Serving the Quantitative Finance Community

 
User avatar
tagoma
Posts: 18587
Joined: February 21st, 2010, 12:58 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 10:03 am

Aliter, a small number of authors is also not a panacea.

https://arxiv.org/pdf/1706.04702.pdf
This paper is quite technical at a glance. What it is that you don't like about it, please?
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 11:11 am

Aliter, a small number of authors is also not a panacea.

https://arxiv.org/pdf/1706.04702.pdf
This paper is quite technical at a glance. What it is that you don't like about it, please?
1. “I do not mind if you think slowly, but I do object when you publish more quickly than you think.”
2. "Das ist nicht nur nicht richtig; es ist nicht einmal falsch!"

3. Image
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 11:20 am

4. No numerical examples, its is a paper model.

For the record, both JohnleM and moi contacted them on at least THREE occasions and no response.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
tagoma
Posts: 18587
Joined: February 21st, 2010, 12:58 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 3:41 pm

4. No numerical examples, its is a paper model.

For the record, both JohnleM and moi contacted them on at least THREE occasions and no response.
Thank you for your flowery comments. It is indeed frustrating when researchers don't reply to questions on their work. I mean while uthors email address written down at the start of the paper (strongly) suggest they are opened to discussion.
 
User avatar
Paul
Posts: 11261
Joined: July 20th, 2001, 3:28 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 5:36 pm

Aliter, a small number of authors is also not a panacea.

https://arxiv.org/pdf/1706.04702.pdf
This paper is quite technical at a glance. What it is that you don't like about it, please?
They clearly have a shallow belief in deep learning if they think anyone or anything can be fooled by (at) in an email address.
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 5:53 pm

Even worse, they are civil servants and the tax payer is footing the bill. I hope you won't invite them to your QUANT conferences, never ever.

And it's not like these guys didn't come down in the last shower. Publish or perish revisited?
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
JohnLeM
Posts: 515
Joined: September 16th, 2008, 7:15 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 8:54 pm

Aliter, a small number of authors is also not a panacea.

https://arxiv.org/pdf/1706.04702.pdf
This paper is quite technical at a glance. What it is that you don't like about it, please?
These authors come from a quite serious institution (ETH Zurich). They issued maybe 50 papers concerning ML methods for PDE. None of these papers succeeded showing that ML methods for PDE converges. Indeed, they do not converge. Their method just fail, even in the first dimensional case.

I am personnaly angry against them : they claimed solving a problem, the curse of dimensionality, that we solved years before them. 

Above personal griefs, I just don't understand how they could have advertized so much this method without taking the slightest mathematical precaution. How many reseachers among us followed their paths and wasted their time and money ? 
 
User avatar
tagoma
Posts: 18587
Joined: February 21st, 2010, 12:58 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 9:06 pm

Aliter, a small number of authors is also not a panacea.

https://arxiv.org/pdf/1706.04702.pdf
This paper is quite technical at a glance. What it is that you don't like about it, please?
These authors come from a quite serious institution (ETH Zurich). They issued maybe 50 papers concerning ML methods for PDE. None of these papers succeeded showing that ML methods for PDE converges. Indeed, they do not converge. Their method just fail, even in the first dimensional case.

I am personnaly angry against them : they claimed solving a problem, the curse of dimensionality, that we solved years before them. 

Above personal griefs, I just don't understand how they could have advertized so much this method without taking the slightest mathematical precaution. How many reseachers among us followed their paths and wasted their time ? These methods are a pure waste of time and money for a lot of people.
Hello JohnLeM. Thank you for your reply.
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 9:18 pm

They issued maybe 50 papers concerning ML methods for PDE. None of these papers succeeded showing that ML methods for PDE converges. Indeed, they do not converge. Their method just fail, even in the first dimensional case.

I didn't even get past the PDE part!

What really did it for me was calling it "Deep Galerkin Method" Meshuggah.
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 23rd, 2021, 9:23 pm

What you need to do JLM, is take 1-factor (not 100 factor) Black Scholes and work it out in great detail. I know that the Gallic maths style is top-down and love of generalisation but give a concrete case for us mortals.

I think it is necessary and sufficient.


Like the fella said,

The heart of mathematics consists of concrete examples and concrete problems. Big general theories are usually afterthoughts based on small but profound insights; the insights themselves come from concrete special cases.”
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
JohnLeM
Posts: 515
Joined: September 16th, 2008, 7:15 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 24th, 2021, 9:00 am

What you need to do JLM, is take 1-factor (not 100 factor) Black Scholes and work it out in great detail. I know that the Gallic maths style is top-down and love of generalisation but give a concrete case for us mortals.

I think it is necessary and sufficient.
I did it numerically for the 2D case, and there are already some theoretical results around there showing non convergence. I wonder however if I should push the tests. The main benchmark can be found here, slide 10.
 
User avatar
Cuchulainn
Posts: 64394
Joined: July 16th, 2004, 7:38 am
Location: Drosophila melanogaster
Contact:

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 24th, 2021, 11:20 am

What you need to do JLM, is take 1-factor (not 100 factor) Black Scholes and work it out in great detail. I know that the Gallic maths style is top-down and love of generalisation but give a concrete case for us mortals.

I think it is necessary and sufficient.
I did it numerically for the 2D case, and there are already some theoretical results around there showing non convergence. I wonder however if I should push the tests. The main benchmark can be found here, slide 10.
Write it up, in a 3-page summary. A counterexample. A bit like "bananas don't grow on trees".
"Compatibility means deliberately repeating other people's mistakes."
David Wheeler

http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
JohnLeM
Posts: 515
Joined: September 16th, 2008, 7:15 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 28th, 2021, 10:55 am

What you need to do JLM, is take 1-factor (not 100 factor) Black Scholes and work it out in great detail. I know that the Gallic maths style is top-down and love of generalisation but give a concrete case for us mortals.

I think it is necessary and sufficient.
I did it numerically for the 2D case, and there are already some theoretical results around there showing non convergence. I wonder however if I should push the tests. The main benchmark can be found here, slide 10.
Write it up, in a 3-page summary. A counterexample. A bit like "bananas don't grow on trees".
I wrote a more complete test. Would you like to open a dedicated thread to discuss it while pushing it to SSRN?
 
User avatar
bearish
Posts: 6430
Joined: February 3rd, 2011, 2:19 pm

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 28th, 2021, 6:18 pm

This feels like it deserves to be promoted from the OT into either Numerical Methods or Technical.
 
User avatar
katastrofa
Posts: 10063
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Advance Planning for Covid Redux: What We Had, What We Needed

March 29th, 2021, 11:33 am

"Why should current Covid-19 vaccines not be used for mass vaccination during a pandemic?"

The guy is a veterinary doctor specialising in virology. What he's basically saying is that that the current vaccines are not suitable for mass vaccination programmes during a lasting epidemic. He explains a common and simple mechanism called immune escape (in certain situations, the vaccine helps for the virus to find workarounds) and points at data which suggests that Covid indeed manages to do that.


I found a rebuttal of Bossche's claims (with a porr-style ad personam attack by the end). It is based on a perfect vision of how the current strategy will develop:
"we vaccinate as many people as we can and as quickly as we can. Cases and deaths go down. If new variants of concern that escape from the protection granted by the vaccines emerge, scientists reformulate the vaccines. This strategy is based on vaccines that target the adaptive arm of our immune system, the same principle that allowed us to slay smallpox and bring polio and measles down to their knees."

A few problems with that: cases don't go down; our brave scientists and vaccine quality control systems may not withstand the pressure. And finally we may forget the old biological laws governing our immune system, but discover/produce something entirely new - what a beautiful way to die.