comment 0

How to get your money back from Emma Mattress?

Disclaimer: This is my story (originally posted in mid August) and what I experienced. There is no intention of giving unsolicited legal advice. Experiences are to be shared so that people can judge and use any and all information they acquire. That being said…

I Almost Sued a Company!

Internet shopping always comes with the possibility of fraudulent businesses and unfulfilled promises. Long story short, I encountered one of these companies. Emma Matratzen GmbH is a mattress company all around Europe with lots of internet presence. They advertise their products in all the social media platforms as well as on classical media. They are proud of their product and flash their awards on every advertisement. On the other hand, you meet hoards of upset customers if you even pay a little bit of attention to the comments of this company’s advertisements online.

Well… I should have paid some attention to those internet testimonials. I recently moved apartments. I was in the need of a bed and ordered one from Emma Matratzen GmbH. They failed to send my bed and made me sleep on my couch for a month. When asked, they said they will not cancel my order either! The full version of the story is below. Let me move onto how I got my money back from Emma right away.

Here is the Algorithm of How I Got My Money Back From Emma:

This is what worked for me and my business with Emma ended in a record breaking 45 days. I see people are struggling for months/years with this company.

  1. Send Emma regular customer support requests, ask for updates on your unfulfilled order/cancellation/refund.
  2. Find and contact your nearest Customer Rights Office and ask for their advice.
  3. In a formal language, send Emma Matratzen GmbH an email formally asking for the cancellation and give them a formal 14 days grace period to refund you. Cite the following law + reason:

    European Union Trade Law: [CELEX-No .: 32011L0083 ] Chapter 33, Items 37, 40, 42: The present contract is a business without the physical presence of the contracting parties, outside of the business premises of the entrepreneur. With this type of business the customer has a right of withdrawal in accordance with [CELEX-No .: 32011L0083]. If the withdrawal has been declared, the company is obliged to repay it.

  4. No matter what they say/do after point 3. remind them that you have given them the 14 days formal grace period and your decision is final.
  5. If they fail to refund you in the 14 days, contact your nearest Customer Rights Office again.

I feel like this is a solid algorithm. I hope you will not let them keep your honestly earned money.

Here is What Happened to Me:

This is a story as old as time. A company does some customer wrong, that customer has lots of quarantine time on their hands, and next thing you know, they decide to procrastinate on everything they are actually supposed to do and fight back. Does it make this a Don Quixote or an actual standing for your rights story? You decide.

I moved apartments recently and in that I needed to purchase a mattress. I decided to purchase a mattress from Emma Matratzen over the internet due to their 100 days no-questions-asked return option and the one award they keep flashing on their advertisements. It looked like a safe bet. I ordered a mattress on June 29th, 2020, to be delivered between 7th to 11th of July. This was supposed to be my new mattress in my new apartment. I found having a mattress in an apartment to be quite important. Silly me. This order got unfulfilled. I contacted the Emma Matratze GmbH through their customer support system many times and finally I got a new delivery date of 22nd of July. I have not received the order on that date either. I told them that I want to cancel this order on July 22nd for the first time to no avail. I tried to contact them many times through their online customer support but they did not respond to my inquiries.

After the second failed delivery date and unanswered customer support emails, I called the company directly and asked the representative what was happening. They told me that the mattress was in their warehouse, it was supposed to be sent soon. I was told that I was going to receive a tracking number on July 24th and receive the mattress exactly on July 27th, 2020 (jokes on me, this did not happen). I clearly stated that their business conduct was not appropriate and asked them to cancel my order. They told me that they will not cancel my order. Then I asked for a return and refund my order, if they will not cancel it. This company offers 100 days of no-questions-asked refunds. They said they will not refund my order either. To add insult to injury, the representative hung up the phone on me when I asked “What would you like me to? You are not sending the bed, not cancelling it, not giving my money back, what would you do if you were in my shoes?”.

Right after my phone conversation with the Emma Matratzen GmbH, I contacted my credit card company and told them about this issue. I was hoping that we can dispute this fraudulent charge at the credit card level. I explained the situation. They suggested that I contact their Risk Management Department and also to contact the Consumer Rights Office. Sadly, the Risk Management Department were late to respond and I did pay for a bed that I will never see. Hence, Emma Matratzen GmbH really got my money without providing any service and left me in an apartment with no mattress to sleep on.

At this point I was already sleeping on my couch for weeks. Now think about this, you have a new apartment, new life, new hopes, a new bed frame, a new box spring, new sheets but no mattress… for weeks.

I decided enough was enough and contacted the Upper Austria’s Customer Rights Office on July 29th. Exactly one month after my initial order. They told me that I was in the right and that I should cite the law and ask for my money. I wrote a formal email to Emma and told that I have the Customer Rights Office involved, and they have guided me to give them a formal 14 days to cancel my order and fully refund me. Trust me, you’ll start getting responses from Emma Matratzen GmbH once you tell them that you contacted the Customer Rights.

Emma replied with the same shebang, how they have the mattress advanced in the shipment process, whatever and ever. But you have o understand that they are so sneaky and will do anything to keep the money. Here is the BS they suggested I do to get my refund (even after I told them that I have the law on my side):

  1. I wait for the shipment to arrive,
  2. Reject the delivery,
  3. Wait for the shipment I rejected to go back to Emma,
  4. Then wait for them to start the refund process.

This is for an item they failed to ship after a month. Oh, the idiocy of this company’s practices! How dumb do you think I am? I flat out rejected this and reminded that they have been given a 14 days grace period to refund me and that was final. And in each email from that point on I reminded them that they have been given this grace period to refund me and I expect my funds reinstated by the end of August 12th.

I interacted with couple of Emma Matratzen victims on the company’s Instagram page. I told them that I had enough of this company and I already contacted Customer Rights Office on August 2nd or 3rd. I advised everyone to do the same.

Surprise! Surprise! My comments disappeared on Instagram and furthermore, on August 4th, I got an automated email from Emma saying my mattress is on its way with a DPD tracking number. This is 15 days after I first asked them to cancel the order and a week after I got Customer Rights Office involved. This was easy though, I told Emma that they sent a mattress into the void and I would reject their shipment even if it comes and their grace period was still in effect. I contacted DPD and told them that I will reject the item and this was Emma Matratzen GmbH’s pitiful attempt to delay my cancellation and refund. DPD cancelled the shipment for me. After I told Emma I even got their shipment cancelled, they finally (and unwillingly) told me that they will refund me.

They still waited till the literal end of the grace period. I got an automated email from Emma stating that they started their refund process and that I will get my money back on August 12th at 20:20 (8:20 pm). I got my money reinstated couple of days later. In my mind they still failed to finalize the refund in time. I am callused by now though.  After all this is Emma Matratzen. No surprises here. They seem to suck at everything business related. They are late, but at least they have done one thing that they were legally obligated to do. All in all, I finally got my money from the claws of this fraudulent company and I am happy now.

Postlude and Reflection

I kept on thinking if my expectations were too high in the times of Covid19. Then again, all the other companies I purchased stuff from delivered their shares to my apartment on time.  Moreover, I was not the one that put the delivery estimates on Emma Matratzen’s website. If only they wrote:

If you purchase this item on June 29th 2020, we will lie to you constantly, force you to the brink of a legal battle with us, and maybe ship your order on August 4th after you asked us to cancel it many times.

I would have decided not to order it. Instead they wrote, guaranteed delivery dates to be in 7 to 10 days.

To make it even more clear that shipping a mattress is/was possible even at this time: I actually ordered a mattress from another company on 27th of July and got it delivered on 29th of July successfully, while I was still trying to get my money back from Emma.

I also thought about sharing this online or not. I am not known for being an activist. I do not know myself as an activist. This time though, I found a huge database of people that fell victim to this company on social media platforms (especially on Instagram and Facebook, but Emma Matratzen seems to actively suppress bad publicity especially on Instagram, so this group of people keeps on changing), on,, and the list goes on.  I felt the need to share my experience at least with them.

I hope this story would entertain someone, or maybe even give the courage to take a step against wrongdoing.

I would be happy to provide the documentation to everything I have written above.

If you are from the Emma and got hurt feelings, you have my phone number. Call me.

comment 0

Where do the Maximum Absolute q-series Coefficients of (q;q)_n occur?

Dedicated to the MACH2 group, in particular to Prof. Wolfgang Schreiner and Johann Messner.

In mathematics research, some questions are too easy to ask that one cannot help but feel amazed when the answers to these questions are nowhere to be found. The title is an example of such a question, I and Alexander Berkovich (my former advisor from the University of Florida) asked in 2017. Our account on this question recently got published in Experimental Mathematics. I suggest checking the ArXiv version of this paper or the actual published article, if my explanation of the problem sparks your interest.

I want to motivate the problem by introducing a well-known q-series identity first. The Euler Pentagonal Number Theorem (EPNT)

\displaystyle \sum_{i=-\infty}^{\infty} (-1)^i q^{\frac{i(3i-1)}{2}} = (q;q)_\infty,

where for any non-negative integer n

\displaystyle (q;q)_n := \prod_{i=1}^{n} (1-q^i) and \displaystyle (q;q)_\infty = \lim_{n\rightarrow\infty} (q;q)_n.

When one looks at the sum side of the EPNT, it is clear that all the coefficients that appear in this formal power series are merely one of the three options: -1, 0, or 1. This property is not visible from the product representation. Moreover, it is highly surprising to see such cancellations to appear in the series expansion of an infinite product!

But wait, there is more! The products (q;q)_n will always have a coefficient that is

  • > 1 in size if n\not= 0,\ 1,\ 2,\ 3,\ 5 or \infty.
  • > 2 in size if n\not= 0,\ 1,\ 2,\ 3,\ 4,\ 5,\ 6,\ 7,\ 8,\ 9,\ 11 or \infty.
  • > 3 in size if n\not= 0,\ 1,\ 2,\ 3,\ 4,\ 5,\ 6,\ 7,\ 8,\ 9,\ 10,\ 11,\ 13,\ 14 or \infty.
  • > 4 in size if n\not= 0,\ 1,\ 2,\ 3,\ 4,\ 5,\ 6,\ 7,\ 8,\ 9,\ 10,\ 11,\ 12, 13,\ 14, 15 or \infty.
  • ...

This is what Berkovich and I proved in 2017* using elementary observations and simple calculations.

For me, the above list highlights how special the infinite product (q;q)_\infty that appears in EPNT once again. There is no way of seeing these unbelievable cancellations that lead to small coefficients of the infinite product when we look at the finite products. Furthermore, the size of the coefficients of the (q;q)_n polynomials grow, and they grow exponentially. In 1964, Sudler studied the size of the maximal absolute coefficients of these products and proved their exponential growth*.

Let’s get a little more technical and write out the polynomial

\displaystyle (q;q)_n = \sum_{i=0}^{n(n+1)/2 } a_{i,n} q^n.

We can observe how fast these coefficients grow with the following animation where we plot (i, a_{i,n}) from n=0 to n=100 in steps of 5. On the top of the image, you can also see the Maximum Absolute Coefficient (denoted as M) and the first location at which this coefficient appears (denoted by L).

At n=100, this polynomial has 5051 terms, about 1/3 of which are huge and all around the place! The maximum absolute coefficient at n=10 is 3. This becomes 11,493,312 when we reach n=100. They grow really fast; just as Sudler discovered.

Whenever we have a list of integers, it is a good idea to check the On-line Encyclopedia of Integer Sequences. We already had the absolute maximum coefficients (Ms) of (q;q)_n for n\leq 50 for our 2017 paper, so we made a quick search of this sequence only to find that this sequence was already listed* and our main question-to-be was right in front of us in the comments.

This is how the website looked in 2019:

The comment is clear. People believe that the absolute maximum coefficient of (q;q)_n appears at the midpoint if n is even (I haven’t seen a proof of this, please contact me if you find one) and it is a mystery if n is odd. Berkovich and I studied these locations by calculating all the (q;q)_n for n\leq 75,000. With that, now, we have confirmed the even index belief all the way up to 75,000 and we actually have a conjectural answer to the location question for the odd cases too. There are so many interesting properties of this elusive location that we supposedly found, but this is not the place to discuss those. I suggest checking the article.

The calculation of these polynomials one by one, finding the maximum absolute coefficient and its location was a huge task. We started calculating (q;q)_ns on Maple, which gave up around n=10,000. We moved to Mathematica and a larger server at Research Institute for Symbolic Computation (RISC) and we got stuck around n=15,000. So I abandoned all the computer algebra systems and started programming in Python and later in C++. Carrying out these calculations correctly is/was not an easy task. I doubt it will ever be in my lifetime. After using all the symmetries and simplifications, one still needs at least \lceil n(n+1)/4 \rceil operations to calculate (q;q)_{n+1} from (q;q;)_n. There is no way of splitting the data. The numbers are growing exponentially and each calculation needed to be carried with arbitrary precision. RISC IT Ralf Wahner dealt with all my incompetence on the RISC side. I need to thank him a lot.

I attended a presentation by Wolfgang Schreiner on the MACH2 massively parallel shared-memory supercomputer project while I was struggling to make progress in my calculations. Using MACH2 was clearly what I needed. My calculations were already crashing computers with 1.5 Tb Ram at RISC. It would have been a great challenge to crash a computer with 20 Tb of shared Ram (I did crash a 10 Tb section of MACH2 once, so I can rest happily).

I contacted the MACH2 team, got an account, and gave a brief explanation of what I wanted to calculate. After a couple of emails, Johann Messner and I met in person and went through my code, line by line. I took his feedback and suggestions and started working on those. In the meantime, he started running tests with then-code. It took so much back and forth. Johann kept on asking me to parallelize my code but I didn’t want to do that in order to save used memory. I resisted as much as I can but he won. He knew the MACH2 beast the most after all. Also, he made the great point that MACH2 has slow processors but it has tons of memory at our disposal.

I didn’t know how to parallelize anything in C++ so I put it on the backburner. It was already more than a year at this point since we first started. The project could have easily sat for another couple of months… and it did. Later in the year, at a conference, I accidentally learned how to parallelize my code from a couple of numerical analysis postdocs. All of a sudden the missing link was there. After Johann’s okay, we cranked the last portion of the calculations (with lots of intermediate checks) from n=40,000 to n =75,000. It still took 3 months even though we were using a little more than 200 cores and 10 Tb of memory.

The mathematical claims of the article came after all these crazy calculations. I came together with Prof. Berkovich at the University of Florida in October 2019. We compiled the data, looked at it left and right. In a short time, we made the best claims we can make about the elusive location. While we were at it, we refined Sudler’s growth constants as well. This ended my first 2 year-long experimental math project.

Long story short, the mathematical claims I made stand on the shoulders of many computers, so much wasted electricity, many sleepless nights, and the input of many people many of which haven’t even been mentioned here.

If you’d like to confirm or disprove that of the maximum absolute coefficient when n=5,050,029 appears as the coefficient of the q^{6,375,697,892,084} term, be my guest. 🙂

I shared this work with some audiences:
  1. Austrian High-Performance Computing Meeting 2020, IST Austria, Klagenfurt AUT. February 19, 2020
    Where do the maximum absolute q-series coefficients of (1-q)(1-q^2)...(1-q^{n-1})(1-q^n) occur?
  2. The 85th Seminaire Lotharingien de Combinatoire, Strobl AUT. September 6, 2020
    Where do the maximum absolute q-series coefficients of (1-q)(1-q^2)...(1-q^{n-1})(1-q^n) occur?
  3. Hudson Colloquium, Georgia Southern University, Savannah GA. September 23, 2020 (Online Meeting)
    Where do the maximum absolute q-series coefficients of (1-q)(1-q^2)...(1-q^{n-1})(1-q^n) occur? (Pdf)
Of course there are some simple typos here and there in the article, let me address the ones I know about here:
  1. There are two 68,324s written by accident. Those should be 62,624s as in everywhere else in the paper.

comment 0

15 minute mathematics advice (13) /w Christian Krattenthaler

Another amazing researcher with an incredible list of accomplishments. I caught him at the SLC 85 conference and had a nice chat. I think we need a second part of this interview because I couldn’t even cover half the topics I’d like to.

I know Christian for a number of years now, and wanted to interview him in this series. I was too shy to ask for some reason… so I thank Wadim for actually being the instigator.

I hope that you will enjoy this video.

comment 0

AMS Special Session on Experimental and Computer Assisted Mathematics

Joint Mathematics Meetings 2020, Denver CO

Here is the pdf version.

I love the Joint Mathematics Meetings. I was fortunate to be invited for these meetings 5 times so far and I get to be there for 4 of these events. For me it is a meeting point to come together with friends and (mathematics) family. I cannot count the number of times that I have seen an inspirational talk or accidentally have run into someone that I was longing to see in these meetings. I can characterize these meetings as huge, disorienting, and sometimes too much but over all this is an event I wouldn’t want to miss.
In a conversation much earlier than JMM 2020, Chris Jennings-Shaffer (Univ. Denver) mentioned how he was applying for a special session at this event. He was already at the conference location and it was making much sense. He suggested that I should also think about applying for one too. It was a well timed suggestion. I wanted to organize something for more than a year but I was always second guessing myself about whether I will be invited to give a talk, take the over Atlantic trip, etc. By this time, I also had some experience in organizing conferences and workshops therefore applying for a special session, where I only need to find speakers didn’t sound bad. We figured that there will be a q-series/partition theory session and a number theory session. If Chris could get his session, that would make it a second q-series/number theory session. I just wanted to advertise a seemingly non-overlapping topic.
From the get-go, I know I wanted to have a different session, one that has a mix of combinatorics and q-series researchers. These two fields have too much in common but somehow don’t get to mingle in general. This was the case in Turkey. This was the case in University of Florida. This was the case in the conferences I have attended. One exception to this informal segregation was RISC/RICAM, Linz Austria, where I was doing my postdoc. I know I learned a lot from the diverse research environment. The Austrian Science Fund’s (FWF) special research project (SFB F50) consists of 12 mixed projects, where combinatorics, q-series, number theory and symbolic computation are mixed and encouraged to come up with solutions together. I joined this group with a pinhole view of what my research area was, and learned that it was touching much more. That was the goal for the special session too. I wanted people to see some related “unrelated” research.
I settled on experimental mathematics as my session subject early on as its definition is open to interpretation. The next big thing was to find a co-organizer that I would like to work with. Thankfully there are many names. Honestly, I figured that applying with someone from an American university would raise the chances. I had no connection with AMS at the moment after all. Shashank Kanade (Univ. Denver) and Matthew Russell (Rutgers Univ.) was a natural duo to approach. They are dear friends before anything, and great researchers that are known for their experimental mathematics publications on top of all the other things they accomplished. Mind you, these are busy people that are also in high demand. Shashank already was organizing/applying for another session and speaking in another, Matthew was busy with his institutional duties. Thankfully it didn’t take much convincing to organize a session together.
We submitted our request to AMS JMM board after a short discussion about the scope and the vision of what our session should be. It must have been a nightmare for the JMM board to decide who gets what because they took about an extra month and a half to decide on the schedule and the special sessions. We got our session. It was official scheduled for 5 uninterrupted hours on the second half of the last day of the conference. It was great news regardless. Not surprisingly all our earlier guesses with Chris was realized too. He got his session. Shashank also got his other session. There were partition theory and number theory sessions too. It was great news all around, there were going to be so much to learn and so many people to see. Just the way I hoped for.
By then, it was clear that Matthew was going to be quite busy and he was not going to be able to join JMM. Shashank and I sent out our invitations and build up a nice mixture of everything. It was certainly much better than I initially planned. From number theory to machine learning and from automated/computerized provers to combinatorics, we had everything. The funny thing is, although many people turned down our invitation in the process, the session filled really quickly and it now feels like we could have easily filled another 5 hours with just as great speakers (I have to admit, I remember us being slightly worried about being able to fill the session back then). Maybe in another JMM in future.
The session averaged 30 people per talk. In a conference which attracts more than 5000 mathematicians a year, this may sound like a small number. In perspective, the special session was actually a hit. The mathematician swarm of downtown Denver was releasing that day, conference attendees were vacating their hotels and heading towards the airport. Even under these circumstances there were 42 other conference events scheduled at the last afternoon of the conference. Still including some really nice research sessions with overlapping research topics with our experimental mathematics session. Our session was one of them. Shashank’s other session was one of these too. He needed to juggle and, I can only imagine, got some decent exercise between two seminar rooms.
Kate Stange, technically a local, opened our session with the demonstration of the project Numberscope, where we were bombarded with many nice visualisations of Diophantine equations related questions. Then Zafeirakis Zafeirakopoulos and William Severa presented machine leearning approaches to nowadays problems and a glimpse of DOE’s research in lowering the energy consumption of high performance calculations. Petr Vojtěchovský, this time definitely a local, talked about his research on universal algebras and the automated prover implementation Prover9 and its strengths in finding and writing out proofs with certificates for some algebra problems. The session followed with number theory. Jakob Ablinger and Michael Mossinghoff presented new computer generated and proven identities involving Riemann zeta value  and experimental observations about divisor sums related constants and problems, respectively. Then we saw an elementary constructive approach to finding extremal values of the statistics related to hypercube orientations as directed graphs with maximal or minimal number of special nodes. We then reached my comfort zone, q-series portion of the session. Drew Sills presented his postdoc project RRtools package that he implemented in Maple and finite Rogers–Ramanujan type identities. He suggested some future research directions, which I am working on to this day. Then Chris Jennings-Shaffer, another local to the conference, presented the proofs of  Kanade-Russell (do these names look familiar?) mod 12 conjectures he found with Kathrin Bringmann and Karl Mahlburg. It still amazes me how a simple guess and prove approach and heavy computer use cracked these conjectures. We closed the session with Frank Garvan, and his new results on the mod powers of 5 and 7 congruences for the rank and crank functions.
Exchanging pleasantries and goodbyes with the participants at the end of the conference always takes some extra time. The conference room kept being filled with nice friendly chat and occasional laughter until a convention center employee politely kicked us out. Carrying our conversations to the convention center aisles was nothing new to us of course. This just meant it was closing to dinner time with friends before saying our final goodbyes of the conference. We ended up going to a local diner. It was the first chance from the beginning of this 4 day, 5 day conference I had with Shashank to do some actual catching up, which was great.
Maybe this is the only downcast of the conference. It is always a cheerful reunion but there is no time to catch up with everyone. Oh well… I bet I am not the only one feeling that way.

15 minute mathematics advice (8) /w Kwai-Lee Chui

Mrs Chui is an amazing University of Florida lecturer.She is loved and well respected in UF. Not to forget that she is powerhouse when it comes to her teaching; teaching around a 1000 in a semester.

She is the one lecturer I did most of my teaching with in my UF time. We have created the Online Calculus II lectures for UF together. Turns out it is still our infrastructure and still our faces, that are teaching the online students since the FALL 2014.


We talked about how the education is changed, how the increasing numbers create more paperwork and sometimes take the fun out of teaching, how I missed teaching with her, etc. Hope you enjoy it.