layout: wip
title: mj part time coding Q4 2021
author: mj
date: Oct 20, 2021
amount: 72.0 XMR
milestones:
- name: Month 1
funds: 24.0 XMR
done: 30 November 2021
status: finished
- name: Month 2
funds: 24.0 XMR
done: 31 December 2021
status: finished
- name: Month 3
funds: 24.0 XMR
done:
status: unfinished
payouts:
- date: 4 December 2021
amount: 24
- date: 5 Jan 2022
amount: 24
- date:
amount:What
In the same way as previously, I propose to work for 3 months, spending 30 hours a week on Monero Core, specifically on topics such as (in this order):
- statistical simulation (see Proposal)
- addressing user issues (whenever I can help)
- enabling and helping new developers
- code reviews
- CI fixes
- extending my Monero health report
- adding Monero-GUI to the health report
- general firefighting, whatever problems we face in near future
Why
I need to raise the stakes unfortunately, since I'm strongly considering ditching my job, which became a pain in my back. This means, that I will be more dependent on the XMR money. Also, since I live in Western Europe, I pay my bills in EUR, which is seemingly going down against USD in a long term trend. I normally hate hearing from people, who I have to hire to do things, which they can do better than me, that "they Have to earn more". I prefer saying first what additional stuff I'll deliver. Since my job will not get into the way as it did before, I will have more mental capacity to learn and read about the internals of Monero (outside of the proposed 30h/w). So far I've been doing (and gaining experience in) a lot of support tasks and I'm happy with the results as far as they can be reviewed and merged by the Maintainers. However there are now also some new challenges in Monero, which deal with statistics. I have some self-taught knowledge the field and I could provide some time series simulations, that would help in verifying if a given statistician's (or my) solution is able to yield the expected results in multiple alternative scenarios, as opposed to relying on a fixed history, which always leads to overfitting. Please refer to the wonderful work of Nassim Nicholas Taleb, especially Fooled by Randomness for more background. I already have a working software for such simulations, but it has to be adapted to fit Monero. I'd gladly reuse it for you guys, while you'd only pay for the adaptation itself. I think it's a good deal. I've been working on this software for many years for the purpose of statistical analysis of financial data, and a huge amount of them.
Some more details and summaries of other work packages are below:
Statistical simulator
Here are the most relevant features of the simulator, that I'd adapt for Monero:
- Parallel Monte-Carlo simulation of alternative scenarios. The result of such a simulation are median and standard deviation of all experiments (in other words: the expected value as well as best and worst case scenario)
- The resulting value(s) are of a unit, defined by whatever objective function that is being fed into the simulator. In my current case of the financial simulation, it's the trading profitability.
- Smooth interfacing with Python (either via Boost or TCP), as Python is the (reasonable) language of choice for Statisticians. This means, that the Statisticians will not have to immediately rewrite their code to C++, if they want to have them checked by a very fast C++ simulator, compared to an analogous task, performed by Python code.
- Very fast loading of serialized data