SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 10th, 2020, 5:33 pm

The same idea as in economy, but here applied to the flows of people between places. In Ferguson's model the probability of transmitting the virus between two individuals is generated by kernel functions, whose values decrease with the distance between those individuals. I'm not sure why they call it a gravity model, but ok.
Maybe because it's based on gravity?? [$]O(1/r^2)[$] interaction.

And it is indeed mentioned in the code.
/**
 * @brief Holds microcells.
 *
 * Keeps track of susceptible, latent and infected people (in addition to details like who
 * is vaccinated, treated etc.) Also contains data for the spatial gravity model for social
 * interactions (probability distributions).
*/
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 10th, 2020, 5:49 pm

Question: Regarding computing probabilities with Kernel functions it is necessary to parallelise the loop computations? and lookup tables?
Last edited by Cuchulainn on May 10th, 2020, 6:11 pm, edited 1 time in total.
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 10th, 2020, 6:11 pm







So, weighing all of the above: yes,--  I am still a big fan. 
I think a big problem is that there is no one(?) in Congress with STEM degree. So, being comfortable with numbers is non-existent.
Same with leaders who studied Greek and the Classics. 

So, how do they link cause and effect?

The consequences are dire: it means software projects will fail (what's new) because of politics, lack of right people on the team (from requirements analyst and beyond) and lack of expertise on all fronts.

A counterexample: I was chief architect of the largest social services projects in NL in1990. We had a very small core team and we were quarantined from external noise. The project was  a success mainly because of this fact. We fired our first project leader and replaced him by one who wanted to get the job done, and he lived up to his promise.
These days, government projects are horror stories.
Well, not exactly. The STEM people make up around 10% of Congress. Still not a lot.  

https://www.npr.org/2018/11/18/66794400 ... w-congress
OK, still.
STEM people ted to be not so strong at (playing) politics, And how many get an audience at court?
Is it true that 60% of the population  believe that the earth is 4000 years old, give or take some daylight-saving days?
 
User avatar
bearish
Posts: 5623
Joined: February 3rd, 2011, 2:19 pm

Re: Models for Covid-19

May 10th, 2020, 7:06 pm

Nah - that sounds like an exaggeration. I’ll buy the 60% number for Republicans in the South and Midwest, though, which may be enough to determine national government policy. So we may still be doomed...
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 10th, 2020, 8:29 pm

I am regularly visited by those nice people Jehovah's Witnesses and we drink coffee and I show them my 8 kilo 19th century Dutch Bible (they use laptops..). Anyways, they also don't agree on that 4K thing .. the earth according to them is 8000 years old.

68% of Republicans and 40% of Democrats do not believe in evolution.

Not much chance that they will believe an ODE pandemic model.
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 11th, 2020, 9:29 am

/**
 * @brief Airport state.
 *
 * Not used for COVID-19 right now. Might be more relevant for USA and
 * other countries that have lots of internal flights. Slows the simulation.
 */

Ouch
 
User avatar
Alan
Posts: 10209
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Re: Models for Covid-19

May 11th, 2020, 7:46 pm


Alan, 
Are you still enthusiastic about this IHME. Some say it is completely useless.




So, weighing all of the above: yes,--  I am still a big fan. 
I think a big problem is that there is no one(?) in Congress with STEM degree. So, being comfortable with numbers is non-existent.
Same with leaders who studied Greek and the Classics. 

21 currently with STEM background, says this site

Interesting background list from a decade ago
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 12th, 2020, 9:52 am







So, weighing all of the above: yes,--  I am still a big fan. 
I think a big problem is that there is no one(?) in Congress with STEM degree. So, being comfortable with numbers is non-existent.
Same with leaders who studied Greek and the Classics. 

21 currently with STEM background, says this site

Interesting background list from a decade ago
And no philosophers, the only group that can think straight?
"Attorney + ocean scientist" sounds like a good combination.

Here in NL the country is run by jurists and economists. The STEM studies are considered to be unimportant, especially for girls.
Germany, on the other hand, is a technology-driven (German precision) country and in many walks of life you need to have a PhD if you want to reach the top. Angela Merkel has a PhD in Physics.

In UK, a study In Eton and Oxford is an automatic guarantee for advancement.

In Ireland, a while back at Trinity College, one of our maths professors was also a senator and he helped to introduce laws to protect that 5% of the population that were not belong to the dominant persuasion. 

The punch-line: decisions are made based on various criteria... e.g. it is possible to change/reverse engineer the requirements to fit reality.

And President Eamon de Valera (front, middle) had a PhD in mathematics, He brought Schroedinger, Cornelus Lanczos to DIAS. 

Image
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 12th, 2020, 11:59 am

Another question is how [$]R[$] is calculated? Seems like a well-kept secret. 
Anyways, herd immunity has fallen out of favour.
 
User avatar
katastrofa
Posts: 9329
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Models for Covid-19

May 12th, 2020, 11:54 pm

That's a good question. R0 (basic reproductive number) can be calculated in many ways and lead to different results.

BTW, that's the problem with the whole Ferguson's model - it's an element of political and ideological war: you are for or against it (or rather you are with or against people you like or not like who do/do not like his model for the same reason). I don't like people in general, so I'm not biased: in my opinion the model is poor, just like the code - and in the microsimulation domain, the working paradigm says that the code is the model (we haven't even seen the original code, but the Microsoft edition is also not sufficiently tested - and looking at the github account activity they continue this malpractice despite wide criticism; there's no clear documentation either - you can trace the assumption in the code, but you don't end up 100% sure what the code does).
The model's "micro-" level is blurred by the approximations and simplifications, which practically puts it in the same bag with reduced-form models like SIR. I would even venture to say that it's inferior to them, because we can't understand it fully owing to the poor design and implementation.
The country-wide lockdown intervention, disregarding social, economic, etc. aspects is not a good microsimulation prediction. Microsimulations give detailed distributional answers: who, where, when, how and how much... Better take clear and simple SIR and a panel of experts from all relevant fields to advise what to do. Of course someone could write a proper microsimulation of this country and its operations, but we all know it's not going to happen.
Last edited by katastrofa on May 13th, 2020, 12:53 am, edited 1 time in total.
 
User avatar
Paul
Posts: 10787
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

May 13th, 2020, 12:15 am

Very well said.
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 13th, 2020, 6:09 am

Flashback to March 17, I wrote

"Today on BBC they actually showed the (udocumented) ODE model (an ODE system with 7 unknowns). Here is a snippet; Can it model Corona? sound too good to be true. And almost linear. My guess is that it cannot model a lot? It looks awful linear, almost. 1st year electrical engineering syllabus??"

This was the purported SIR model being modelled Seven weeks later the Imperial source code has now been published and is maintained by Microsoft.

The issue seems to be that the code is the implementations of an otherwise undocumented model but is not the SIM model that the BBC announced with such bombarderie last March 17.

So, something got lost in action/in translation on the vicus of recirculation from  requirements to code.

//
Professional/industrial projects

1. Requirements elicitation A-Z
2. Analysis and data model
3. Detailed design
4. Coding
5. Acceptance testing
6. Product delivery
7. Planing for project updates.

8. GOTO step 1.

Those here with a background in the Classics will immediately see the analogy with Theseus and his ball of thread. The difference now is that each stakeholder has been allocated his own ball to play with.
Last edited by Cuchulainn on May 13th, 2020, 9:08 am, edited 1 time in total.
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 13th, 2020, 6:40 am

Remark; I have not found references to [$]R_0[$] in the software implementation of the model. Can it be computed without finding equilibrium points of ODEs??

Indeed, I do remember Paul having doubts about R_0; everyone uses it now (even physicist Dr. Angela Merkel) as if it were a law of gravity. And the shorter the formula the easiest it is to believe that it is true, e.g. [$]E = mc^2[$].

the working paradigm says that the code is the model 
Anyone who says that will be fired!
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 13th, 2020, 11:27 am

Some brainstorming..

Damning remarks by Professor Johan Giesecke on the Imperial College model and the resulting fallout.
He seems to be saying that > 90% is already mildly infected. So in SEIR model [$]E(t=0) = 0.9?[$]

www.youtube.com/watch?v=H2TEOnOnHVM&fea ... u.be&t=154
@5.48

- article never publisher (it was an internal memo)
- not peer-reviewed
- underestimation of E big time

"The UK government made a U-turn off the paper overnight".

This project would never have seen the light of day if they had done a thorough Requirements Analysis and FEASABILITY STUDY. Those kinds of things I used to experience and it does not take long to find out that it is high risk. Chico Marx would call it the "SANTY CLAUSE". It is not even funny anymore.
 
User avatar
Cuchulainn
Topic Author
Posts: 62394
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Models for Covid-19

May 13th, 2020, 2:35 pm

Looks as if it's under control in Sverge.

Image
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On