dcp | python main

I am trying to run the following optimization using CVXPY:

Problem: I have 50 text files, each with thousands of lines of text, each line has a value on it. I am only interesting in a small section near the middle (lines 757-827 - it is actually lines 745-805 I'm interested in, but the first 12 lines of every file is irrelevant stuff). I would like to read each file in. And then total the values between those lines. In the end I would like it to print off a pair of numbers in the format (((n+1)*18),total count), where n is the number of the file (since they are numbered starting at zero). Then repeat for all 50 files, giving 50 pairs of numbers, looking something like:

(18,77),(36,63),(54,50),(72,42),...

Code:

Problem

I have a set of 50 files, which contain 8192 lines of integers (after skipping the first 12, which are irrelevant, there are also 12 more lines at the bottom which can be skipped). I am only interested in a patch of 70 lines in each file (lines 745-815, or 757-827 if you include the 12 at the start). The files have a naming pattern of 'DECAY_COINC000.Spe','DECAY_COINC001.Spe' etc. Desired output below.

Existing Code (Minimalised Example)

A pet store keeps track of the purchases of customers over a four-hour period. The store manager classifies purchases as containing a dog product, a cat product, a fish product, or product for a different kind of pet. She found. a. 83 purchased a dog product b. 101 purchased a cat product c. 22 purchased a fish product d. 31 purchased a dog and a cat product e. 8 purchased a dog and a fish product f. 10 purchased a cat and a fish product g. 6 purchased a dog, a cat and a fish product h. 34 purchased a product for a pet other than a dog, cat or a fish. i. How many purchases were for a dog product only? ii. How many purchases were for cat product only? iii. How many purchases for a dog or a fish product? iv. How many purchases were there in total? enter code here

I have a query like this :

This is a follow-up to an earlier specific question, but as I add more complexity to the problem formulation, I realize that I need to take a step back and consider whether cvxpy is the best tool for my problem.

What I'm trying to solve: create the largest cluster of a category and company where the average values are above a particular threshold. The trick is that if we include particular categories for a company in the cluster, in order to add another company, that company also should have high values for the same categories.

I've formulated this as an integer linear optimization problem, where I've expanded out all the variables. There are 2 main problems:

1. Constraint 8 violates the DCP rule. Constraint 8 is intended to keep the values included in the cluster above a particular threshold. I check this by taking the average of the non-zero variables.
2. The actual problem will require thousands of variables and constraints to be specified (there are 100 categories and >10K companies). This makes me question whether this is the correct approach at all.

I am designing a system which needs to track tables of different types. It would be highly convenient to have one authoritative list of the tables*, and the best way to have this list would - I think - be as a list of Proxies, as at times I do need the type. I suppose HLists are an option, but even after 2 years of using Haskell a fair amount, they look unwieldy, so I was reaching for Dynamic.

• Actually, it would be best to have 0 lists, and just have a way to query all available instances of a particular type class at run time, but I'm not sure there's a way to do this (last I checked).

My code currently looks like this:

I'm trying to solve a SOCP problem using cvxpy and integrating it to cvxpylayers. I'm looking at this SOCP problem (problem 11) (here is the scihub link in case you can't access), and here is a snippet of the problem (note min (p-t) comes from an adaptation of problem 4 using expression 8 in the link above):

Go to ---> EDIT 2

OLD

I've looked at this example, but is still stuck and can't get the problem to be solved. Here is a sample code at my attempt:

I would like to write the log likelihood of the Dirichlet density as a disciplined convex programming (DCP) optimization problem with respect to the parameters of the Dirichlet distribution alpha. However, the log likelihood