would continue to rise.

Post reply on HN